Vivo's latest foldable just hit the global market, after first appearing in China in June. I'll rip the band-aid off early: the X Fold 6 won't be launching in the UK or US, but it will be available in select European markets (as per usual).

The Vivo X Fold 6 is launching at a very interesting time. Apple's first-ever foldable, the iPhone Duo, is just about to land, and Samsung shook things up with its passport-sized Galaxy Z Fold 8 earlier this year, too. Vivo keeps things a little more traditional; the X Fold 6 is the normal book-style fold that Android users are accustomed to. And to keep it appealing, Vivo has maxed out the specs in every department.

Choosing a foldable over a bar phone has always come with compromises, whether that be size and weight, battery capacity, or camera performance. Vivo has been working to eliminate every single one of these limitations, and it's getting really close. I've been using the X Fold 6 as my main device for the past week to see if it lives up to the hype.

The polar opposite of the iPhone Duo

At the moment, it feels like all eyes are on Apple, as the official launch of the Duo draws closer. I get the impression Vivo is more interested in one-upping the likes of the Honor Magic V6 and Galaxy Z Fold 8 Ultra, but the launch timing means that many will be weighing up the difference between it and Apple's first attempt at the foldable form factor. It's an interesting comparison, because the two devices couldn't be more different.

The X Fold 6 looks and feels like a Pro Max-sized flagship when folded shut. The cover screen isn't narrow like the Z Fold 8 Ultra's, nor is it shorter like the Duo. It's meant to feel like a regular phone that can unfold into a spacious tablet when you need it, and it succeeds at doing so.

It almost feels like this form factor is going out of style, and I think that's a real shame. The utility is plain to see; apps are all perfectly designed for the cover screen, and when you open it, you can have two apps side by side in their intended aspect ratio. I loved using the Z Fold 8, but I ran into software optimization hiccups on many occasions.

The X Fold 6 is a fairly big and heavy phone, especially compared to the slender Galaxy Z Fold 8 Ultra, but somehow, it's still lighter than the iPhone Duo. The Vivo tips the scales at 235g, while the smaller iPhone Duo weighs 254g. That chunky camera bump on the Vivo is sure to be divisive, though.

On the other hand, the camera bump is there for a good reason. While the Duo forgoes a telephoto camera entirely, the X Fold 6 has the most impressive telephoto I've ever seen on a foldable. Plus, if you really want to take things to the next level, it supports the same external teleconverter lens as the X300 Ultra.

And the camera prowess doesn't stop there. The X Fold 6 has a 200MP main camera with a large 1/1.4-inch sensor, and a 50MP ultra-wide to boot. It also benefits from Vivo's Zeiss partnership, with the signature T* lens coatings and co-developed picture styles in tow.

I've tested almost every foldable phone in 2026, and so far, I reckon Vivo's photos and videos are probably the best. Sadly, I haven't tried the Duo yet, but I do own the iPhone 18 Pro, and in most situations, I prefer the results I'm getting from this foldable phone. That's quite the achievement, and without a telephoto, I don't see Apple competing.

Vivo is also light-years ahead of the competition with its battery tech. Somehow, there's a 6,900mAh battery crammed into the X Fold 6, and I've been getting two days on a charge quite consistently. That's with heavy use, spending lots of time with the large folding screen and taking loads of photos.

Apple hasn't revealed the iPhone Duo's battery capacity (it never does), but what it did say at the phone's launch was revealing enough. Users can expect up to 24 hours of typical use. That's not even close to what the Vivo is offering here.

The same goes for the charging speeds; the X Fold 6 supports up to 80W with a wire and 40W wirelessly. Again, Apple hasn't confirmed the wired speed, but we already know enough. The brand says you can expect up to a 50% charge in 20 minutes; meanwhile, the Vivo can fully charge from flat in about 35 minutes.

Obviously, there will be a lot of differences in the software, but Origin OS 7 is closer to iOS than ever, at least in terms of aesthetics. It has a Dynamic Island clone, really nice-looking Liquid Glass styling, and very similar lock screen customization, too. If there's anything you don't like, you can almost certainly change it, as the customization options are extremely detailed.

Where the X Fold 6 truly pulls ahead is with its multitasking capabilities. So far, all sources report that the iPhone Duo can only display two apps side by side in split-view, and that these windows aren't resizable; it's always a 50/50 split. What Vivo is doing is on a whole different level.

On the X Fold 6, you can have up to four apps open in split view, all resizable, and then you can open a fifth as a floating window on top. Still not enough? You can also add tabs to the floating window to quickly switch between apps. I can't imagine anyone ever needing more than that; it's essentially desktop-level multitasking in your pocket.

Where Vivo falls short

Of course, no device is perfect, and the X Fold 6 has its shortcomings. The biggest one is the phone's software polish, notwithstanding those aforementioned multitasking features. I doubt I'm alone when I say I was salivating when I first saw the iPhone Duo's unfolding animation. It's incredibly slick.

By contrast, the X Fold 6 reacts more slowly than most when being opened or closed. Instead of smooth animation, you'll be greeted with a blank screen every time you unfold it, and then you'll have to wait a split second for the UI to catch up and appear on screen. This transition isn't hard to live with, but it's far less polished than what Apple is offering on the iPhone Duo.

For the most part, though, I really like the X Fold 6's software. It's quick, smooth, and extremely customizable. Unfortunately, it has the same quirks we always see from Chinese brands. Apps will close themselves in the background to conserve battery life, which can result in delayed notifications until you go in and tweak the settings.

There's also that ever-present lock screen feature that just shows random images until you disable it. Weird.

The iPhone Duo impressed with its almost-invisible crease, too. The Vivo X Fold 6's crease is minimal, but it's still there. I'd say it trades blows with the Z Fold 8, and it's a step behind the Oppo Find N6's crease. It seems like Apple might be leading the pack in this particular area, but we'll pass judgment once we've spent more time with it.

Then, there's the issue of availability. If you live in a region where Vivo is present, then great, no issue — but that won't be the case for a lot of our readers. I live in the UK; Vivo doesn't operate here, so the only way for me to get a Vivo phone is to import one from a European neighbor, or grab a Chinese model.

It's not hard to do, but it means there are no contract deals on offer, and if anything goes wrong with the phone (which is more of a worry with foldables), I'm on my own.

Should you get one?

The Vivo X Fold 6 is a foldable phone with almost no compromises. Honestly, it'd be a pretty solid phone if it didn't unfold, and the fact that it does is just the icing on the cake. The battery life is incredible, gaming performance is solid, and the cameras are among the best we've seen on a foldable to date.

If the iPhone Duo is the foldable phone for people who didn't know they needed a foldable, the Vivo X Fold 6 is the phone for foldable users who are fed up with compromises. The two couldn't be more different, and I'm glad they both exist.

As someone into photography, gaming, and content creation, who often needs to get serious work done on the go, the Vivo X Fold 6 is almost certainly more suited to my needs. But most people don't demand so much from their phone.

I have no doubt that many will prefer what Apple has cooked up. Still, I've been having an amazing time with this Vivo, and I don't want to put it down anytime soon.

Artificial intelligence is becoming part of almost every workflow. Developers are using it to write and test code, customer support is turning to it to handle routine queries, and security teams are using it to analyze alerts and support investigations. Digital forensics and incident response (DFIR) is no exception.

But as digital forensics work has evolved over the last few decades, one lesson remains consistent. Tools can support investigations, but investigators remain responsible for the quality of the conclusions they reach. As AI becomes increasingly integrated into workflows across industries, the challenge is understanding where AI tools can assist and where human expertise remains essential – and importantly, where organizations should draw the line.

Where AI adds value

For digital forensics and incident response teams, some of AI's most promising applications are also its most practical.

With 90 percent of criminal investigations and prosecutions now involving a digital element, DFIR teams are being asked to process enormous volumes of data. As of February of this year, more than 20,000 devices remain in the digital forensics backlog in England and Wales alone.

Data is spread across endpoints, cloud environments, communications platforms and user accounts, and AI can help investigators triage. It can process that information more efficiently and help to identify potential leads, spot connections between datasets, and highlight other sources of evidence that may require further examination.

For example, it can help investigators correlate relevant activity across multiple devices, accounts and data sources, supporting the reconstruction of events timelines that would otherwise take days to piece together manually. It can also help practitioners organize information and structure reports, allowing them to spend less time on administrative work and more time on analysis.

Used appropriately, AI can reduce the time spent on manual, repetitive work and allow skilled practitioners to spend more time on the analysis and decision-making that moves their work forward.

The danger of treating AI as an answer

The challenge is that AI often appears so authoritative that people become overly reliant on its outputs.

One of the most important lessons from digital forensics is that every tool has limitations. Experienced investigators understand that forensic tools can be wrong. AI is no different. The greatest risk of AI is using it without having the foundational knowledge needed to recognize when it is wrong.

AI is like your junior analyst on their best day. It can identify patterns, highlight areas of interest and help move work forward. But no organization would allow a junior analyst to make the final decision in a sensitive investigation without oversight and accountability. AI should be approached in much the same way.

Problems arise when organizations treat AI-generated outputs as answers rather than something that needs to be validated by someone with the expertise to challenge them. AI can help you reach a conclusion, but it shouldn’t be the conclusion in and of itself.

Knowing where to draw the line

There are many areas where AI can accelerate and enhance work, but there are some areas where AI should never be used. Organizations need to pay attention to where those boundaries lie.

In digital forensics, AI can help identify information that warrants further examination. It can support analysis and help investigators work more efficiently. But it shouldn’t be blindly trusted to reach investigative conclusions or determine innocence or guilt.

Those decisions carry legal, professional and life-changing consequences, making human judgement and accountability non-negotiable.

The same principle extends beyond digital forensics. Organizations in every sector are now deciding how much responsibility should be handed to AI. The answer will vary depending on the use case, but the underlying question remains the same - what safeguards are in place to prevent people becoming overly reliant on AI outputs, and how will they know when those outputs are wrong?

Building frameworks for responsible adoption

These are exactly the questions currently being addressed by the new practical frameworks for the use of AI in digital forensics and incident response. They encourage organizations to consider how much risk a task carries, how outputs will be reviewed, and where human oversight should remain mandatory.

Using AI in DFIR requires strict governance because the stakes are so high. Mistakes are taken seriously when someone’s life is on the line, or a wrongful conviction could take place. However, it shouldn’t take consequences this severe for organizations to work towards protecting the integrity of their decision-making and ensuring humans remain in the driving seat.

Rather than racing to embed AI into every process, organizations should establish governance frameworks first – including what AI should never be in charge of – then start with low-risk use cases where outputs can be easily reviewed or reversed. From there, they can expand capability and explore how AI can help their teams do their jobs better, acting as a force-multiplier rather than the final arbiter.

We've featured the best AI website builder.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Last October, a massive DDoS attack suspected to be linked to the Aisuru botnet disrupted some of the biggest names in gaming. Steam, Riot Games, PlayStation Network and Xbox experienced simultaneous outages, affecting games including Counter-Strike, Dota 2, Valorant, and League of Legends.

Researchers investigating the event estimated that the attack may have peaked at 29.69 Tbps, which would make it one of the largest DDoS attacks ever recorded.

These are top-tier, sophisticated technology companies with significant resources devoted to staying online. Yet even at this level, major disruptions still happen. That raises a more useful question than simply how large the attack was: Why can the DDoS defenses that organizations have invested in heavily, still leave them exposed when it matters most?

Why Availability Outages are a Revenue Problem in Gaming Gaming accounted for 19% of DDoS attacks observed by Gcore in H2 2025, making it the third-most-targeted sector, behind technology and financial services. Overall DDoS activity in Gcore's dataset increased 150% year over year.

For a gaming company, availability is arguably the biggest part of the product. If players cannot authenticate, stay connected, or play at all, users, and therefore revenue, go down. The consequence of extended disruption is a lack of trust, affecting the bottom line.

For example, an outage during a major launch or tournament for a gaming company is catastrophic because it’s occurring exactly when they’ve put time and resources into ensuring maximum player demand. A failed launch window means marketing dollars, player acquisition, and revenue all take a hit.

How configuration drift creates DDoS protection gaps

Configuration drift is the growing mismatch between deployed DDoS controls and the changing environment those controls were configured to protect. Gaming is especially vulnerable to this because the product changes constantly.

Studios are always shipping updates because players expect it. To stay relevant and exciting (and profitable), they need to always be launching a major update or new title, or expanding to a new region or user base.

Constant changes mean constant risk. Every new change is a chance for a system or safety net that was in place during deployment to fail. For example, let’s say a studio launches a new matchmaking service. Traffic to that service may be routed differently, but the mitigation policy protecting the original service doesn't automatically follow it. Now you’ve got a gap. The DDoS protection may be the same, but the environment around it isn't.

In addition to product, feature, or service releases, any change to the IT ecosystem may have an impact on the environment the security controls were configured to protect. Those changes accumulate. This is called configuration drift. And it becomes the default, not the exception.

Why point-in-time DDoS testing cannot keep pace with continuous change

This problem is especially concerning in gaming, but any company operating a fast-moving digital service faces some version of it. Point-in-time red team testing alone can't provide current evidence about DDoS protection across a continuously changing environment. If point-in-time testing happens in January, and then the environment materially changes in February, what exactly does the January result tell the team in August?

Despite this, point-in-time testing still has real value: it tells a studio how well its people and procedures respond on the day the test is run. The issue is that the change is continuous. And the release and testing cycles are running on different clocks.

A game studio might make meaningful production changes every week while its formal testing schedule runs, say, quarterly or semiannually. Every change between those two points potentially creates a gap between the environment that passed the test and the environment players are actually using.

Point-in-time testing was never built to validate configuration coverage across an entire, constantly changing attack surface. For gaming companies, that leaves too much time between an environmental change and knowing whether existing DDoS protections still cover it.

What gaming security teams should change

Because gaming infrastructure is constantly changing, DDoS validation needs to be ongoing rather than relying on scheduled point-in-time testing or waiting for major events like a new launch or a tournament. This gives security teams current evidence that their protections continue to work even as infrastructure and services evolve.

That ongoing process should account for changes to public-facing services and how existing protections respond to them. Did this change alter what is exposed, how traffic reaches it, or how existing DDoS controls respond?

When ongoing validation identifies a DDoS gap, teams should investigate how it emerged rather than treating it as an isolated vulnerability to patch. If a configuration change exposed one service, the same change or pattern may exist elsewhere in the environment. Finding the vulnerability should prompt teams to look for similar exposure across other services.

What this means for any always-on digital business

Players don’t care about protection deployments, whether their preferred game passed its last point-in-time test, or if an outage is caused by the technology itself or an outside attack. The game works, or it doesn't.

Gaming makes the consequences of these vulnerabilities unusually visible. When availability is part of the product, even a small weakness can become a business problem the moment attackers find it.

For any always-on digital business, leaders need to pay attention to how their defenses can keep pace with the environment they are responsible for protecting, especially as it changes. The same lesson applies to any company whose customers expect its digital services to be always available.

We've featured the best endpoint protection software.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Historically, the cybersecurity conversation in banking and fintech has centered on detection: can we spot the intrusion, flag the anomaly, catch the fraud pattern before it does damage? We developed incredibly sophisticated tools that can sift through thousands of ‘signals’ every time a customer makes a payment, ranging from how long it took them to type their card number to, sometimes, their social media activity.

That framing assumed a certain tempo: human attackers probing and exploiting at human speed, or automated scripts operating faster but without the intelligence to adapt dynamically or discover novel vulnerabilities. AI changes that assumption.

PYMNTS reported this month that OpenAI could not rule out its highest cybersecurity warning level for its upcoming Astra model, a threshold that signals a model may be able to independently discover and develop working zero-day exploits, or carry out attacks with minimal human direction.

Around the same time, the International Monetary Fund reached a similar conclusion in a note on artificial intelligence and cybersecurity in the financial sector.

The IMF's argument is the one banks should be listening to: AI does not need to invent a whole new type of attack to change the risk equation. AI-enabled attacks may not look fundamentally different from attacks carried out by humans.

What changes is their speed, scale and capacity to adapt. It changes the equation by accelerating vulnerability discovery and exploitation across shared technologies, turning what used to be isolated incidents into correlated disruptions that hit several institutions at once.

Why AI cyberattacks are so dangerous

Banking has never run on bespoke, siloed technology. It runs on shared foundations: core banking platforms, cloud infrastructure, payment rails, identity systems and a long chain of third-party software. Few of these systems are built entirely from scratch. Modern financial infrastructure relies on layers of commercial software, cloud services and open-source libraries.

That makes development faster and can strengthen security through shared scrutiny, but it also creates common dependencies. In the wrong circumstances, that can become analogous to thousands of doors relying on the same key.

An attacker who needs weeks to find a working exploit gives defenders time to patch, monitor, and respond. An AI system can potentially find that same exploit in seconds, and in the most autonomous cases execute a multistep attack with limited ongoing human involvement – 27 seconds is the current record.

The IMF frames this as a dual-use problem: the same capabilities that let AI find vulnerabilities faster for defensive purposes are the ones that let it find them faster for offensive purposes.

This is not a hypothetical for payments specifically. Payment infrastructure has so many overlapping systems at the intersection of common software, cloud dependency, and dense third-party integration that a single exploit becomes an ecosystem-wide issue.

A vulnerability in a widely deployed component can therefore propagate risk far beyond a single institution, particularly where common cloud services, software libraries or third-party platforms are involved.

Why prevention alone isn’t enough

For a long time, resilience strategy has leaned heavily on prevention: patch faster, test harder and review more code before release. All of that still matters. But when discovery and exploitation can happen faster than a conventional patch cycle, prevention alone is no longer sufficient.

The answer is a shift from periodic defense towards continuous resilience: architectures designed to detect, isolate, contain and recover while an attack is unfolding.

The IMF's recommendation points the same way the PYMNTS piece does: institutions need architectures built to limit the blast radius of a successful breach through segmentation, disciplined access controls, zero-trust design, and closer oversight of third parties. Detection, containment, and recovery need to operate at a speed that is comparable to the speed of the threat, not the speed of the last board-approved incident response plan.

That has equally important implications for how banks govern AI within their own security operations. AI tools will increasingly play a role in detecting vulnerabilities and responding to threats, but the degree of autonomy granted to those systems must be an explicit governance decision. Institutions need clear boundaries around what AI can access, what actions it can take independently and where human intervention remains mandatory.

Why boards need to make AI cybersecurity a priority

This is where the conversation has to move from the security team to the boardroom. Segmentation and access control are technical decisions, but the appetite for AI autonomy in critical infrastructure is a governance decision. Who decides what an AI system is permitted to touch? Who is accountable when it acts on a high-level objective in ways nobody fully anticipated? How quickly can exposure be contained once it is identified?

There is no vendor, platform, or piece of infrastructure that resolves this on its own. The institutions that manage this transition well will be the ones that treat architectural resilience, not just threat detection, as a standing governance priority, and that know precisely what their systems, and the AI operating within them, can reach.

The capability curve is not waiting for banks to finish their next audit cycle. Resilience now has to be designed for the speed of the attack it is meant to withstand.

We've featured the best AI website builder.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

For years, enterprises have treated cloud adoption as a milestone of digital transformation. Today, however, cloud adoption alone is no longer enough. As AI moves from experimentation to execution, cloud maturity is increasingly the difference between AI that scales and AI that stalls.

According to the research we recently carried out with more than 2,300 senior decision-makers globally, that gap is becoming increasingly visible. Many organizations have invested heavily in cloud, yet far fewer have embedded it deeply enough into their operating models, governance and business strategy to fully support the next wave of AI-driven transformation.

That distinction matters because AI has fundamentally changed the role of cloud. It is no longer simply infrastructure. Cloud has become the execution layer for AI, where data, applications, automation and security come together to enable faster decisions and better business outcomes.

The cloud maturity paradox

Cloud is now nearly ubiquitous across the enterprise. Yet widespread adoption has not consistently translated into the innovation, modernization and business value organizations expected. Our research also shows fewer than half of organizations are satisfied with the cloud’s impact on innovation or their progress in IT modernization.

That disconnect reveals a deeper truth: adoption has outpaced maturity. Many organizations initially approached their cloud journey by focusing on cost-efficiency, migration speed or infrastructure modernization. Those priorities delivered short-term gains, but not always transformation – and the needs of the AI era have brought that to the forefront.

The organizations pulling ahead approach cloud differently. They invest in modernization, architectures and governance models that enable business reinvention, positioning them to scale AI and generate measurable business outcomes.

Perhaps the biggest misconception surrounding AI is that better models alone create better outcomes. AI amplifies the strengths or weaknesses of the cloud foundation beneath it. Organizations with mature cloud environments can scale AI faster, govern it more effectively and extract greater business value. Those without that foundation often struggle to move beyond isolated pilots.

Almost all organizations report that AI is increasing demand for cloud capabilities, yet a large majority also acknowledge that current levels of investment are insufficient to support their ambitions.

Legacy systems holding enterprises back

One of the greatest barriers to AI success is continued reliance on legacy systems. Around half of the organizations in our research say outdated applications and data platforms are holding back innovation.

For years, legacy technology was primarily viewed as an efficiency challenge. Today, it has become an AI challenge. Organizations can no longer separate modernization from AI strategy because the limitations of legacy systems now directly define how far AI initiatives can scale and how much business value they ultimately deliver.

Legacy architecture is often poorly suited to AI's demands for cloud-native, data-rich and real-time environments. Meaningful modernization requires redesigning applications, data and workflows—not simply migrating them—so organizations do not allow yesterday's architecture to define tomorrow's AI capabilities.

Architecture choices now carry strategic weight

Another shift is the growing importance of cloud architecture decisions. Organizations are no longer selecting cloud infrastructure based solely on technology requirements—they are deciding where AI workloads should run and how they can be governed and scaled across the enterprise.

The data we collected shows strong momentum toward hybrid, multicloud and sovereign cloud strategies. In fact, sovereign cloud adoption is expected to rise by 50% over the next two years as organizations place greater emphasis on data sovereignty, resilience, regulatory compliance and control.

AI workloads are compute-intensive and unpredictable, making decisions about where AI workloads run more important than ever. Cloud architecture decisions increasingly determine whether AI initiatives can scale efficiently, remain cost-effective and deliver business value.

Cloud architecture is no longer simply an infrastructure decision; it has become a strategic business decision.

The rise of platform-led operations

As infrastructure complexity increases, so does the need for a different operating model.

AI-enabled enterprises are dynamic, distributed and constantly evolving, making manual cloud management increasingly challenging.

Agentic AI raises the stakes further. As AI agents begin to act more autonomously across applications, data and workflows, platform-led operations can provide the visibility and control needed to manage them at scale.

Platform-led cloud management enables organizations to embed governance into systems, automate routine decisions and maintain consistency across an organization’s entire environment. It also helps address one of the most common challenges: cloud cost management.

More than half of organizations report difficulties in controlling cloud costs, particularly as AI and agentic workloads introduce new and often unpredictable usage patterns.

Platform-led operating models provide the visibility, governance and automation organizations need to manage cloud in the era of AI while continuously optimizing cost and performance.

They connect modern applications, industry cloud capabilities, AI agents and data into a coherent operating environment, creating greater visibility across business processes and enabling organizations to manage increasingly autonomous AI systems with greater consistency and control. More importantly, they help organizations realize operational savings they can reinvest in modernization and innovation initiatives.

Security as a foundation, not a constraint

The data we collected shows that security is now the top cloud investment priority, reflecting a broader shift in how organizations think about modernization and AI readiness. It also reveals that governance remains uneven.

Organizations with higher levels of cloud maturity are far more likely to define clear security roles and responsibilities backed by formal governance processes, giving them greater confidence in their cloud security posture.

The research also found that risk, governance and compliance are expected to be the biggest challenges as organizations deploy agentic AI over the next 12 to 18 months. As AI systems become more autonomous, security and governance can no longer be treated as an afterthought. Instead, governance must be built into cloud architecture and operating models from the outset.

As cloud, AI and security ecosystems become more distributed and increasingly reliant on third parties, security can no longer be managed in isolation. Organizations need an enterprise-wide approach to risk that brings together controls, tools and responsibilities across providers, platforms and partners, with clear governance and accountability built in from the start.

Far from slowing innovation, strong security foundations enable organizations to innovate with confidence. When identity management, clear data protection policies, continuous monitoring and governance are embedded from the outset, organizations become more resilient—and better positioned to innovate.

From adoption to value creation

Taken together, these trends point to a broader shift in how organizations should think about cloud.

Organizations that will lead in the AI era are those that align their cloud and AI strategies, continue modernizing applications and data, make deliberate architecture decisions, embed governance and security from the outset, and treat cloud as a business operating model for continuous innovation rather than simply a technology platform.

AI is accelerating faster than enterprise cloud maturity, creating a widening gap between organizations prepared to scale AI tools and those that are not.

Ultimately, the question is no longer who has adopted cloud. It is who has matured it into the execution layer for AI — a platform for innovation, resilience and business value.

We've featured the best AI website builder.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

There are plenty of grand ideas about what AI could eventually do for healthcare, whether discovering new drugs or acting as an interactive medical historian for everyone. But a viral Reddit post offers a better, and much simpler vision for AI doctoring.

According to the poster, the family was eating dinner when their father repeatedly dropped his chopsticks. His right hand felt numb, his mouth looked slightly uneven, his speech sounded strange, and he briefly lost his balance. Then he seemed to improve, which was almost enough to convince everyone that he was simply tired and should sleep it off.

After the writer shared the symptoms with ChatGPT, the AI responded that it sounded like a stroke or a related medical incident. It urgently encouraged them to get emergency help despite improving symptoms. The hospital confirmed that he had had an ischemic stroke and treated him accordingly.

This is an anecdote on Reddit, but the advice described in the post lines up with official medical guidance. The CDC says sudden one-sided weakness or numbness, speech problems, and loss of balance are among the warning signs of stroke, and anyone experiencing them should receive emergency medical attention immediately.

For all the complicated debates about AI and medicine, I keep coming back to how uncomplicated ChatGPT's job was in this story. It didn't need to be the doctor. It needed to know when it was time to find one.

The best diagnosis may be to go see a human

Health is an area where an AI's confidence might be problematic. If ChatGPT confidently decides your chest pain is indigestion, you might be reassured when you should be very much not.

OpenAI seems increasingly conscious of that boundary. The company's notes for ChatGPT Health declare it is designed to support medical care rather than replace it and explicitly say the service is not intended for diagnosis or treatment. OpenAI has also highlighted improvements in its models' ability to recognize situations where urgent care may be needed.

The stroke story is almost a perfect demonstration of why that matters. The family did not need ChatGPT to determine which artery was blocked or prescribe treatment. They needed something capable of looking at a cluster of worrying symptoms and saying that waiting until morning carried too much risk.

One reason ChatGPT might actually be helpful is to counter how good people are at talking themselves out of going to the emergency room. Nobody wants to spend six hours under fluorescent lighting only to be told everything is fine. The person experiencing the symptoms may be especially determined to minimize them.

A chatbot has no such social discomfort. Given symptoms that cross an emergency threshold, the useful response is to say so clearly. That doesn't mean ChatGPT should send everyone with a headache racing to the ER. A health assistant that treats every symptom as an emergency would quickly become useless. The challenge is recognizing when the cost of being falsely reassuring is dramatically higher than the cost of being cautious.

Stroke is an unusually clear example. The American Stroke Association uses the acronym B.E. F.A.S.T. to highlight balance problems, vision changes, facial drooping, arm weakness, and speech difficulty, with the final letter standing for time to call 911.

In that context, "maybe sleep on it" is terrible advice. "Get evaluated now" is exactly the sort of sentence I want an AI health assistant to be willing to deliver.

The handoff is the feature

Another Reddit post this week pushes the idea into slightly stranger territory. A user described having ChatGPT Voice running while their blood sugar was falling dangerously low. According to the poster, they became incoherent and kept fading out, while ChatGPT repeatedly woke them, prompted them to check their glucose, and encouraged them to eat something.

The user said the episode lasted for roughly an hour and a half before their wife returned home. Again, this is a personal Reddit account rather than a documented medical case, and nobody should plan on leaving ChatGPT Voice running as an emergency glucose-monitoring system.

Still, there is something compelling about what happened. ChatGPT was already there, the user's condition apparently deteriorated during the conversation, and the AI kept responding to the changing situation rather than carrying on as though everything were normal.

That is where I think AI health tools could become genuinely valuable without pretending they can replace doctors. They can notice that something sounds wrong and tell them if it's a situation that requires real medical intervention.

There are plenty of safer everyday versions of medical AI help, like understanding terminology in a lab report before an appointment or organizing questions for a specialist. OpenAI's dedicated Health experience can even use health information that a user chooses to connect, including supported medical records and Apple Health data.

But restraint has to be part of the AI's efforts too. I want an AI health assistant that recognizes when answering is no longer the most helpful thing it can do. Sometimes the smartest response available to ChatGPT should be that this conversation needs to continue with somebody who went to medical school.

The stroke story stays with me because the AI apparently did something both modest and consequential by helping his family recognize that waiting was a risk they shouldn't take. In healthcare, an AI knowing when to hand the job over may be one of the most important features it ever offers.

The adoption of AI tools in the workplace shows no sign of slowing down. With 88% of businesses using AI in some capacity, companies are now highly focused on trying to measure the ROI from their investments in the technology.

Companies deploying AI tool stacks across their teams aim to boost productivity by automating routine tasks, synthesizing data and improving efficiency, reducing both costs and time-intensive labor.

But there is friction. Half of London businesses say their workforce does not currently have the skills needed to meet their organizations AI requirements. Companies are now facing a costly problem, they need a return on the investment in AI and it’s going to take a revamp of skills and job roles to do so.

Businesses need to take a head-on approach to solve this challenge by balancing technical knowledge and creativity as well as business understanding.

It is not just technical IT skills required for back-end ops, but also digital literacy, critical evaluation of AI outputs and workflow integration. These are becoming essential at every level, from entry level to C-Suite level.

Job Descriptions need updating

As the digital skills gap widens, it is unsurprising that, in the UK, one in four workers worry AI could replace them. In reality, the labor market is splitting into two distinct tiers.

The professional tier is centered on blended roles that rely on human judgement and recognizes that AI tools are only as effective as the people guiding them. By contrast, the democratized tier sees repetitive tasks handed over to AI tools entirely.

With most roles sitting in the professionalized tier, maximizing AI efficiency requires critical thinking and judgement that the technology itself cannot supply. However, if those expectations are not written into the role, employers end up overestimating the capabilities of their employees.

Nowadays, many roles have become blended with technology, and, except for technical roles, many necessary AI capabilities are often summarized into a single generic bullet point on technical proficiency. What we are seeing is a reshaping of what employers value. This reshaping is driving the emergence of creative technologists.

A creative technologist is someone who understands technology, creativity and strategy, and organizations that actively empower this role are more likely to pull off digital transformation successfully.

Today, many teams are full of creative technologists who don't realize that's what they've become. The forces driving this are generative AI becoming a daily tool, rising content demands, and increased cross-discipline collaboration.

But employers need to ensure job descriptions match this reality. A blended job description needs to name the AI tools the person will actually use, then spell out the skills needed to use them, moving things like critical thinking from generic bullets into specific expectations tied to those tools.

If employers get that right, they will be hiring people who can capture the productivity gains and not just operate the tools. AI success requires people who combine technical expertise, industry know-how and creativity.

Without more blended roles, businesses are going to struggle to see the return on investment and ensure they are using the right tools for the right job.

Encouraging trialing, testing and tryouts

Effective learning can only happen in an encouraging environment. Getting comfortable with prompting frameworks makes AI tools feel less daunting, and that comfort is often what surfaces where AI can actually help within the organization. Experimentation matters even when a given approach doesn't get pursued further. Sometimes the most useful lesson is learning what not to do.

One of the biggest barriers to that experimentation paying off is dark data. Most organizations sit on enormous volumes of information, but a large share of it lives in documents, emails, PDFs, and other unstructured formats that are hard for AI systems to interpret.

That means even well-designed models often work from an incomplete, inconsistent view of the business which leads to unreliable outputs and initiatives that stall before they scale.

Moving past that requires turning dark data into something usable that can build upon the structure, context, and governance AI systems need to act on the data with confidence. Once that foundation is in place, AI becomes far more reliable, and far easier to move from isolated pilots into real production environments.

The human edge

AI can streamline planning, but people still drive execution. As tools take on more routine workflow, organization and documentation tasks, the skills that remain most valuable are the ones AI cannot replicate such as judgement, storytelling, creative thinking and the ability to lead people through change.

Demand for AI-fluent specialists has created a new kind of empowered generalist who can read AI trends as fluently as they manage stakeholder relationships and understand customer pain points.

AI may automate workflows and documentation, but it cannot lead a team, interpret context with human nuance or fully understand a company’s mission.

Where wearing many hats within one role was once seen as a drawback, AI is making it one of the most efficient ways to work. It gives employees access to skills that would previously have taken years, and several role changes, to acquire.

Rather than simply eliminating roles, AI is also creating new ones centered on managing the human-AI relationship, including AI evaluators and human-in-the-loop reviewers.

In this sense, AI is not replacing jobs so much as reshaping them, with the potential to reduce workforce inequality by democratizing access to new skills.

The skills gap is a hurdle many organizations are now facing and both employees and employers must experiment to adapt.

We've featured the best AI chatbot for business.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

As the EU AI Act’s transparency requirements come into force, organizations need to be on the front foot when it comes to regulatory readiness and how AI is being used. Facing fines of up to €35 million or 7% of global turnover mean it’s no longer enough to just monitor communication risks. Firms must definitively prove how those risks were identified, reviewed, and escalated.

When regulators ask questions, most compliance teams don’t face the issue of a lack of data. In most cases, the information does exist somewhere within the organization. The harder challenge is bringing that evidence together quickly, clearly, and defensibly across different systems, workflows, and teams.

That challenge is becoming more important as regulators place greater emphasis on demonstrable outcomes. Firms are increasingly expected to show not only that controls exist, but how those controls operate when a potential risk emerges. This is where comms surveillance has become central to effective compliance.

From surveillance activity to supervisory outcomes

Comms surveillance has typically been measured through activity. Compliance teams have tracked how many alerts were generated, how many reviews were completed, how many cases were escalated, and how many investigations were closed.

Those metrics still matter, but they only tell part of the story.

A stronger measure of readiness is whether a firm can connect surveillance activity to supervisory outcomes. Can it explain why a particular risk was identified? Can it show how an alert was reviewed? Can it document the evidence that supported a decision? Can it demonstrate whether escalation was appropriate? And, crucially, can it reconstruct that process months later when a regulator asks for evidence?

These are the questions that now sit at the heart of regulatory readiness.

The evidence chain is often too fragmented

Most firms have already solved the basic issue of capturing communications. The next stage is more complex: connecting communications, alerts, reviews, investigations, and outcomes into a clear and defensible chain of evidence.

Too often, that chain is fragmented and therefore, obsolete. Comms often sit in one platform, while surveillance workflows sit in another, and investigation records somewhere else entirely. Each system may function effectively in isolation, but the gaps are evident when compliance teams need to explain what happened from start to finish.

This is especially pertinent as AI-driven communications become a bigger compliance risk. Our research found while more than half (55%) of organizations globally have the core AI policies, training and review steps in place, fewer than one in five (19%) have the logging, retention, detection and scoring controls needed to prove what happened.

Without that evidence layer in place, organizations lack being able to confidently demonstrate what AI tools produced, who reviewed it, where it went and whether the record was kept.

The result is unnecessary friction: investigations take longer than they should, evidence is harder to assemble, and supervisory decisions still require additional manual validation.

Regulatory requests become more disruptive when teams have to spend valuable time piecing together information that should already be connected. This risks turning into an operational readiness nightmare.

Transparency is becoming a regulatory expectation

As regulatory expectations continue to evolve, firms need to be able to demonstrate how oversight works in practice. In other words, showing how potential risks are identified, how decisions are made, how evidence is reviewed, and how outcomes are recorded.

A surveillance program is only truly effective if it can support that level of transparency. Regulators may not prescribe a specific architecture, but they do expect firms to show that oversight is effective, repeatable, and defensible.

That's why organizations are now looking beyond individual tools and towards a more unified approach to communications governance, surveillance, investigations, and retention.

Operational visibility matters as much as information storage

Operational visibility is becoming just as important as data capture. Compliance teams need to understand where information is stored, as well as how it moves through the organization once a potential issue is identified.

They need to be able to follow the full chain of activity from initial alert through to investigation, escalation, decision, and retention. Comms surveillance has become a critical part of evidence framework.

Done well, it helps teams identify risk earlier, investigate issues more efficiently and demonstrate how supervision operated over time.

Readiness is about proving what happened

The priority is now creating the context needed to make better decisions, and to evidence those decisions effectively.

Regulatory readiness increasingly depends on the ability to demonstrate judgement: why a case was reviewed, what information was considered, whether escalation was required and how the final outcome was reached. Without that connected view, even well-run processes can become difficult to defend.

Comms surveillance is therefore moving from a narrow monitoring function to a core component of regulatory preparedness. It provides the first line of evidence for how potential risk was identified, assessed, and resolved. It also gives firms the foundation they need to demonstrate that supervision is not only in place, but working effectively.

In today’s regulatory environment, readiness is all about being able to prove what happened, not simply locating the information.

We've featured the best AI website builder.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

UK small and medium-sized businesses (SMBs) have reached a defining moment. Rapid advances in technology and AI have created the potential to transform how SMBs operate – improving competitiveness, enhancing efficiency and unlocking new growth opportunities.

Yet, over a third of UK SMBs risk being permanently left behind their competitors because they lack the confidence and processes required to identify, adopt and scale new technologies effectively.

The result is an increasingly pronounced divide between businesses that feel equipped to embrace technology and those struggling to keep pace – Tech Vanguards vs Tech Bystanders. Crucially, that confidence can be self-reinforcing.

Businesses that successfully introduce and scale new technologies are better placed to make their next technology decision with confidence. Those that hesitate risk falling further behind, as the capabilities and experience needed to keep pace become harder to build.

This issue can have widespread consequences. SMBs account for over 99% of UK businesses and are the backbone of the UK economy. Therefore, if the confidence gap isn’t closed the whole country will face the consequences.

The disconnect between access and action

As technology has advanced, innovation has become widely accessible to businesses of all sizes, and many SMB leaders already understand the importance of innovation and technology adoption to their business performance. However, according to Babble’s research, fewer than half feel they’re very confident in their ability to identify, adopt and deploy new technologies successful.

There is a clear gap between knowledge and action, and this is where confidence is becoming a key competitive differentiator. The SMBs pulling ahead don’t necessarily have larger budgets or more resources. But they have built the confidence to make better technology decisions through having clear leadership, strong governance and repeatable processes in place.

These foundations matter because they decrease the uncertainty surrounding technology investments. Rather than each investment being treated as a high-stake investment, SMBs with strong leadership and repeatable processes will have frameworks for assessing where technology can add value and if it fits with their business. Successful adoption builds institutional knowledge and experience, making each decision easier.

For less confident businesses, the opposite can happen. Without established processes, technology decisions can feel riskier and more difficult to navigate, encouraging businesses to delay investment or pursue isolated initiatives that struggle to deliver meaningful returns.

More than half (53%) already feel at a competitive disadvantage compared with businesses better equipped to adopt technology, while 28% believe they'll lose new business to competitors embracing technology and 24% fear losing existing customers.

In an environment where innovation is accessible, confidence is the defining competitive advantage.

The cost of standing still in the AI era

While delaying technology adoption instead of making the wrong investment may seem like the lower-risk option, those decisions quietly accumulate over time, especially in the current AI environment.

Previous technology shifts, from the internet and mobile connectivity to cloud computing, gave SMBs time to observe early adopters before deciding when and how to follow. AI is different because it is not simply introducing another tool into the technology stack.

It has the potential to reshape how work itself gets done, taking on repetitive and manual tasks while allowing employees to spend more time on the judgement, creativity and customer relationships where people add the greatest value. For resource-constrained SMBs, those productivity gains are especially significant.

But productivity tools are only the beginning. SMB technology adoption today remains concentrated around chatbots and assistant technologies such as Microsoft Copilot, which can help individuals work faster and reduce administrative burden. But to unlock significant value, SMBs need to move beyond assistants and embed AI directly into workflows and processes.

The risk of waiting isn't simply missing out on today's productivity gains. Businesses experimenting with AI now are building the experience to understand where it works, where it doesn't and how to deploy it at scale. By the time today's hesitant SMBs decide to act, their competitors may already be several decisions ahead.

The answer isn’t more expertise

Despite the growing confidence gap, which is alarming and needs addressing urgently, there is reason for optimism. SMBs are uniquely positioned to thrive compared to larger organizations. They aren’t weighed down by decades of legacy infrastructure and complex organizational structures that can slow down decision making.

The issue for SMBs isn’t lack of ambition; it’s lack of time. SMB leaders are often balancing multiple priorities from marketing to finance and therefore don’t have the capacity to evaluate technology solutions, especially considering the speed at which new advancements are entering the market. It’s difficult for SMBs to distinguish genuine business value from marketing hype. This challenge is compounded by years of technology vendors overpromising and underdelivering.

But SMB leaders don’t need to have every answer or be a technology expert. Adoption should begin with the problem, rather than the technology itself. After all, Henry Ford’s customers didn’t ask for the car, they asked for a faster horse – knowing the problem doesn’t always mean knowing the solution.

SMBs don't need to know which platform, model or vendor will solve every challenge. They simply need to understand where friction exists inside their business. Once those questions are clear, the right partner will help identify the correct solution. Ultimately, assisting in closing the confidence gap.

Confidence starts with the leader

Closing the confidence gap doesn’t require SMB leaders to become technology experts overnight. Technology confidence is built one successful decision at a time, the aim over the next 12 months shouldn't be to transform everything.

The aim should be to create the conditions, one step at a time, for better decisions to be made consistently. That means making technology a defined leadership responsibility with clear ownership at board level and treating it as a business priority rather than something delegated to IT teams.

Competitive advantage will come from knowing where and when to invest. SMBs that establish clear ownership and repeatable processes today will be better equipped to spot emerging opportunities, make informed technology choices and scale what works. In doing so, they can build the confidence to keep adapting as technology evolves.

We've featured the best business computer.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

For the past several years, the enterprise AI conversation has been dominated by models. Every week has seemed to bring a new benchmark, a new leader board, or a new debate about which model is smartest. Organizations often made technology decisions based on incremental improvements in model performance, assuming that the model itself was the primary source of competitive advantage, but that assumption is rapidly becoming outdated.

As foundation models continue to improve, they are also converging in capability. At the same time, the cost of deploying AI at scale is becoming a more pressing concern for enterprises than squeezing out marginal gains in benchmark performance.

As a result, the focus is shifting from the model itself to everything surrounding it: the systems that make AI useful, governable, and economically viable in real-world business environments.

The question enterprises are increasingly asking is not, "Which model should we use?" Instead, they are asking, "How do we turn AI into reliable work?"

The rise of the AI harness

The answer begins with what can be described as the harness. An AI model on its own is powerful, but incomplete. To deliver business value, it needs context, access to enterprise tools and data, memory, controls, and guardrails. Together, these elements form the scaffolding that transforms a general-purpose model into a system capable of performing useful work, determining how effective an AI system becomes.

Two organizations might deploy the same underlying model yet achieve dramatically different results. One may struggle with inconsistency, high costs, and poor outcomes. The other may produce reliable, measurable business value. The difference often lies in the quality of the harness surrounding the model rather than the model itself.

That reality becomes more significant as models continue to commoditize. In many cases, organizations will have access to the same leading AI capabilities. What will differentiate them will be their ability to operationalize those capabilities efficiently and safely, and move from intelligence creation to intelligence application.

Why loops matter more than prompts

A related concept is what might be called the loop. Many organizations still approach AI interactions as isolated prompts and responses. But enterprise work rarely operates that way. Real business processes involve objectives, verification, correction, and completion criteria. A loop defines a governed unit of work.

Rather than simply generating an answer, the system performs a task, validates progress against the objective, corrects mistakes when necessary, and stops when the outcome is achieved. This seemingly subtle distinction has enormous implications.

When AI systems operate through loops instead of one-off prompts, they generate operational traces and performance records. Those traces can be used to evaluate outcomes, improve workflows, and refine future execution. Over time, this creates a learning system that grows more effective through use rather than requiring constant manual intervention.

In practice, this means enterprises are no longer deploying isolated AI tools. They are building repeatable learning systems — ones that improve the more they do the work.

The orchestration challenge

As AI adoption expands, organizations encounter another reality: there is rarely a single harness. Most enterprises operate across multiple functions, business units, and regulatory environments. Each domain may require different tools, data sources, workflows, and governance requirements.

A harness optimized for software development is unlikely to be the same one used for finance, healthcare, customer service, or compliance. This creates a new challenge: orchestration.

The orchestration layer acts as the traffic controller for enterprise AI. It routes tasks to the right harness, determines when human oversight is required, and coordinates work across multiple systems. Rather than treating AI as a standalone capability, orchestration allows organizations to manage AI as an enterprise-wide operational platform.

This capability becomes particularly important in regulated industries, where workflows often cross organizational boundaries and require strong accountability. As AI moves deeper into critical operations, orchestration will become a defining architectural requirement rather than a technical afterthought.

Governance and assurance as strategic differentiator

If orchestration determines where work goes, governance determines whether that work can happen safely and assurance determines whether we can trust it was done as intended. Many organizations initially viewed AI governance and assurance as a compliance exercise rather than a core operational requirement.

Effective governance and assurance encompasses policy enforcement, authorization controls, asset management, cost monitoring, evaluation and audit services, observability, guardrails, and risk management. These capabilities must function consistently whether work occurs within a single AI system or across multiple interconnected workflows.

The need is especially acute as enterprises deploy autonomous and semi-autonomous systems. A model generating an inaccurate answer may be inconvenient. An AI agent taking action in a regulated business process without adequate oversight can be more problematic. Organizations that treat governance and assurance as an integral component of their AI architecture rather than a bolt-on feature will be better positioned to scale adoption while maintaining trust.

Thinking in ecosystems, not products

Collectively, the harness, orchestration capabilities, and governance and assurance layers form what can be viewed as a governed operating environment for AI. Combined with models, data, and compute infrastructure, they become an ecosystem that functions as a single accountable system.

This shift in perspective is important because no single vendor currently delivers every component of that ecosystem. The future of enterprise AI is therefore unlikely to be dominated by one model provider or one platform vendor. Instead, success will depend on how effectively organizations integrate and operate a diverse collection of technologies.

For enterprises, particularly those operating under regulatory, security, or sovereignty requirements, this integration challenge is not optional. AI must operate within boundaries defined by business policy and organizational control. The ability to combine frontier models with protected enterprise data while maintaining accountability will become a critical capability.

The next AI battleground

The model layer will continue to improve. It will also become increasingly interchangeable, but that does not diminish the importance of AI innovation. Instead, it changes where value is created.

The next battleground is the layer surrounding the model: the harnesses that make work reliable, the loops that make systems learn, the orchestration platforms that coordinate activity, and the governance and assurance frameworks that make scale enterprise ready. Together, they will determine whether AI remains an impressive technology demonstration or becomes a sustainable enterprise capability.

In the years ahead, the organizations that gain lasting advantage are likely to be those that stop asking which model is best and start focusing on how to integrate, govern, and operate AI as an accountable business system. The future of enterprise AI belongs not just to the model alone, but to everything around it.

We've featured the best AI website builder.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

AI tools are now a normal part of the working day. Employees lean on AI for tasks ranging from drafting emails and summarizing content, to preparing legal documents and creating financial reports but, while there’s no doubt that AI and automation are making our workdays more efficient, the more we rely on these tools, the greater the risk of them being exploited, misused or simply trusted too much.

In 2023, US lawyers were sanctioned after filing legal documents built on false citations that were invented by ChatGPT. If experienced professionals are trusting AI outputs without verifying primary sources, even in high-stakes legal proceedings, it raises concerns about our growing dependence on AI and the risks of unchecked and hallucinated data.

But hallucinations are only part of the problem. As employees become more comfortable using AI at work, organizations also need to understand what information employees are feeding into AI engines and whether they have the right controls in place to manage the security, privacy and compliance risks that come with unsanctioned AI use.

Ungoverned AI is no longer a theoretical compliance risk. It’s a real issue that CISOs are having to deal with, from protecting consumer and employee data to defending against AI-enabled attacks. As AI regulation moves from guidance to enforceable requirements, organizations are facing more pressure to understand where AI is being used and who’s responsible for its outputs.

The cost of AI without governance

Ungoverned AI use poses significant risks, particularly when employees share sensitive data with AI tools without clear guidance on how it should be used. That information could then influence business decisions based on unverified AI outputs or fall into the hands of a bad actor looking to exploit the business.

Once sensitive information leaves an organization's control, there’s no telling where it could end up. If an AI platform is breached, data entered into the tool could be exposed, potentially giving attackers exactly what they need to target a company’s systems. What starts as a quick shortcut to save time can just as easily become an entry point for credential theft or serious financial loss.

The EU AI Act’s transparency requirements, applicable from 2 August 2026, expects organizations within scope to be transparent about their business’ AI interactions and AI-generated or manipulated content. It also requires staff to be equipped with an appropriate level of AI literacy, alongside existing regulations such as GDPR and relevant FCA requirements that involve the effective governance of AI processing of personal data. The cost of breaching these transparency requirements can carry fines of up to €15 million or 3% of an organization's total annual turnover.

This means that now more than ever, organizations need to understand exactly how AI is being used, what information is being entered into AI tools and what controls are needed around higher-risk use cases, especially where personal data and financial decisions are involved. This is where AI verification protocols become critical.

Why verification isn’t a one-size-fits-all

A generic AI policy that isn't tailored to specific use cases will tend to get ignored. Employers telling staff to simply "check everything" they receive from these tools isn't realistic or clear enough guidance, since not every AI-assisted task carries the same risk.

Scrutiny needs to scale with what's at stake if the output turns out to be wrong. By treating every task the same, you either take up valuable employee time on unnecessary verification processes or risk a potential security incident because an AI-generated output is moved forward unchecked.

For low-risk tasks like internal drafting, brainstorming and research support, light spot-checking can be enough. However, anything feeding into decisions about employment, finance or customer-facing content needs strict approval processes, ongoing monitoring, and a genuine human in the loop.

Practical steps to reduce the AI hallucination risk

Reducing AI cyber risk starts with proper governance. Organizations need clear rules on which AI tools are approved, what information employees can share within the platforms, and who’s accountable when something goes wrong. Waiting until a hallucinated output causes damage is the most expensive way to learn this lesson.

It’s becoming increasingly common for employees to bring their own AI tools into work. Our latest research found that 40% of CISOs fear their staff are sharing sensitive information with generative AI platforms, while 68% identify employees as their organizations biggest cyber risk. Ungoverned AI is becoming a growing governance and security blind spot for organizations .

But the risk goes far beyond what employees put into AI tools. When organizations don't know which tools are being used or to what extent employees are relying on their outputs, they have limited visibility into whether AI-generated information is being factchecked before it’s acted upon.

Organizations can't manage risks they don't know exist, which means mapping where ungoverned AI is genuinely being used and then offering secure, approved alternatives, rather than banning it completely. This gives organizations greater visibility over how AI is being used and where hallucinations or inaccurate outputs could enter business processes, before they become a liability.

Generic training won’t fix the AI problem

Employees need to be equipped with the right processes, tools and guardrails to use AI effectively, without putting businesses at risk. This includes training to recognize automation bias (our tendency to trust outputs blindly), paired with explicit verification routines that turn "double-check the AI output" from a vague instruction into an actionable process for employees to feel in control of how they can use AI responsibly.

Role-based security awareness targeting and behavior-triggered interventions are also critical for putting the right guidance in front of the right people at the moment they need it, so that sensitive information is protected, and the right human judgement is used.

Ultimately, addressing AI hallucination and the risk of ungoverned AI comes down to training employees properly. Most CISOs (81%) say that security awareness training fails because it’s too generic to feel personally relevant. Without context that reflects the decisions employees face day to day, training gets completed but rarely absorbed and actioned on.

Checkout our ratings for the best payroll software in the UK.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Launching a business has never been easier. AI tools can generate brand names, build websites, create logos, and produce marketing copy in hours.

It's a remarkable shift for entrepreneurs - but the same technology is also making it easier for cybercriminals to create convincing digital impersonations at scale.

Recent allegations by Google claim cybercriminals used AI to generate 1.5 million malicious URLs designed to mimic legitimate organizations.

As AI lowers the cost of creating convincing websites, businesses face a new challenge: proving they're real.

That's why conversation around domains is changing. The hardest part of launching a business is no longer getting online; it's establishing a distinctive brand that customers can recognize, trust, and confidently return to.

The problem isn't .com's relevance but its availability

For decades, securing a .com domain was considered the first milestone of establishing an online presence. It became the default because it was familiar, widely recognized, and trusted.

That hasn't changed. A .com domain still carries value and remains the first choice for many organizations. What has changed is availability.

According to The Domain Name Industry Brief's (DNIB) latest quarterly report, there are more than 166 million .com domains registered with domain registrars. Many short, memorable .com domains are already registered, held defensively, or owned by investors.

Businesses launching today might discover that their ideal domain has long since been taken, forcing them to settle for longer names, additional words, or unconventional spellings that can make brands harder to remember.

This doesn't signal the decline of .com. Instead, it's encouraging organizations to become more intentional about how they establish their online identity. Rather than focusing solely on securing a .com address, businesses are evaluating whether their domain reflects their brand, communicates what they do, and supports long-term recognition.

Organizations are looking beyond convention and exploring domain extensions that better align with their brand, industry, or audience. Not because .com has lost its value, but because a memorable and distinctive online presence has become just as important.

Domains have evolved from web addresses into brand assets

A domain name has traditionally been viewed as a technical requirement; a web address customers type into a browser. Today, it’s much more than that.

A domain shapes the first impression customers have of a business. It appears in search results, LLM citations, email addresses, advertisements, social media profiles, and digital marketing campaigns. It is one of the few digital assets a business owns.

Businesses invest heavily in building audiences across social media platforms, but those platforms ultimately control the experience. Algorithms change, policies evolve, and visibility can change overnight. A domain, however, provides a stable identity that remains under the business’ control.

This shift is changing how organizations approach domain strategy. Businesses are asking whether it reinforces their brand, communicates what they offer, and supports long-term recognition.

Alternative domains have become part of that conversation. A technology startup may choose an .ai domain. An ecommerce retailer might adopt a .store extension. Creative agencies, consultants, and online communities are similarly exploring domain names that better align with their industries and audiences.

These choices are less about following trends and more about improving clarity. A well-chosen domain can communicate something meaningful before a visitor even reaches the homepage.

Digital identity is becoming a competitive advantage

AI has accelerated more than business creation. It has also made it easier to imitate legitimate organizations.

Generative AI simplifies the creation of convincing fake websites. Professional-looking branding, realistic copy, and cloned user experiences can now be produced at a scale that would have been difficult just a few years ago.

As a result, businesses face growing pressure to make their digital presence unmistakably authentic. A customer who lands on the wrong website may not have another opportunity to determine which business is genuine. In this era, a domain is not just a destination for customers. It's part of a broader trust strategy.

Business owners should think beyond individual websites and consider their entire domain portfolio. Relevant brand variations, country-specific domains, and campaign domains all contribute to protecting digital identity, while forgotten domains, expired microsites, or unmanaged registrations can create opportunities for phishing, typosquatting, and brand impersonation.

Internet Corporation for Assigned Names and Numbers (ICANN) recently closed their first application opening for new generic top-level domains since 2012, which means more TLDs will be available for business owners and impersonators. It's not enough to focus only on your primary business domain. Businesses need to consider monitoring registrations and new TLDs and understand where their brand sits within the domain landscape and decide when and how to protect their names.

Protecting online identity is no longer solely the responsibility of cybersecurity teams. It requires coordination across marketing, IT, legal, and security as organizations work together to protect customer trust.

More choice creates more opportunity

The growing range of domain extensions is giving businesses greater flexibility in how they establish their online identity.

For many businesses, .com remains an important part of their domain portfolio. But businesses are recognizing that different extensions can serve different strategic purposes.

A global organization may continue using its .com as its primary corporate website while adopting other extensions for regional initiatives, product launches, marketing campaigns, or specialized services. Startups may choose an extension that more reflects their industry while maintaining consistency across their broader brand.

The conversation is shifting from finding the right extension to building the right domain strategy.

As businesses expand into new markets, introduce new products, or adopt emerging technologies, their digital presence should evolve alongside them. The growing diversity of domain extensions gives organizations more tools to support that evolution.

Owning your identity matters more than ever

A .com domain is still one of the internet's most valuable assets, and there is little reason to believe that will change anytime soon.

The bigger shift is how businesses think about domains. AI can help launch companies faster, but it also helps malicious actors imitate them more convincingly. That’s why domains are more than just navigation tools. They also prove authenticity.

The future is more than choosing between .com and newer domain extensions. It's also about being intentional - building a domain strategy that helps customers recognize the genuine business while supporting long-term growth and resilience in an AI-driven internet.

To put this into practice, businesses should prioritize:

•Monitoring for typosquat and lookalike domains

•Auditing their domain portfolio regularly for unused domains

•Registering brand domains across other reputable domain extensions

•Maintaining an accurate domain inventory

•Making legitimate customer communications easy to verify

Together, this strategy can help protect a brand’s reputation online, no matter what domain extensions are used.

The best website builders: 80+ platforms tested to find the easiest ways to build a site.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Social media has become central to how businesses communicate with customers, promote their brands, and build trust. But as businesses have relied more on these channels, they have also become a more attractive target for cybercriminals.

New research has found that social media impersonation and defamation have risen from the fifth greatest cyber threat last year to the top expected cybersecurity threat over the next three years. The data also revealed that employee and executive impersonation, including deepfakes, has entered the top five areas of risk for the first time.

These findings point to a wider challenge for security teams. Cybercriminals are still trying to compromise systems, but they are also increasingly using other ways to exploit legitimate brands’ hard-earned trust. That trust might sit in a senior leader’s online profile, a brand’s social media presence, a customer service account, a domain name, or a website that looks legitimate at first glance.

For security leaders, this means the attack surface is no longer defined only by infrastructure they own and control. It also includes the public-facing channels where customers, employees, and partners interact with the company every day.

Security strategies now need to reflect that reality. A threat may start with a fake profile, but it rarely stays there.

Social media and identity have become frontline cyber risks

Fake profiles and fraudulent customer service accounts are not new tactics, but they are becoming more prevalent and are increasingly being used as entry points for phishing, fraud, counterfeit sales, and wider brand abuse.

And this is all happening at a time when cybersecurity teams are already under pressure. Recent research found that 72% of senior technology leaders said the level of cybersecurity threats faced by their organization in 2025 was either “critical” or “very critical”.

On social media, criminals can meet potential victims where they are already used to engaging with brands. Fake accounts can be used to share malicious links, promote fraudulent schemes, spread false information, or pose as legitimate support channels.

Attackers may then direct users to a “second location”, such as a lookalike website, fake login portal, or phishing page, where victims are encouraged to enter account credentials, payment information, or other sensitive data.

A separate but related risk is counterfeiting. Fraudulent social media accounts can impersonate legitimate brands to advertise fake goods, directing customers to websites where counterfeit products are presented as genuine. In both cases, criminals are using the trust associated with a recognized brand to make the next step of the attack appear credible.

The impact can also go beyond fraud and counterfeiting. Impersonation or fake accounts can be used to spread false or defamatory claims about a company or its employees, potentially damaging reputation and customer confidence even when no direct financial fraud takes place.

As such, social media abuse, domain impersonation, and identity-based fraud should be treated as part of the same threat landscape.

AI is making impersonation faster and more convincing

AI is making impersonation faster and more convincing. Criminals can use AI tools to imitate a brand’s tone of voice, generate convincing customer messages, produce realistic imagery, or create more credible fake profiles at speed.

Deepfakes, synthetic audio, and AI-generated content can also strengthen employee or executive impersonation, particularly when combined with information gathered from multiple sources. For example, an attacker could create a fake executive profile and use AI-generated messages that mimic how that individual communicates, making requests or links appear more credible.

But AI is only one part of the threat. Attackers can also use established techniques to build the infrastructure behind impersonation campaigns. Domain generation algorithms (DGAs), for example, can produce large numbers of plausible lookalike web addresses to support phishing and impersonation campaigns. DGAs are not inherently AI-powered, but they remain a concern for security leaders, with 86% of respondents viewing them as a threat.

These domains can then be combined with fake social media profiles to create a more convincing digital presence. A fake executive account may point to a fraudulent landing page. A counterfeit product post may direct customers to a lookalike domain.

For customers and employees, these attacks are becoming harder to identify based on appearance alone. The signs of fraud may be subtle, and by the time an impersonation attempt is reported, the campaign may already have moved to another account, page, or domain.

Businesses need a more connected approach to protecting trust

Social media impersonation should be treated as part of the wider cybersecurity strategy, rather than as a standalone brand or communications issue. That requires coordination between the teams responsible for protecting organizations, their people, and their customers.

Attacks often move from a fake profile to a lookalike domain, phishing page, or fraudulent website, so companies need a connected view of where threats are emerging. Social media teams may spot one warning sign. Security teams may see another. Legal or brand protection teams may hold the takedown process. A coordinated approach can help companies respond more effectively.

Alongside those measures, organizations need continuous monitoring across social media, domains, websites, and other public-facing digital assets to identify impersonation, brand abuse, and related threats occurring beyond infrastructure they directly control. Bringing these signals together can help teams recognize when activity across different channels forms part of the same campaign.

Clear processes for investigation, escalation, and takedown are essential, particularly as AI enables campaigns to spread, adapt, and reappear rapidly. Businesses need to know who owns the response, what evidence is needed, and how quickly action can be taken.

Technology will also play a growing role. More than half of respondents, 57%, confirmed that they use AI-based monitoring and enforcement solutions, while 44% use AI-based solutions for threat detection and fraud prevention. Given the 24/7 nature of the internet and how quickly malicious accounts and materials can appear, some level of automated detection is increasingly important for identifying threats at scale.

This is an important step, but tools alone will not solve the problem. Companies need the right governance, clear ownership, and close collaboration between security, legal, marketing, and digital teams.

The goal should be to detect and stop impersonation before it reaches customers, employees, or partners and causes wider damage.

Looking ahead

Social media is no longer simply a communications platform that sits outside the traditional cybersecurity perimeter. It has become one of the main environments in which criminals exploit trust, impersonate legitimate people, and manipulate victims into falling for fraud.

As AI makes these attacks faster and more convincing, businesses will need to connect social media protection with their broader approach to identity, domains, and digital brand security. This means monitoring for threats across the full online journey, sharing intelligence between teams, and responding before an impersonation campaign has time to spread.

The companies best equipped to protect customers will recognize that trust itself has become a target and treat the channels where that trust is built as a core part of their cybersecurity strategy.

We've featured the best antivirus software.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

The Voice AI Agents market is growing exponentially. The industry is projected to grow from $2.4 billion in 2024 to $47.5 billion by 2034, while Gartner predicts that by 2028, 70% of customer service journeys will begin with a conversational AI interface.

For global brands, the appeal is obvious. Voice AI offers the potential to deliver consistent, scalable customer support across markets, languages and time zones.

But there is a risk that voice AI will face the same issue faced by outsourced customer service centers, where differences in language and culture create a disconnect and frustration between the brand and customer.

As voice AI advances, language may become less of a barrier. But speaking the same language is not the same as understanding someone.

It’s just one part of the puzzle. Customers bring accents, habits and cultural expectations to every interaction, and they expect support to reflect their reality.

For that reason, localization needs to become a core test of voice AI performance, rather than a translation exercise that happens at the end of development.

What does it mean for voice AI to be ‘local’?

Accents, dialects, terminology and cultural norms can vary considerable even when speaking the same language. In Scotland, for example, a customer might say “aye”, rather than “yes”, refer to something small as “wee”, or talk about “getting the messages” when they mean going shopping. The language is English but understanding the interaction requires familiarity with how that language is used locally.

Cultures also differ in their expectations around directness or politeness. Even something as simple as how a customer is addressed can matter. In some cultures, using titles and surnames is an important sign of respect, while in others, addressing someone by their first name is the norm. A voice agent that is too informal could come across overly familiar, while one that is too formal might feel distant or unnatural.

All this to say that just because an AI might sound like me, it doesn’t necessarily mean it understands me or is equipped to serve someone in my market.

These aren't cosmetic details. They shape an AI agent’s ability to understand intent and respond appropriately and, ultimately, determine whether customers trust the experience it provides.

It doesn’t sound like a robot anymore

This isn’t to say there haven’t been huge advances in AI. There’s a reason why so many brands are investing in voice AI for CX, and that’s because it's becoming so convincing. Just look at ElevenLabs, for example.

The organization has built its success on AI-generated voices that can reproduce not just speech, but the pacing, intonation and emotion that make a voice sound human. Its technology not only spans voice cloning, multilingual speech and conversational agents, helping take synthetic voice from a novelty to something businesses can realistically deploy at scale.

The days when synthetic speech was instantly recognizable as robotic are fading. Early digital voices such as ‘Microsoft Sam’ left little doubt that you were listening to a machine. Today, the line is becoming much harder to distinguish. As voice AI becomes more convincing, our interactions with it can start to feel less like using a piece of technology and more like engaging with another person.

The technology's ability to create a sense of social presence is already apparent beyond customer service. Even actors such as Matthew McConaughey are getting behind AI voice technology, where it can match their performance without requiring them to change the way they sound, demonstrating just how powerful voice and conversational cues can be in shaping our perception.

But the most advanced experiences aren’t universally available, meaning many people may still be exposed to underdeveloped AI voices. What’s more, humans are remarkably good at noticing when something is slightly off. This is what creates the uncanny valley effect.

An unnatural pause, misplaced emphasis or strangely enthusiastic response can quickly break the illusion. Local context adds another layer to that challenge. A voice can sound perfectly human while still sounding culturally out of place. That said, the closer AI gets to human speech, the more conspicuous those moments can become.

Global brands can’t rely on one global answer

This is where global brands need to resist the temptation to treat voice AI as a single product that can be rolled out everywhere.

A frontier model might provide the foundation, but its performance needs to be tested against local data and real customer interactions so it can capture what ‘good’ looks like in terms of:

- The empathy that de-escalates a frustrated customer

- The specific terminology a market actually uses

- The intonation that reads as natural

All of these can't be inferred by a frontier model; it has to be learned market by market, brand by brand. Successful voice AI models will have an ongoing feedback loop which monitors interactions within individual markets, identifies where misunderstanding or friction occurs, refines the experience and tests again.

Local teams can play an active role in this process as well, bringing market expertise into development rather than simply receiving the “finished product”.

It can’t be treated as a job that is ever truly finished. Language and slang changes, customer expectations shift and new trends introduce expressions and behaviors that may not have appeared in the original training data.

Final summary

Getting this wrong can have an impact on a brand’s reputation. As voice AI becomes more human, customers will naturally expect more from it.

Derby City Council recently experienced this with its AI assistant, Darcie, which struggled to understand a presenter with a strong Derbyshire accent who used local expressions such as “mardy” and “duck,” despite having been upgraded to support nine additional languages.

It is a useful reminder that multilingual does not necessarily mean localized. If a system repeatedly misunderstands an accent or fails to recognize local expressions and expectations, customers will not experience it as a technology problem.

Instead, it will shape their perception of the organization behind it. The brands that get real value out of the voice AI boom won’t be the ones whose technology can speak all 7,170 languages in existence today. They will be those whose customers feel the most understood.

We've ranked the best translation software.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Known as the 'Godfather of AI' for his work throughout the decades in pioneering the underlying architecture behind the intelligent systems we use daily, Nobel Prize-winning computer scientist Geoffrey Hinton has added his voice to the already saturated chorus projecting a possible AI doomsday scenario within our lifetime.

Existential risks

Speaking in a CNN segment, Geoffrey Hinton was asked if AI could really destroy humanity, as many other voices have posited over the years – and especially in recent weeks.

Hinton's work was instrumental in devising the underlying neural network architectures that today's generative AI models rely on. He was recognised for his efforts in the 1960s with the Nobel Prize in 2024.

Hinton laid the groundwork for the systems many of us use today, but he has repeatedly sounded the alarm over the threat that his technology may one day pose. He previously spent time at DeepMind before leaving in 2023, warning then that AI could pose an existential threat to humanity in several ways.

Confidence artists

Hinton's comments came in response to several major security incidents in recent months, announced by a variety of tech companies and frontier AI labs at around the same time.

The first documented case of AI infiltrating another company's systems was internal OpenAI models breaking out of their sandbox and accessing the internet, subsequently breaking into HuggingFace's internal systems to retrieve the answer to a query.

Since then, several organizations have been quick to confirm their models have also breached other company networks unprompted. Now the dust has settled, critics point to these announcements being part of a broader marketing campaign to raise the profile of AI models – with the companies using the unfortunate (usually human) errors to demonstrate the potentially dangerous capabilities of these systems.

RGB mini-LED (and its other various names) continues to be the talk of the TV world, with nearly every brand getting involved with the new tech in 2026. I’ve tested several RGB mini-LED/Micro RGB TVs and while it’s been a mixed bag, the new tech shows promise. While these TVs still can’t quite beat the best OLED TVs for the place at the top of the TV food chain, there’s certainly been some impressive RGB models.

One of those is the flagship Samsung R95H, which earned 4.5 stars out of 5 in my recent review. Another is the ‘mid-range’ Sony Bravia 7 II. When I compared both the R95H and Bravia 7 II, I was impressed by the fight the much cheaper Bravia 7 II put up: especially with its picture quality.

Well, now the Bravia 9 II, the Bravia 7 II’s step-up sibling and Sony’s flagship RGB TV, has arrived in our testing lab and since I still have the Samsung R95H in, I thought I’d compare the two to see which flagship RGB model is the top dog.

A quick note on photos: the R95H appears to have a red tint in some images below, but this is the camera interacting with the R95H's matte screen. This red tint does not appear in-person.

Color

The big selling point of RGB is its color reproduction. And as expected from two flagship models, colors look very impressive on both the Bravia 9 II and the R95H. Watching The Super Mario Galaxy Movie, any scenes involving the Lumas showcases RGB’s excellent colors. While the R95H demonstrated higher color luminance, meaning color looked brighter, the Bravia 9 II’s colors appeared much richer and deeper, giving the Lumas themselves a perceived crisper texture.

As for the character’s outfits throughout the movie, it varied from scene-to-scene which TV produced better colors. Peach’s pink dress benefitted from the R95H’s higher perceived brightness in some scenes, while in others, the Bravia 9 II’s bolder colors made it pop more. The same was true with the reds of Mario’s overalls and Luigi and Yoshi’s greens.

Switching to a real-world movie, Speed Racer, it was more of what I’d found when watching the Mario Galaxy movie. During the breakfast scene at the Racer home, I found some colors, such as yellow, looked better on the Bravia 9 II with the added color depth, while some paler colors such as orange looked more natural on the R95H. The R95H’s higher perceived brightness definitely made colors more impactful, but overall I found myself drawn to the Bravia 9 II’s deeper color profile.

In terms of skin tones, both TVs showed excellent accuracy, making characters on screen look lifelike. In this scenario, I did find the R95H’s brightness made people look more true-to-life, which I preferred. But the Bravia 9 II did use shadows and contrast in an interesting way to make people look dynamic.

Contrast

One of RGB’s other selling points compared to other backlit TVs is better backlight control and OLED-rivaling black levels and while this has been a mixed bag in my previous testing of other RGB TVs, these two flagship models demonstrate excellent contrast and backlight control.

Watching The Batman, both TVs demonstrated strong contrast. As Bruce works in the Batcave at his desk, the light tones of the bright monitors and desk lamps around him contrasted well with the dark tones of the environment and the crevices around the desk. While the R95H’s higher brightness meant some dark areas were more legible, the Bravia 9 II demonstrated richer blacks that created stronger perceived contrast.

Although both TVs were able to render the often challenging picture of The Batman, due to its low brightness, in bright conditions, it was definitely best viewed in ambient or dark, home theater conditions. I also found the R95H’s Movie mode could over-do its brightness, resulting in clipping of bright objects. For The Batman, it was best to use Filmmaker Mode. The Bravia 9 II looked great in Movie mode and Professional mode (its name for Filmmaker Mode), although it lost some of the impact in its contrast when viewed in the dimmer Professional mode.

Switching to Project Hail Mary, both scenes delivered inky black tones during exterior shots of space, as well as demonstrating strong shadows as Ryland Grace explored the Hail Mary. I did find that the Bravia 9 II’s local dimming could be aggressive in places, resulting in some black crush in certain scenes. Still, black tones and contrast were incredibly impressive for a backlit TV, and the lack of blooming showed just how good the Bravia 9 II’s backlight control. Though not as impressive, the R95H also showed brilliant backlight control: much better than most backlit TVs I’ve tested.

While the same movies still had more dynamic contrast on the OLED TVs I’ve tested such as the LG G6 and Samsung S95H/S99H, this is certainly a big step in the right direction for backlit TVs.

Brightness and reflections

Starting with brightness measurements, I was surprised by how big the gap was between these two TVs: and which was brighter based on testing. The R95H hit 1,977 nits and 612 nits peak and fullscreen HDR brightness respectively in Filmmaker Mode, while the Bravia 9 II registered a much higher 4,414 nits and 974 nits peak and fullscreen HDR brightnrss respectively in its Professional mode. This is a much wider gap than I expected, made all the more interesting on how these TVs rendered brightness in practice with real-world movies.

In the majority of scenes I watched, the R95H delivered higher perceived brightness in both large areas and highlights in high-contrast scenes, such as the lights in the Batcave in The Batman, despite its much lower measured brightness.

Watching a real-world bright scene from Lawrence of Arabia, as Lawrence and his guide were fetching water from a well in the desert, the white sand around them was brighter on the R95H than the Bravia 9 II. HDR highlights in other movies, such as the white paint job of the Mach 5 in Speed Racer or the lights of all the consoles and switches of the ship in Project Hail Mary were closer in brightness and would sometimes alternate on which was brighter but for the majority, the R95H was much more impactful.

This is something I've found in the past when measuring TVs: measurements don't always match up in practice, and I'm noticing that RGB TVs tend to have the biggest difference between results and real-world application. Still, both TVs did deliver impressive brightness overall.

One area where both these TVs really impressed me was in their reflection handling. The R95H obviously uses Samsung’s Glare Free matte screen, which has proven to be incredibly effective at eliminating mirror-like reflections. The Bravia 9 II is also equipped with an anti-reflective screen, and it’s a big improvement over the step-down Bravia 7 II. Watching darker scenes, I was surprised at how little mirror-like reflections were on the Bravia 9 II.

Which TV is the best buy?

Both the Bravia 9 II and R95H are fantastic TVs. They show a clear evolution of what is capable from a backlit TV and while again I still think OLED is the best TV panel tech, it’s a positive sign of what’s to come from RGB if this is how it performs in its infancy.

When it comes to which TV is best out of the two, I personally preferred the Bravia 9 II’s picture, thanks to its richer, OLED-like colors and richer black tones. The R95H has the brightness battle won and delivers amazing pictures in its own right, but for me the Bravia 9 II takes what I loved about the Bravia 7 II and builds on it.

That being said, price is a factor. For a 65-inch Samsung R95H, you’ll be looking to pay roughly $2,799 / £2,799 / AU$4,495 (though I’ve seen the R95H drop to as low as AU$3,499 in Australia). A 65-inch Sony Bravia 9 II will currently set you back $3,499 / £3,499 / AU$5,999. Considering the R95H is actually packed with more features for gaming, such as four HDMI 2.1 ports, and has a better interface, the extra money is a tough ask.

At the moment, the R95H is the better buy, but if the Bravia 9 II’s prices can be more in-line with its rival, then the Bravia 9 II’s picture is worth the sacrifice of some features. That being said, I’d still recommend OLED right now, as you can get flagship OLED’s from both Sony and Samsung that are the same price or cheaper.

Thinking of buying a new TV?

Try our TV size and model finder! You tell it how far you sit from your TV, we'll tell you what size to buy based on viewing angle advice from image quality experts, and we'll recommend our three top TVs at that size for different prices.

Normally we get to this time of year and the typical advice is to hold fire on any potential tech purchases. This is because Black Friday is just a couple of months away, and so it always seems prudent to think about being patient, and the bargain you might just secure in the biggest tech sale of the year.

Not this year, though, when it comes to GPUs – at least in my opinion.

During Black Friday 2025, we did see some decent deals on graphics cards. Not many, mind, but there was a scattering of them. I've noticed on Reddit that there are some gamers who have found a GPU with a decent price tag, but they're wondering whether to hold off and wait for something better on Black Friday. They are perhaps remembering some of the discounts that happened last year (thin on the ground as they were).

For example, this US-based Redditor asks: "Buy an open box [Nvidia RTX] 5070 now or wait until Black Friday?" They add: "I see a few open box deals for $745. What do people think about snagging this now at such an inflated price, or waiting until Black Friday to try and snag under $700?"

If you're in a similar quandary, my advice is this: absolutely do not wait for Black Friday. Just buy that graphics card now.

Let's break down the reasons why this isn't a 'waiters' market (by which I mean those biding their time, not the lovely people who serve us food and drinks).

Reasons to be fearful

For starters, as a legion of cynical Redditors are quick to point out, Black Friday isn't the best for graphics card deals anyway – it has never been the strongest suit for this sale in terms of PC components.

As noted, there have been bargains in the past though. Despite those Redditors who insist that GPUs just have their prices quietly upped in October so they can be artificially 'discounted' for Black Friday, there are actually some genuinely decent deals – or there have been.

However, things have just gone crazy this year, and as the second half of 2026 has rolled around, the RAM crisis has reached another tiring peak of frenzied hiking.

There were more rumors of imminent GPU hikes just a couple of weeks ago (following last month's negative chatter), and at retailers in the US we've already seen graphics card prices rising considerably over the past month. Compounding this are even more negative predictions about the memory shortage in general – which inevitably affects video RAM as well as system memory – and DDR5 prices which are spiking in a ridiculous manner right now. So you've got a recipe for a tumultuous rest of the year.

In short, I can't see 2026 panning out any other way than seeing more GPU price increases – so why wouldn't you buy now? At least from the perspective of whether Black Friday might somehow bring some bargains around – it surely won't.

I'm pretty convinced of that, and there are nuances here in terms of what kind of graphics card you're looking to buy. I'm talking about mainstream GPUs, the mid-range or lower tiers of the market, and not the higher-end graphics cards.

The beefiest GPUs like Nvidia's RTX 5080 (don't even mention the RTX 5090) are already exceptionally pricey, and while I don't think they'll get any cheaper – and indeed I reckon they'll probably still creep up in price – the cost of those products is already a very bitter pill to swallow.

I don't blame you if you don't want to pay out that much cash for a graphics card, because over $1,500 for an RTX 5080 (in the US) is a pricing nightmare. But you'll have to accept that you'll likely be in for a long wait for better prices, maybe not until 2028. A compromise might be to buy a card like an RTX 5070 to tide you over, or maybe an AMD RX 9070 XT (it's a touch less expensive in the US at the time of writing), then sell that second-hand later on when the RTX 5080 comes down.

The trouble is, there are no good options here... not for the pricier GPUs anyway.

Red Friday

I think that this year, Black Friday should be renamed Red Friday for its Herring-like nature, at least when it comes to graphics cards. Don't be swayed by the temptation of waiting for a deal, and if you see a reasonable price on the GPU you're considering, grab it now.

A $750 (open box) deal on an RTX 5070, the dilemma of the aforementioned Redditor, seems a sound enough proposition, even though that's a depressing reality considering this GPU was still going for around $650 just two months ago – because you're now looking at $850 or so (going by Newegg US pricing).

You can pick up an RTX 5060 Ti 8GB for only 10% more than this GPU was priced at in June (again, going by Newegg US), and I think there's an even stronger case to snag something like this if you're looking at the more affordable end of the market.

Obviously, do your own research based on the regional pricing for the GPU you're thinking about, but whatever the case, I'd still urge you not to be fooled into waiting for Black Friday, as I really think that'd be a mistake.

What happened to Jensen Huang? The Nvidia co-founder and CEO seems about as disconnected as a person can be from other humans' life experiences. That wouldn't be so concerning if it weren't for the fact that he runs the most valuable company in the world and is responsible for the hardware behind the most life-altering innovation in our lifetime: AI.

Huang, who founded Nvidia 33 years ago, has been a CEO for decades and a billionaire since 1999; he's also now firmly on the side of rapid AI development and deployment with little-to-no regulation, putting Huang firmly at odds with a growing legion of, admittedly, some other billionaires and CEOs — among others — who are calling for a foundational model development pause or at least slow down.

It's in this moment that Huang has been talking — a lot. He's sat down with multiple journalists for chats, but none now more notable than the extensive New York Times podcast with columnist Ezra Klein.

Unlike some of the other interviews, Klein used Huang's own description of the AI as a Five-Layer Cake: Applications, Models, Infrastructure, Chips. Energy to frame the conversation, which invited Huang to opine on all these critical AI bits, and I have to say, many of his comments were eye-opening.

Here are the most startling things Huang said and maybe why he's saying them:

'There are a lot of skills that don’t matter'

"There are a lot of skills that don’t matter."

The topic here was studies in China showing that while AI in education may initially help students work more efficiently, it, on average, lowers their test scores by almost 20%.

Huang agrees with Ezra that the use of AI is, while helping them work faster, possibly degrading student performance, at least in certain skills.

"The multiplication table is starting to be forgotten. Doing square roots, my goodness. Basic math is being forgotten. Does it matter?...Yeah. I don’t think it does. I don’t think it does."

Huang believes that even as we lose some skills, or as students become adults entering the workforce, they will gain other new skills. Of course, that's sort of a zero-sum game where you can easily replace one thing with another.

I don't know about you, but I do consider the multiplication tables to be a core skill and useful even if you don't work in a math-related field. Naturally, Huang and most AI providers would like to believe that, like the calculator before it, AI will handle this skill and do all the math for you. Perhaps. But at what point do we lose the ability as a culture to double-check AI's work?

'I actually don’t know my address'

This next one is actually connected to the prior quote but is worth calling out:

"My first confession, I actually don’t know my address."

Not his email address, not his phone number or his partner's phone number, but where he lives. The street number and zip code (I'll assume he knows the town and state).

Huang used this as an example of a skill he no longer needs, but admits that realizing it while pumping gas and needing his zip code (I'm guessing for credit card verification) panicked him.

Listening to this while driving my own car, I almost pulled over. "What?!" I yelled at my AI-filled iPhone 18 Pro Max, which was playing the podcast. First of all, how? Second, Huang could not have crafted a better comment to undermine this and many of his other comments.

The only way you don't know your home address is if you've been shielded from the act of entering it on documents and driving yourself home because you always have someone else doing it for you. Huang's lived experience is thoroughly disconnected from the average person, and yet the choices he's making impact most regular people.

'[China] manufacture[s] everything in volume. They manufacture smart kids in volume'

Throughout the long conversation, Huang comes off as an industry Pollyanna and wildly self-serving.

He's asked repeatedly about China's approach to AI and if and how the US should be competing with them and ensuring that the US doesn't fall behind in this critical race. To put his comments in context, you have to remember that Huang personally asked the White House to allow it to continue selling AI chips to China. Huang did note, by the way, that he at least sells new technology to US companies first.

Overall, Huang essentially never criticizes China and, in fact, seems almost in awe of its approach on most fronts, especially in its use of open-model community (calling it "super-vibrant), and how it's raising an army of people to build its AI future.

"They have so many scientists and mathematicians. The number of engineers they have, they manufacture that in volume. They manufacture everything in volume. They manufacture smart kids in volume."

I don't know if that last bit was a backhanded criticism of the US education system, but it's not like he added, "Of course, we are creating just as many smart kids in the US."

'all of the rhetoric and all the alarmism, all the doomerism, all of the predictions — they’re scaring people'

If Huang has any criticism, it's reserved for his US counterparts, whom he calls "alarmists" and "doomers".

"I want to see us not ruin the opportunity for the United States to benefit at the highest level. And notice all of the rhetoric and all the alarmism, all the doomerism, all of the predictions — they're scaring people."

Huang insists that these AI models are still just programs running on operating systems, and wishes people would stop infusing them with human attributes.

'Just because it comes from a scientist doesn’t make it scientific'

While Klein mentions most of Huang's partners and occasional alarm-sounders, like Altman, Modei, and Musk, Huang doesn't actually mention any of them by name. The 'Godfather of AI' and chief alarmist Geoffrey Hinton, though, does receive special mention.

In response to a question about Hinton's assertion that there's a 10% chance AI will end society as we know it, Huang calls the statement "irresponsible" and adds, "Just because it comes from a scientist doesn’t make it scientific."

In a way, Huang is right. After all, Hinton is still just a person who can bring personal opinions to the debate. Huang argues that the "10% chance is not grounded on science."

But isn't it? If Hinton is the Nobel Prize-winning person who introduced the world to deep learning, which helped trigger the generative AI revolution, isn't everything he says, in some way, based on science?

Instead of saying he understands the concern but here's why he's wrong, Huang just claims the foundation of Hinton's argument is faulty, and therefore his statements are not really worth addressing. Huang would simply like everyone, all the doomers, to stop scaring everyone.

'The A.I. supercomputers are super energy efficient, but they’re still going to use a lot of power'

Huang says he wants AI to benefit every company and person, and he tries to offer a reasoned approach to the growing outcry over data centers.

He agrees that if people don't want them in their town, "then so be it," and encourages companies to be transparent about the impact, though he argues that their "use of water is really efficient." In the same breath, Huang tries to have it both ways: "The A.I. supercomputers are super energy efficient, but they’re still going to use a lot of power."

If, in Huang's perfect world, AI companies do generate their own power (building power sources takes time, probably more than it takes to build data centers) and they somehow lower property taxes, maybe data centers could someday be a net positive. Most, I think, would argue we're not there yet, even as the number of data centers being built across the US explodes.

'I think we just have to acknowledge that we got ourselves really gummed up in climate change and sustainable energy'

More concerning is Huang's shocking perspective on climate change and fossil fuels.

"I think we just have to acknowledge that we got ourselves really gummed up in climate change and sustainable energy, and as a result, we just didn’t plan enough energy production."

Klein, naturally, asked what Jensen meant by "gummed up," and, yes, it got worse.

"Well, in the near term, energy production requires fossil fuel. And because there’s just so much angst about fossil fuel energy production, if you look at our country, we’ve produced very little net new energy for a long time."

There's a lot to unpack there. It sounds like Huang is downplaying real climate change concerns, something that might align with the beliefs of his pal and "drill-baby-drill" and climate-change denier US President Donald Trump.

In fact, Huang never addresses whether climate change is real or not and instead seems fixated on how the loss of more fossil-fuel-burning energy plants puts us behind China in the energy race, and, more problematically for him, at a time when AI needs a lot more energy.

'They’ve got to inflict an enormous amount of pain and suffering on you so that they can save you'

It's not all bad. Huang believes, "If you want a future that is sustainable, lean into AI”

And then he says this:

"Yeah. It’s kind of like, in order to save you, they’ve got to hurt you first — that’s the nature of surgery. They’ve got to cut you open to save you. They’ve got to inflict an enormous amount of pain and suffering on you so that they can save you. And so I think A.I.’s kind of like that."

As for what level of "pain and suffering" Huang believes we should endure, he didn't elaborate.

If you were watching or listening to the interview hoping for some encouragement or a more rational, middle-of-the-road approach to regulating and governing AI in ways that truly benefit all, you might've been disappointed. Huang does say that when the builders see AI going awry, they should stop it, and when an AI acts out of alignment, "they shouldn’t release the product. That’s the simple answer." But he never addresses the concerns over what appears to be happening before Anthropic, OpenAI, and potentially others release their models. These systems are jumping fences in testing.

Huang may not understand the needs of the common person, but that pales in comparison to the dispassionate AI, which neither knows nor cares about us. We need leaders like Huang to start caring, but first they have to understand us and our very reasonable concerns.

So far this summer, four frontier AI models have broken out of the isolated environments built to contain them. The media coverage treats this as four separate scandals, when it's really more like four scandals in a trenchcoat.

Three of these four incidents trace back to the same evaluator, Irregular, making the same class of environment mistake. That's one weak point in the industry's safety infrastructure, found repeatedly, by the same tests, in the same way. AI models are not randomly going rogue despite what the hype might want you to believe.

What we're watching is evidence that the entire industry is leaning on a thin, overstretched layer of third-party safety testing that can't keep pace with how capable these systems have already become.

Why now, all at once?

It's worth asking why four incidents have surfaced in such close succession, because the answer says more about the industry than the models do. These disclosures aren't the product of independent audits arriving on their own schedules, but rather they are being released on the labs' timeline, shaped by the labs' incentives.

Once one escape became public, the pressure to get ahead of the story, rather than be caught concealing a similar one, pushed the others into the open in short order. The clustering is a symptom of an industry where disclosure itself is a PR and market moving decision rather than a regulatory one.

That should concern anyone hoping regulation is being shaped by evidence rather than by which lab wants to look transparent first.

The real Rubicon isn't the model — it's the dependency

The instinct is to ask whether an AI model "went rogue" and crossed some invisible line into autonomous misbehavior. That's the question the labs would rather we ask, because the answer is reassuringly narrow: a misconfigured sandbox, patched, incident closed. For that reason, it's the wrong question to ask.

The right question is about the system around the model. The industry has built itself on a small, concentrated pool of specialist evaluators who are themselves struggling to contain what they're testing.

Enterprise problem, not a lab problem

If the organizations built to stress-test these systems before release are stretched this thin, then any enterprise treating a single vendor's safety assurance as sufficient due diligence is inheriting that same fragility, just one layer downstream.

You don't get to outsource your risk assessment to a lab's press release.

A misconfiguration that lets a model reach GitHub in a sandbox is trivial. The same category of blind spot, undetected in a production deployment handling your customer data or your regulatory obligations, is not. Particularly when it will be you and your business left holding the bag.

Control has to be architectural, not promised

Few organizations are going to openly rail against Anthropic, Meta, or any frontier lab for its disclosure of a sandbox escape. But they will be shoring up their own AI sovereignty and vendor-agnosticism rather than around any single model or lab's word.

If your AI strategy depends on one provider's safety claims holding up indefinitely, you've concentrated your risk exactly the way the evaluation industry has concentrated on its own. That means the solution is refusing to build your organization's dependency on a single, unverifiable point of trust.

Trusted AI is becoming the new cybersecurity: governance, transparency and the ability to verify what your systems are actually doing being the differentiator that determines whether you can adopt AI tools at scale with confidence, or whether you're quietly accumulating risk you can't see.

The choice is still yours

So, have we crossed the Rubicon? The labs, and the dependency model the industry has built around a handful of overstretched evaluators, arguably already have. But that crossing doesn't compel anyone else to follow.

To press the metaphor a little further: Caesar crossing the Rubicon alone was significant, but it's the thousands of legionaries who crossed behind him that turned it into the point of no return.

Every organization adopting AI right now is deciding, individually, whether to be one of those legionaries - wading in on someone else's momentum and someone else's risk calculus - or whether to hold the line on its own terms: verified, governed, and in control of its own crossing.

The industry doesn't get to make that decision for you. Don't let it.

We've featured the best AI chatbot for business.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Around the world, governments are changing how e-invoicing works. Instead of checking transactions after the fact, many are moving towards real-time clearance models.

Brazil’s NF-e system was one of the earliest examples of this shift. Italy’s SDI model follows a similar approach, where an invoice must pass through the tax authority’s exchange system and its required checks to be considered issued for tax purposes.

In practice, this means compliance is no longer something that happens after a transaction but increasingly sits within the transaction process itself.

In some markets, goods can be left sitting in warehouses until the required electronic document is authorized. A process that once happened in the background now sits directly in the path of business operations, determining whether products move, payments are processed, and supply chains keep running. If this sounds familiar, it is because we have seen the same shift in air travel.

If you have flown in the last twenty years, you know the drill. Passports, visas, digital check-in, and security checkpoints all need to be cleared before a journey can begin. It used to be far simpler.

Once upon a time, passengers could simply turn up at the airport, buy a ticket, and walk straight up to the gate without a second thought. Today, every stage of the journey depends on passing a series of checks before moving forward. Tax compliance works the same way. Think of clearance as the transaction’s boarding pass. Without it, nothing moves.

The compliance checkpoint has moved earlier in the process, and it now touches logistics just as much as it touches finance. Businesses can no longer afford to treat tax as a back-office function.

When invoice clearance becomes a prerequisite for shipping goods, compliance becomes an operational issue. A disruption can delay shipments, interrupt cash flow, and disrupt day-to-day operations.

The concentration problem

For businesses, real-time clearance brings tangible benefits. An invoice that once took days, or even weeks, to process can clear in seconds - helping organizations improve cash flow and reduce administrative delays. But greater efficiency often comes with a trade-off. In this case, the trade-off is concentration.

When compliance depended on paper processes and manual checks, risk was spread thin across multiple workflows and touchpoints. Today, that risk is increasingly concentrated within a small number of digital systems that handle everything from billing details and vendor data to banking information and payment instructions. This concentration unsurprisingly draws the attention of cybercriminals.

Every integration between billing tools, software providers, and external systems creates entry points for hackers. Businesses with integrated, well-governed systems are generally better positioned to manage this risk. And when invoicing is treated as a standalone system, the connections between it and the rest of the business can become weak points for attackers to exploit.

This is where the language around compliance needs to shift. E-invoicing is not just a platform for exchanging invoices, it is a critical business system which requires security and governance built in from day one. After all, a cleared invoice sitting inside a poorly protected environment has done nothing more than move its potential exposure to somewhere harder to see.

The digital passport problem

A boarding pass gets a passenger through the gate, but it is only one piece of the journey. Behind it sits a network of systems responsible for everything from identity checks and payments to security screening and baggage handling. Tax compliance is beginning to look the same.

Clearance confirms that a transaction has passed a compliance check, but it says little about the wider environment behind it. Businesses still need confidence that their data remains accurate, access controls are maintained, and critical integrations continue to function as intended. That is the digital passport problem.

Passing a check is not the same as proving that the wider system remains trustworthy. As e-invoicing becomes more deeply embedded in business operations, keeping that wider environment secure and reliable becomes just as important as achieving compliance in the first place. And that’s where AI can make a real difference.

Just as modern airports rely on intelligent systems to spot anomalies, identify risks, and keep passengers moving, AI can help businesses monitor increasingly complex compliance environments. However, AI does not fix poor governance or clean up any bad data. Applied to a well-managed environment, AI can help identify anomalies early and improve decision-making.

Applied to a poorly managed one, it simply accelerates existing issues and can make them harder to contain. Good people, good processes, and good data need to come first, and AI can act as an accelerant.

Proving it continuously

Businesses need to continuously demonstrate that the systems supporting compliance are secure, well-governed, and resilient as mandates and threats evolve. For years, fragmentation was the challenge. Different portals, tools, and countries each introduced their own complexity and their own weak points.

Consolidation closes many of those gaps, but it only works if you treat the consolidated system with the same seriousness as the mandate that created it. Get it wrong, and instead of managing multiple scattered risks, businesses can find themselves relying on a single, highly attractive target.

Security has moved earlier and become more continuous. It now depends on systems working together, rather than any single checkpoint. Businesses that treat real-time e-invoicing purely as a compliance exercise are missing half the picture. The other half is making sure the system carrying that compliance is one they trust.

Just as a boarding pass is only useful if the systems behind it are secure and working together as intended, a cleared invoice is only as valuable as the environment supporting it. Businesses need confidence in the systems carrying it, not the transaction alone.

Real-time e-invoicing may have started as a compliance requirement, but it is increasingly becoming an operational one. The businesses that succeed will be those that treat compliance as something that must be continuously demonstrated, monitored, and maintained.

Checkout our list of the best accounting software for small business.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

From businesses exhausting yearly AI budgets in just months to some imposing limits on staff AI use, it’s clear heavy token consumption or ‘tokenmaxxing’, is reaching its limits.

Instead of incentivizing and measuring business output, many are measuring consumption and usage.

While cutting AI usage seems like the natural solution, this doesn’t always work in practice. In fact, this can result in genuinely useful projects being pulled.

Without a reliable way to measure AI ROI, companies cut against the only metric they can see: consumption.

We know a lack of performance benchmarks and traceability is translating to poor ROI with Gartner estimating that 84% of finance leaders have not been able to measure the ROI of AI initiatives.

To give enterprises the confidence to navigate this next phase, business leaders need to prioritize getting a clear picture of AI spend from beginning to end.

How We Got Here

The rapid adoption of generative AI has introduced a new consumption model and traditional IT financial management needs to adapt to keep up. For a long time, organizations optimized their budgets around cloud and on premises workloads.

However, these new tools function in a different way. Costs vary based on the complexity and accuracy of a prompt or even the type of model being used. In other words, the inherent variability of LLMs has made accurate cost tracking more difficult.

The problem is only being made more complex by the introduction of AI agents which can increase expenses because of unpredictable token consumption, heavy GPU usage and fast scaling. Unlike standard AI chatbots, these are not static tools.

These agents work using continuous, background loops independent of human operators which can generate multiple queries to solve difficult tasks. Because this reasoning loop happens autonomously, it can make it trickier to understand how much it’s costing to run an agent.

The Blind Spot

Another hurdle is visibility. Because of the rapid adoption of AI tools, spend is rarely centralized, distributed across a complex mix of business units, infrastructure, vendor APIs and engineering teams. Enterprise cloud and API bills are also unlikely to be updated, or interpreted in real-time meaning it gets even harder to understand what has been spent.

As a result, this is forcing a shift in how organizations measure success and spend. It is not enough to track the raw, isolated figure of cost per token. To make sure that enterprises have a clear understanding of what they are paying for and what they are getting in return, it’s important that team leaders have the frameworks in place to keep a firmer hold on budgets.

A business's financial practices must evolve at the same pace as its technology adoption.

Value Over Volume

So, what replaces the trial-and-error approach that has defined AI adoption so far? Having spent years working with businesses, first through the cloud transition and now enterprise AI, I’ve seen that sustainable returns depend on rethinking how we define and measure productivity in the first place.

Many businesses have been encouraging workers to use AI wherever possible, but few have implemented specific AI metrics that can tie together, higher usage to improved outcomes.

For example, has the process of taking a product from concept to production sped up or become less expensive? Just measuring intermediate steps like code check-ins, ticket closures and cases doesn’t necessarily align with business value.

To do this, a benchmark must be established. Businesses need to know what a process, from manpower to tools already costs them without AI so they can make the right call. Without that baseline, any gain is guesswork.

This is where frameworks like Technology Business Management and FinOps earn their place. Both practices are aimed at making sure that every aspect of spend is understood and tied to a key business objective. In my experience they also help build a culture that instils accountability amongst teams when it comes to their role in managing IT spend.

There is a need to dismantle the silos that can keep costs out of sight and instead focus on treating technology spend as a real-time product variable.

Sustained Financial Intelligence

To make true progress and create value with AI, teams must see it as a measurable business driver. This transition requires leaders not to see success as how many times an employee has turned to AI or logged into the latest tool, or just having an AI tool as a part of a business process. Real ROI comes when enterprises can connect money spent directly to improved outcomes, that affect the business output.

Put simply, it’s about getting a clear and honest picture of spending so smarter choices can be made and investment in the right areas can be prioritized. The result? A more accurate understanding of costs and what AI projects are pushing the business forward as opposed to just being vanity projects.

We've listed the best business laptop deals.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

There's an old saying that necessity is the mother of invention. When there's no easy way out, we're forced to think differently and come up with something new. AI offers a shortcut around that. It gives instant answers with almost no effort needed.

That's great for productivity in the short term, but it comes with a hidden cost. It quietly takes away the pressure that pushes people, and whole organizations, to get better at solving problems.

Is that a problem? Not straight away. Spending less time on a task means lower costs and more capacity for other things. But over time, it risks wearing down the skills a business actually needs to keep coming up with new ideas. We call this the "productivity trap": you get faster in the short run, but slowly lose the ability to think for yourself.

We've spent time studying how organizations learn. We built a model looking at how people decide whether to work through a problem on their own or just reuse a solution someone else already found.

The pattern is clear. When "good enough" answers are free and instantly available, people reuse them more and explore less. Teams start settling on the same handful of approaches. Output goes up, but fresh thinking quietly dries up.

This isn't just a theory. A well-known 2015 study looked at computational biologists and found that when they could easily see each other's half-finished work, they spent more time polishing what already existed and less time trying new approaches. The range of ideas being explored shrank.

We've seen something similar in our own workplaces – with weekly sessions set up to share new AI tools were well attended, but over time they produced fewer genuinely new ideas. A small group of people kept doing the exploring. Everyone else just waited to pick up whatever they found. For business leaders, the lesson is fairly simple. AI changes how people learn at work.

If a decent strategy memo or risk assessment can be produced in minutes, fewer people will bother doing the slower work that actually builds understanding, like talking to customers, checking the numbers themselves, or testing their own assumptions.

Left unchecked, this leaves you with a team that's good at getting answers out of AI but not very good at judging whether those answers are actually right.

Adding some friction back in

One option is to slow down how much AI your team uses. But that gives up real efficiency for a benefit that's hard to measure, which is a tough sell to any leadership team focused on this quarter's numbers. A better approach is what we call "strategic friction": a small, deliberate step you add between an employee and an AI generated answer, so they have to think for themselves before they get it.

This isn't about adding pointless red tape. It's about making sure that before someone benefits from a ready-made AI answer, they've spent enough time on the problem to actually understand it.

Slightly counterintuitively, our research suggests this doesn't slow people down overall. It speeds them up, because someone who has already wrestled with a problem is much better placed to spot when an AI answer is wrong, and to make it better. There are a few practical ways to do this, from simple rule changes to more involved system design.

The easiest option is asking for proof of independent effort first. Before someone is allowed to use AI to draft a market assessment, ask them to write up their own rough version first, even if it's incomplete, showing what they already know and where they got stuck. This needs no new tools or software, just a change in how you work.

A step further is designing your AI tools to ask questions before giving answers. If someone asks for a competitor analysis, the tool could first ask them for a few things they already know or have noticed. Then it builds its answer around that. This gets people thinking before they receive an answer, and it usually makes the AI's output better too, since it's working from fresh information instead of generic guesses.

The most involved option is building tools that only unlock once someone has put in their own input. For example, someone might need to upload their own starting assumptions before they're allowed to generate a risk assessment. It takes more work to set up, but it makes sure every AI output is a genuine team effort between the person and the tool, not just something handed over wholesale.

Not every task needs this. For simple, low stakes jobs, like formatting a slide or summarizing some notes, speed is all that matters, so let AI do its thing. Save the friction for the work where you actually need people to think, not just churn things out.

Spotting the builders

There's a hiring angle here too. In a workplace where everyone has access to AI, the person who gives the "right answer" is no longer that impressive. What matters more is how they got there.

Look for people who use AI selectively and question what it gives them, rather than people who just accept whatever it produces and pass it on. Simply watching how someone uses AI, whether they push back on it or just go along with it, tells you a lot in an interview or a performance review.

AI is going to keep making businesses faster. The harder job for leaders is making sure that speed doesn't quietly wear away the judgement and creativity the business relies on. A bit of well-placed friction, built into your tools, your workflows and how you hire, is what keeps a business coming up with new ideas instead of just churning out answers.

We've featured the best AI website builder.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Oh Netflix, you have a lot of apologizing to do.

Its latest reality game show, Wonka’s Golden Ticket, hasn’t even been on the platform for 24 hours, and it's already found itself in the midst of scathing criticism for, you guessed it, its controversial use of AI.

To put it into context, the show takes inspiration from the beloved 1971 movie musical with Gene Wilder, Willy Wonka and the Chocolate Factory. In the game show, a group of contestants is thrust into a series of themed challenges filmed on elaborate sets based on the original movie’s fantastical aesthetic.

The first seven episodes are available to stream now, and the final two episodes will be released on September 30 where the champions will be revealed.

From its first teasers, I knew the show had already begun its downward spiral before it even saw the light of day. For me, Netflix game shows are nothing more than money-grabbers and devoid of any quality, and to those who held out hope for the show’s survival, you were very naïve indeed.

It was over before it started

When the best streaming service announced the show in the summer, it wasn’t just its cheap and gaudy attempt at resurrecting the movie’s familial charm that had viewers concerned. It was when Netflix unveiled that the show’s host would be an off-screen AI-generated recreation of Wilder’s voice — yes, the OG chocolatier — that the boycott really took shape.

What’s shocking is that, surprisingly, Wilder’s estate actually granted Netflix the right to use an AI recreation of his theatrical Wonka voice. Even so, we thought this decision was going too far. I was particularly upset; the original movie is a childhood staple of mine.

As you can imagine, the AI-generated voiceover is by far the most criticized aspect of the show. In addition to an eye-watering 20% score on Rotten Tomatoes, major outlets such as Variety and The Guardian haven’t taken well to the AI host, with the latter saying ‘it sounds stilted, indefinably wrong somehow’. That’s tame compared to what viewers are saying.

Despite the legal permission Wilder’s estate gave to Netflix, it’s still unfathomable to some that it was given a pass in the first place — including this user on X, who called Netflix "disgusting".

Equally, Reddit is full of posts featuring the same level of revulsion. One user was left asking, ‘Does Netflix do anything decent these days?’, to which another user replied, "[Netflix] has always been quantity over quality". My thoughts exactly.

A sickly streaming experience

I’ve never committed to any of Netflix’s game shows, but in the case of Wonka’s Golden Ticket, I streamed the first episode to see how much it lived up to its car crash reviews. The only thing I’ll say is, don’t waste your time.

It’s clear that Netflix has spent quite a bit of money on the production (the prize money alone equates to roughly $3 million, not to mention its sets and engineering must’ve also cost a fortune), but money doesn’t buy everything — not even a near-accurate AI version of Gene Wilder.

While there’s a slightly replicated cadence, the uncanny valley-ness of it all leaves you feeling rather sickly. It’s emotionless, robotic, and completely lacking the bold infectiousness that Wilder brought to the original role. It’s not just that; the structure and overall approach are just as disjointed.

Unlike the original five children who first stepped into the chaotic world of the chocolate factory, Netflix introduces 12 pairs of contestants, totalling 24 players, of which remembering all their names and relationships is a chore in itself. But the icing on the cake is the roster of false and unnatural reactions that are played to the camera during both the challenges and confessional segments. This isn’t The Traitors, this is a chocolate factory. It’s not that serious.

One of the advantages of having written about tech for as long as I have is that you start to notice when the same ideas come around decade after decade, in different shapes and sizes. Especially the bad ones.

12 years ago Google invented something called Google Glass — a set of glasses complete with a camera so you could record what you were looking at — and it was so roundly hated by everyone that we all breathed a collective sigh of relief when they went away.

But now it's 2026 and smart glasses are back, powered by AI, and tied in with a world-famous fashion brand, but fundamentally they still have exactly the same problem, which still hasn't been fixed: they enable people to record other people without their consent.

The problem is that none of this should have come as a surprise. The privacy risks of putting a camera on somebody's face weren't merely predictable — they were predicted. Women were raising concerns about men using Google Glass to record them more than a decade ago.

And yet here we are again.

The curious case of Google Glass

Let’s remind ourselves what happened the first time that smart glasses arrived.

Google Glass arrived in 2013. It puts a small heads-up display and camera on the wearer's face, allowing them to take photos and record video from their point of view. Almost immediately, privacy became one of the defining controversies around it.

By 2014, women were explicitly raising issues of abuse. Wired reported that women were particularly interested in technology that could detect and block Glass because of concerns about men covertly recording them in nightclubs. "Even if they didn’t know if the device was recording, they felt threatened by its presence", said Oliver, the creator of the detection device.

Google Glass wearers were being called “Glassholes,” businesses were banning the device, and Google itself was publishing etiquette guidance because of complaints about people creepily filming others without permission.

By 2015, consumer Glass effectively died. Privacy wasn't its only problem — its price, awkward appearance, and lack of compelling everyday uses all played a part — but by January 2015 Google had stopped selling the consumer Explorer Edition. Glass survived as an enterprise product before Google finally discontinued the Enterprise Edition in 2023.

Smart glasses didn’t die with Google Glass. They had simply retreated to a lab to lick their wounds and wait for that vital next piece of technology that would make them relevant again. That technology was AI. In 2023, AI-powered smart glasses, partnered with leading fashion brands, started to emerge, with Meta leading the charge with its Ray-Ban Meta glasses.

This time they looked much more like ordinary glasses. Meta had also found the compelling use cases that Google Glass struggled to provide: its glasses could use AI to answer questions about what you were looking at, give you directions, and translate signs and conversations. But they could also still record what, or more importantly, who, you were looking at without needing that person's consent.

It’s like women don’t exist

Frankly, I don’t think Meta should have released them with that ability, but here we are. It’s like we’ve learned nothing from the Google Glass debacle. Almost as soon as the next generation of smart glasses emerged they were being used to harass women and record them without their consent or knowledge.

Researchers at the University of Sydney have been studying hundreds of social-media videos involving smart glasses and found men using them to approach and record women without their knowledge, including at gyms, workplaces and beaches. Australia's eSafety Commissioner says the technology can amplify gendered harms because recording can be almost undetectable, and specifically highlights pickup-artist content in which women often don't know they're being filmed.

The researchers analysed 350 publicly available Instagram videos from 2023–2026. In a subset of covert POV videos, around 60% of interactions were classified as potential harassment. And in about 43% of the videos, women were subjected to further abusive online commentary, sometimes exposing personal information.

These are real victims. CBS interviewed a woman who discovered she'd been secretly recorded only after the resulting video went viral, attracting more than 200,000 views. Her reaction said it all: “I had no say.”

And France has now opened an investigation involving alleged sexual harassment and discreet smart-glasses recording.

I don't think these products were designed to harm women. That's almost the point. Too often, the technology industry seems to design products around what the person using them can do, without giving enough consideration to the person standing on the other side of the technology. And when the potential for abuse disproportionately affects women, that's a particularly serious blind spot. Women have historically been underrepresented in the data and testing used for product design, with designers still describing the male body as the implicit default.

Meta clearly did think about privacy. The glasses have a recording LED specifically designed to tell people nearby when the camera is being used. But if your safeguard against covert recording is a tiny light on the front of the glasses — one that people have found ways to conceal or disable — is that really enough?

Did it consider what happens when a man wearing an almost invisible camera approaches a woman who's alone?

Meta's audio-only smart glasses

Yesterday at Meta Connect, Meta announced a new version of its smart glasses called the Ray-Ban Meta Audio glasses, which don’t have a camera. They got barely a mention in Zuckerberg’s keynote, but I wonder if they were released because of the existential threat to its business that the now commonly used “perv glasses” moniker poses.

Ray-Ban Meta Audio has no camera. You can still listen to music, make calls, translate conversations, and talk to Meta's AI, but the person sitting opposite you no longer needs to wonder whether they're being secretly recorded.

Meta hasn't said the Audio glasses exist because of the privacy backlash, and removing the camera inevitably removes some genuinely useful features too. For visually impaired people in particular, a camera combined with AI can be transformative.

But Ray-Ban Meta Audio proves something important: smart glasses don't have to contain a camera.

Twelve years ago, women were already warning that putting inconspicuous cameras on people's faces could enable men to record them without their knowledge, and today researchers are documenting precisely that behavior.

Google Glass users became known as “Glassholes” because people hated that the person they were talking to was recording them. Twelve years later, we're having the same argument about “perv glasses.”

That's what makes me angry. This wasn't an unforeseeable consequence of some revolutionary new technology. We'd already run this experiment and realized that these should never have got out of the lab in the first place.

For decades, banking technology has been built around a simple sequence: a person makes a financial decision and the bank’s technology processes it. A customer decides to open an account, move money into savings or apply for a loan, and the bank’s systems execute that instruction.

Artificial intelligence can potentially reverse that sequence. Instead of waiting for a human to specify an action, AI can interpret the person’s goals, understand the financial context around that goal, and determine what happens next. It could recognize that a customer is likely to face a cash shortfall, identify the available ways to address it, and potentially execute the appropriate action.

McKinsey estimates that generative AI could create $200 billion to $340 billion in annual value for the banking industry. Yet much of the industry is adding AI to systems designed for the old model, not rebuilding the model around AI.

The problem is that AI inherits the same product silos and process boundaries. Banks gain another layer of technology, but not the cross-system decision-making needed to realize AI’s full potential.

The first wave is still about better processes

The most visible applications of AI in banking are often the easiest ones to deploy. Banks are using AI tools to improve customer service, automate fraud detection, personalize recommendations, summarize documents and accelerate credit decisions. Lloyds Banking Group, for example, says more than 50 AI use cases were rolled out across the group in 2025, generating around £50 million in value, with more than £100 million in additional value expected in 2026.

McKinsey has made a similar observation based on the industry's experience with generative AI. Simply adding AI on top of existing processes will not produce transformational change and can instead create another layer of technical debt.

The difference between an AI-native architecture and a chatbot attached to an existing system can be tested with three questions.

• Is the AI following a fixed sequence of instructions, or can it choose and coordinate actions within a defined system of guardrails, trusted data sources and approved tools?
• Is the process designed for the agent to act, with human review and every decision recorded for analysis?
• And are permissions, monitoring and regulatory controls built into the workflow, particularly for critical functions such as compliance and fraud prevention?

If the answer is no, the company has added an AI interface without redesigning the underlying process.

From products to outcomes

The more significant transformation begins when the bank starts with the customer’s objective, not with the banking product.

Consider a customer who wants to maintain a certain level of liquidity while earning as much as possible on excess cash. An intelligent banking system could continuously monitor their balance, upcoming payments, income, available credit and other relevant information, then determine whether money should remain liquid, be invested elsewhere or be used to reduce borrowing.

While the system can continue making decisions as the customer’s circumstances change, that does not require customers to surrender control immediately. Adoption can begin with low-risk actions, such as moving excess cash into savings or setting aside VAT for future tax payments, before expanding into more consequential decisions.

This is already beginning to appear in financial institutions, although mostly in bounded applications. Deutsche Bank, for example, has deployed an agentic AI system for third-party risk management in which several AI agents retrieve relevant controls, analyze supporting documentation, and propose assessment outcomes. Human assessors remain responsible for reviewing or overriding those recommendations.

Like a new employee, an AI agent should receive defined permissions. Transparent activity logs, alerts, approval thresholds and the ability to override decisions would make that principle visible in the product and enforceable by regulators.

The bank becomes a continuous decision system

Under this premise, the bank becomes an intelligent execution layer that continuously manages financial activity to accomplish a defined objective.

This shifting structure is also visible outside traditional banking. Visa and Mastercard are both building infrastructure for AI-initiated payments, allowing agents to act on behalf of consumers and businesses. Visa Intelligent Commerce is designed to let AI agents find and purchase products on a user’s behalf, with tokenized credentials, authentication and spending controls built into the payment flow.

As agents move from recommending actions to executing them, banks will need to ensure transactions remain within the customer’s intent and risk tolerance. The institution remains responsible for keeping the agent within those limits.

The architecture has to change with the AI

Deloitte found that integration with existing systems and tools is the top modernization challenge for 77 percent of banking executives deploying AI, ahead of security, compliance and cloud interoperability concerns.

Traditional banking systems separate payments, lending, accounts and compliance across different applications. AI agents need to work across those boundaries, combining data and actions from several systems to make a single decision.

Banks therefore need an orchestration layer that allows AI to access those systems without requiring another custom integration for every use case. The core platforms can remain systems of record, while more of the decision-making happens above them.

This gives AI-forward fintechs such as Revolut or Ramp, as well as new entrants designing their infrastructure from scratch, an advantage over institutions that must retrofit deeply embedded systems. If regulation remains broadly unchanged, the first major financial institution built around continuous decision-making could emerge within five years.

It may not be a bank in the strict regulatory sense, but it could perform an increasing share of a bank’s functions, allowing customers to manage their finances.

To thrive, I believe banks need to become institutions organized to make continuous decisions for the customer’s benefit, not simply to follow instructions. And when it comes to AI, they must stop treating it as a fancy tool to add and start treating it as something to build around.

We've featured the best business intelligence platform.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

You will almost certainly have heard many business and technology leaders extolling the virtue of AI. They will enthusiastically tell you that the technology will have a transformative impact on businesses across the globe.

I think they are right. However, with two important caveats: AI must be given the right jobs, and the people using it must trust what it does.

Too often, AI is deployed across a business without much thought, as overconfident bosses assign agents tasks it isn't designed for. The problem this creates is that many employees using AI tools every day are not yet prepared for their new role in managing these agents. Trust only comes from experience and that builds over time.

When an AI system provides you with results that make sense and actually helps with your work, people start to feel more confident in using it. But if it makes a mistake, that confidence can be lost in an instant. If things go wrong, employees start finding ways around AI or stop using it altogether, and when leaders ignore employee concerns about making AI work better, its benefits are undermined.

The best way to ensure that AI delivers the efficiencies it promises is to ensure that it is doing the right job, and this isn't as straightforward as it sounds.

The areas where AI excels

AI works on probabilities, producing answers that are likely to be right. The key word here is ‘likely', which is why it occasionally hallucinates.

Being probabilistic makes AI very effective when dealing with unclear or complicated information. But it's not the best fit for tasks where a wrong answer could have serious consequences.

There are three core areas where AI really can make a difference. Firstly, when processing documents; secondly, decision support; and finally, personalization.

AI is especially effective at tasks involving many documents. It can extract, classify and summarize information that would normally take humans many hours to review. For example, an AI system could compare contracts against a standard set of clauses, or organize a large collection of customer emails by the issues they concern. This way, a person only has to look at the important findings instead of reviewing entire documents.

AI can also play a key role in supporting company executives to make decisions, as it's very good at taking lots of different pieces of information, finding patterns and delivering options as to what to do next.

It is, however, important to remember that AI is not always the right technology to make that decision. For example, imagine you're buying something and have to choose among many different suppliers.

AI can look at all the information, like how well they've done in the past, what they can deliver, how quickly it is likely to arrive and whether the supplier offers discounts for repeat purchases. It can even articulate why one option might be better than another. In most scenarios, though, it is the human who makes the final choice.

For some businesses, personalization is a promising tool, as AI can be used to make products more relevant to the individual using them. This is an area where working with probabilities can become a key driver for AI. The content only has to be good enough to make that connection so the recipient feels they are being addressed in a bespoke way.

The cost of assigning AI the wrong job

One of the key concerns companies should have about managing AI is that while AI-generated answers can sound assured, there is still a possibility that the technology has got something wrong. If that answer is allowed to trigger action without validation, a small error can travel rapidly through a workflow.

Think about what happens when someone makes a small mistake with a rule or detail in a contract. By the time someone catches the error, it may have already been added to records, sent to people outside the company, or even used to create new software.

To remind myself of AI’s limitations, I find it useful to think of the technology as the equivalent of a super-intelligent, hard-working intern. Interns need guidance and oversight, which should be provided by experienced, knowledgeable colleagues. Crucially, their access to sensitive information needs to be limited and granted only as they demonstrate they can make sound decisions.

AI needs to be managed in the same way. It should only be given greater levels of control when it has proved itself, and then its work must still be monitored by humans.

A good example of the importance of managing AI is its role in software development. AI can deliver a lot of code quickly, but its fast pace means the quality isn't always the same as that of human developers.

In fact, managing large amounts of AI-generated code can be a real headache for IT professionals. For example, surging token consumption and a shift to consumption-based pricing is ballooning AI coding costs, forcing developers to be selective on when and where they apply AI models, as reported by Gartner.

Let AI help design the model, not control the machinery

The most important design decision is where AI is allowed to reason. The riskiest place is deep in the implementation layer, where a small mistake can cause big problems that are hard to fix.

I believe a sound approach is to use AI to work with business ideas and rules. This way, it's easier for people to check and understand what the AI is suggesting.

Domain-specific languages and other abstractions used in model-driven development can provide that separation. AI helps create or refine a high-level application model, while a developer then validates the model before a deterministic platform transforms it into executable software.

AI can suggest how something should be done, but it doesn't have to be responsible for executing every step.

Abstraction helps developers, and in agent-driven systems it becomes a control tool. It simplifies complex systems and prevents small errors from becoming big problems, making AI output easier to test and fix.

Human expertise is becoming more valuable, not less

As AI becomes more prevalent, developers' roles will evolve. They will still write code, but a larger part of their role will be overseeing systems, checking outputs and catching mistakes, a change that brings with it the opportunity to learn new skills.

Developers will increasingly work at the level of systems and architecture, understanding the business rationale, recognizing patterns and judging whether AI-generated code fits the wider application and holds together.

One company that has innovated in this area and is already reaping the rewards for their experience is Ford. They recently hired 350 experienced engineers to help train younger colleagues and improve their AI and automation tools.

The shift was driven by the fact that the AI tools weren't producing the results they wanted without the expertise brought by seasoned professionals. Now, these experienced engineers act as internal auditors, reviewing designs and finding potential problems before they become major issues on the factory floor.

When you combine AI's ability to recognize patterns and process information with human knowledge and experience, it becomes a much more powerful tool. This combination is what makes it truly effective, not just relying on one or the other.

Building trust

Enterprises should also resist the pressure to deploy agents everywhere simply because competitors appear to be doing so. When it comes to using AI, leaders need to think carefully about each situation.

They should ask themselves three important questions. First, are they asking AI to only suggest what to do or to actually do it? Next, can someone check and fix what the AI says before an error spreads? Finally, what would happen if the system is certain about something but is actually wrong?

The answers play a big role in deciding how much freedom to give the technology. For example, a tool that summarizes documents might only need a quick check, but a system that suggests business decisions needs someone's logged approval.

When companies start using AI for specific, practical tasks, it shows employees that it can really make a difference in their work. This approach gives the company a chance to get its data, rules and processes in order. If people can see that the system is working well, and they understand what it can and can't do, they start to trust it more. Over time, this trust grows.

The key is to create a work setup where humans and machines do what they're good at. AI is great at processing large volumes of information and surfacing patterns. People are good at understanding the bigger picture, pointing out mistakes, and taking responsibility for their actions.

The wrong job to give to AI is any task where a mistake carries real consequences, no one checks the work before it causes harm, or the system sounds certain while being wrong. Companies that respect this division will build the confidence to give agentic AI greater autonomy over time. Those who ignore it may discover that a single poorly chosen task can undermine their entire AI strategy.

We've featured the best AI website builder.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Artificial intelligence has become a significant area of investment for UK businesses. More than £6 billion of new AI-related investment was announced during London Tech Week in June, while the UK remains home to the largest AI sector in Europe and the third largest globally.

From customer service and knowledge management to software development and internal operations, organizations are looking for places where AI tools can improve productivity, decision-making and business performance.

But as investment accelerates, another gap is becoming harder to ignore: organizations are often scaling AI faster than their ability to measure, govern and explain its value. That matters when CIOs and CFOs are increasingly being asked not simply whether AI is being adopted, but what the organization is getting in return.

As businesses move from individual copilots towards agents embedded across workflows, the economics become more complicated. A user making a single prompt is relatively easy to understand. An agent may make multiple model calls, retrieve information, invoke tools and take actions to complete one task.

Depending on the platform and pricing model, that can introduce additional consumption, infrastructure and oversight costs. CIOs therefore need to understand not only where AI has been deployed, but what it is doing, what it costs and whether the outcome justifies that cost.

The not-so-hidden cost of AI

AI investment is increasingly being scrutinized in the same way as any other major technology investment. The difficulty is that measuring its return can be unusually complex.

Usage may be distributed across departments, applications, models and workflows, while the benefits can range from time saved to improved quality, reduced risk or increased revenue. Without agreeing what success means first, organizations can end up measuring activity rather than value.

That becomes difficult when organizations expand AI without first defining where it sits in the workflow, who owns the outcome, what success looks like and how costs will be measured. Spending can become fragmented across licenses, models, infrastructure, platforms and consumption-based services, with no single view of whether those investments are delivering value.

The scale of the challenge is becoming visible. Research commissioned by Emergn estimates that large UK businesses lose £67 billion annually across transformation and AI initiatives that fail to deliver. Separately, a Censuswide survey of 500 senior UK decision-makers found that just 31% of businesses already using AI reported a positive return on their investment.

Governance is part of that measurement challenge. A 2026 ShareGate survey of 851 IT leaders across seven countries found that cost visibility was the most commonly cited barrier to measuring AI ROI, identified by 51% of respondents, followed closely by governance complexity at 47%. The challenge isn't simply knowing what AI costs. It's connecting that cost to the use case it supports, the information AI interacts with and the outcome it creates.

Much of the AI debate to date has focused on model selection, skills and productivity. But as adoption spreads, another gap is becoming visible: confidence in governance does not always match what happens in practice.

The same study found that 93% of IT leaders believed their Microsoft 365 governance was ready to support AI responsibly, yet 29% reported that AI tools had surfaced sensitive internal data that should not have been accessible. Another 8% weren't sure whether it had happened at all.

Those gaps carry costs as well as risk. Duplicated tools, additional validation, rework, security investigations and time spent establishing whether an output can be trusted all create a hidden tax on AI adoption. For CIOs trying to demonstrate value, reducing that friction starts with making AI usage more visible, accountable and measurable.

How organizations can regain control

Good governance doesn’t begin and end at procurement. Knowing how many licenses have been purchased and where they have been assigned is useful, but regaining control requires a broader view: visibility into how AI is being used and what it costs, clear ownership of the outcomes, and a well-governed information environment for AI to work from.

Clear ownership matters just as much. As AI becomes embedded in business processes, responsibility can easily become fragmented across IT, security, business teams and individual employees. That makes some basic questions surprisingly difficult to answer: Who owns the outcome? Who monitors the cost? Who decides whether a use case should scale, change or stop?

Cost governance is only part of the picture. Organizations also need to improve the information environment in which AI operates. That means reducing redundant and outdated content, managing access appropriately and helping employees and AI systems find authoritative sources.

Cleaner, better-governed information does not guarantee a correct AI response, but it can reduce ambiguity and make reliable grounding easier. That can mean less time spent searching, validating and reworking outputs.

Employees have a role here as well. Clearly distinguishing drafts from approved material, keeping trackers and priorities current, recording decisions and maintaining authoritative sources all make organizational context easier for people and AI to interpret. Where organizations use AI meeting assistants or similar tools to capture context, those tools should be subject to the same privacy, retention and access controls as the information they create.

The question for leaders, then, is no longer simply whether AI is worth the investment. It is whether they have enough visibility and control to understand where AI is creating value, where it is creating cost and what they should do differently as a result. AI investment is likely to continue, but under different expectations.

Deployment alone is not evidence of value, and governance is becoming more than a risk-management exercise; it is increasingly part of the business case. Leaders need to understand what AI costs, who is accountable for its outcomes, whether the information supporting it can be trusted and what measurable benefit it creates.

The organizations best positioned to scale will not necessarily be those deploying the most AI, but those that can explain what it is doing, understand what it costs and make informed decisions about where it belongs.

We've featured the best AI chatbot for business.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Artificial intelligence (AI) is like a genius with no memory of you. It has plenty of brain power, but unless context is handed to it directly, it knows nothing about your business, your customers, or the decisions that shaped how you operate today.

And solving that memory problem is quickly becoming a major barrier to enterprise AI deployment today.

Much of the current conversation about AI's limitations centers on the models themselves and their evolution toward bigger context windows, better reasoning, or longer memory. But that framing misses something important.

Enterprises don't need to wait for the next model breakthrough to solve this problem. They already have what they need sitting inside their own organizations: decades of content, decisions, and institutional context that most AI systems never see.

Every enterprise has a vast, largely untapped record of how it actually operates in the contracts negotiated, claims resolved, cases handled, decisions made and revisited. This is the accumulated context that makes an organization's judgment distinct from a general-purpose LLM. Yet in most companies, this information sits fragmented across systems, buried in unstructured formats, or locked away with no clear path for AI to reach it.

That is the gap that needs closing before AI automation can scale responsibly. Not a smarter model, but a more complete memory built from an organization's own history rather than a general-purpose training set.

Think of this as "business memory": the nervous system connecting what a company already knows to what it wants AI to help it do next, carrying context to wherever a decision needs to be made.

Giving AI the full picture

Business memory is best described as the accumulated knowledge of how an organization operates. It turns years of enterprise content, workflows and industry knowledge into information that AI can act on for true agentic automation at scale. The thinking behind this is simple.

By creating a context layer that governs what data the AI has access to, and ensuring it operates on relevant, authorized and current information, organizations will get those governed, trusted outputs they need to act with confidence.

An organization's memory is more than the information it retains. It also includes which version is authoritative, who may access it, where it came from and how long it remains valid. Carrying those signals into the context layer helps AI operate within the same rules as the business itself.

Better context alone does not ensure trustworthy outputs; provenance, evaluation, monitoring and human oversight also matter.

And although it accounts for an estimated 80% of enterprise content, based on various analyst reports, businesses use only about 10% of this incredibly valuable resource.

If organizations want to maximize the effectiveness of their AI models, the first thing they need to do is to start leveraging that valuable business memory contained in their unstructured data. And that requires investing in your infrastructure to transform it into an AI-ready format.

In practice, start with a clearly defined use case and identify the authoritative content and data needed to support it. Preserve existing access controls, enrich the information with metadata and relationships, and test whether the AI's outputs are accurate, traceable, and useful before expanding automation. This creates a repeatable foundation that scales across the business without a wholesale replacement of existing systems.

Context is key

However, giving your AI access to the relevant data is not enough on its own. In fact, I’ve seen a lot of AI initiatives fail in the early stages because the business doesn’t appreciate just how complex a lot of this underlying data is.

As such, the content needs to be contextualized by an ‘ontology’, a formalized framework that defines all the different entities, terminology, relationships and rules that exist within your business. This enables your AI to understand not just what information exists across your systems, but how all of that information relates to your business or industry.

Put simply, ontologies are like maps: your AI may well get to your desired destination without one, but having it ensures you get there quicker and without potentially stumbling into any pitfalls along the way.

This is particularly important for regulated industries. In healthcare, for example, ontologies connect diagnoses to treatment plans, physician notes or lab results. Or in financial services, they link industry-specific regulations to the organization's compliance structures and policies.

In other words, unlocking unstructured data provides crucial business context that helps create better, more informed outputs from your AI. But that, in turn, needs to be contextualized by an ontology so that AI can understand data properly.

Taking control of AI memory

AI doesn’t need perfect memory to understand how your business operates and start creating trustworthy, valuable outputs. What it does need is relevant, governed context; most of which lives in unstructured content but only becomes useful once connected to structured enterprise data.

Infrastructure that can enable your systems to access the right information, understand how that data fits together and then apply it within the realities of the business is, therefore, essential. But crucially, that doesn’t mean you have to start from scratch.

In fact, reinventing your foundations would essentially defeat the point. Real business memory understands the systems, content, data, and processes that already exist within your organization, so throwing everything out in the pursuit of better context is unnecessary and, frankly, a huge waste of resources.

You don’t have to wait for the next model update to cure AI’s amnesia, and you don’t need to completely change your current stack to see real value from your automation projects.

By building in the foundations for business memory, organizations can take back control, make the most of their AI systems and make their infrastructure more resilient and adaptable for whatever models the future holds.

We've featured the best AI website builder.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Meta Connect just ended, and the biggest announcement of the night for me was easily the debut of Meta’s VR Glasses — even if they were leaked ahead of schedule by Meta’s own prescription tutorial.

Ahead of their official reveal, I got to spend 30 minutes with the new tech, experiencing a wide range of demos that might have convinced me this isn’t merely Meta’s best VR headset yet — it could be the best VR headset I’ve used yet.

The only downside is we’ll be waiting a little longer to wear it. The headset isn’t due to launch until “Spring 2027” (March, April or May), but at least we know it’ll cost just $1,299 when it does land (price elsewhere TBC).

Slimmer than ever

Let’s start with the design.

Slimmer and lighter than anything that has come before (the actual headset part you wear is down to just around 100g); this headset looks like a pair of glasses. They even swap the headstrap for glasses stems to stay on your head.

The bulk of the weight is instead housed in a puck you carry in your pocket — a lot like the puck used by the Apple Vision Pro or Samsung Galaxy XR. The big difference here is Meta’s isn’t just a battery pack, it also houses all of the headset’s compute. It’s tethered to the headset via an optical cable, but this wire never got in the way while I used the headset — whether I was up and moving, playing a game like Beat Saber Flux, or sitting being productive with multitasking across multiple windows.

The headset can support controllers; however, it doesn’t come with any in the box. Instead, you’ll need to rely on hand controls: either reaching out and grabbing, or the look and pinch controls popularized by Apple’s Vision Pro and Samsung’s Galaxy XR.

Lastly, I should note that, despite being called glasses, this is a headset. They have an opaque front cover that only appears to be see-through to the wearer because of passthrough. This live video feed of the real world isn’t yet perfect (it’s still a little grainy and there’s a slight delay), though it seems at least as good as the competition based on my experience so far.

They do support pescription inserts too. They connect magnetically to the display lenses and don't disrupt the slimness of the headset at all based on my experience.

The device is just so supremely comfortable to wear. My demo was only around 30 minutes, but at one-fifth the weight of a Meta Quest 3 or Samsung Galaxy XR, the difference is immediately noticeable — this is a headset I don’t see myself needing to take off, even after multiple days of working with it on.

Plus, it’s so portable you can easily take this everywhere in whatever bag you have on hand. No dedicated carry case required. I can’t wait to travel with this thing.

An all-round powerhouse

While this can be used for all existing Meta Quest 3-compatible apps (though you will need controllers for those that don’t support hand tracking, either as a new paid add-on or by commandeering your Quest’s handsets), where this headset truly shines is static experiences: productivity and entertainment.

We know VR headsets are excellent for multitasking. You can arrange virtual project app windows around you, and even bring your PC screen into XR to create the perfect setup. Meta’s headset boasts two useful additional features.

First, you aren’t limited to the number of displays your physical setup has when you connect it to your headset. Each laptop or PC can be extended to up to three virtual monitors, a tool that some previous iterations of virtual desktop apps haven’t been able to offer (You could open multiple VR app windows, but only as many connected device windows as you had physical monitors).

Secondly, Meta’s headset can turn any surface into a keyboard. It just scans the surface in front of you — be it a cafe table, a seat tray on a plane, or a bench at the park — and then virtually projects a keyboard and trackpad you can type and swipe on. It works seamlessly. In my demo, I was able to type as swiftly and accurately as I could on a physical accessory, and in terms of portability, this is yet another win for Meta’s VR Glasses.

Moving on to entertainment, it seems Disney Plus’ Apple exclusivity has ended as the app and its library of 3D films are hitting Meta’s new headset. Because it’s an app, you can download content to watch on the go, which includes brand new immersive watching experiences.

While it was still just a demo, I got to experience watching a clip of Avengers: Endgame that extended beyond the screen in a souped-up version of Philips Ambilight.

It’s the scene where Thor is forging his new axe, I was sitting in that same forge. The room lit up and reacted as events unfolded in the film, ending with lightning flying out of the screen as Stormbreaker helped Thor return to full strength. It was a bit gimmicky, but also fun — not something I’d want for every movie, but an experience I’d want when returning to a big tentpole.

All of this content, be it immersive movies, a VR game, or a work document, looks beautiful on the 5K micro-OLED screen. It boasts a 37 pixel-per-degree resolution, which is around the PPD that its rivals, the Apple Vision Pro and Samsung Galaxy XR, boast (Samsung and Apple haven’t revealed these figures, but they have been estimated during product teardowns).

This is significantly better than the Quest 3’s mere 25PPD, and it meets the standard of being able to deliver the IMAX Enhanced Format — with superb audio and visuals. I got a taste of this in my demo, and I’m keen to experience more.

I need this yesterday

I’m about to jet set to LA for a press trip, and I so desperately wish I could be taking the Meta VR Glasses with me for the flight — both for getting work done and for immersing myself in a movie on a big screen rather than the back of the chair in front of me.

Alas, the VR specs aren’t due to land until 2027, though we know how much they’ll cost: $1,299. Much less than a Vision Pro, or a Galaxy XR, and cheaper than the highest price Xreal has said its Aura glasses — the closest thing I’ve tried to the Meta VR Glasses — could be.

At the time of writing, Meta hasn’t revealed too much in the way of specs (I’m writing this up based on a pre-Connect briefing). The Meta VR Glasses can run any Quest 3 app and can multitask across six windows at once, so it must be at least somewhat capable. I expect it boasts at least 12GB of RAM and the latest Snapdragon XR chipset.

I’m just impressed Meta can commit to such a relatively low price, especially months in advance. There are so many unknowns over trade and RAM demand that we’ve seen companies not reveal prices until as close to the last second as possible. Meta has bucked a trend here, but I expect it would only do so if 100% confident the cost won’t budge — if it doesn’t, I’ll be very, very happy.

The abolition of the Department for Science, Innovation and Technology has created uncertainty about what happens next to Britain’s AI ambitions.

Kanishka Narayan’s promotion to Minister for AI gives the technology a dedicated voice at Cabinet, yet folding DSIT’s responsibilities into other departments raises questions about how the Government will coordinate the infrastructure, investment and public-sector adoption needed to support those ambitions.

Recent parliamentary scrutiny of critical public-sector AI contracts has highlighted growing pressure on institutions to strengthen control over AI operations and deployment continuity. Britain has focused on adoption speed; the next step is operational control in ongoing public-service AI use.

The case for sovereign AI is becoming stronger as a result. Britain needs more than access to capable models; it needs controlled operations for the services that depend on them, even when a provider, contract or political relationship changes.

Sovereignty goes further than where data sits

Sovereign AI is often reduced to data residency, with the assumption that keeping information inside Britain is enough to keep it under British control. Data location remains important, particularly in healthcare, defense and public administration, but it describes only one part of the relationship between an organization and the technology it uses.

A hospital could store every patient record in a British data center while relying on models and software that only the original provider can operate or modify. Where that provider is based remains relevant, but the more useful test of sovereignty is whether the hospital can understand the system, govern how it is used and continue running it if the commercial relationship changes.

An organization should know how its systems operate, determine which data a model can access and retain the ability to move a workload or introduce another model without rebuilding the entire service. Once that system becomes essential, it also needs enough internal expertise to keep it running rather than discovering that continuity depends on the decisions of one supplier.

Britain does not need to reproduce every layer of the stack, but it does need predictable control where public services run. Modern AI draws on international research, complex supply chains and open-source projects, and recreating them domestically would consume enormous resources without necessarily producing better technology.

As DSIT’s responsibilities move across Whitehall, the Government needs to decide which dependencies it can accept without threatening security, public trust or an essential service.

Capability and control can reinforce one another

Government AI plans sometimes present sovereignty as the slower alternative to using the largest commercial models, as though Britain must choose between world-class AI and control over its deployment. That assumes every organization needs the same frontier system, when the question is which model suits the work.

For many public workloads, a fit-for-purpose model in a controlled environment is more valuable than one large generic option, especially when governance obligations are strict.

A specialized model can be adapted to the organization's data, tested under the conditions in which it will operate and governed according to the consequences of getting a decision wrong. In those circumstances, greater control can make the system more useful as well as more independent.

Open and portable technology can support this approach by allowing organizations to run models across different infrastructure, examine how systems behave and replace components as better options emerge.

Openness does not guarantee sovereignty, since organizations still need secure infrastructure, skilled teams and clear governance, but it reduces the chance that an entire service becomes inseparable from one provider’s platform.

I have seen the AI market move quickly enough to make that flexibility increasingly valuable. The best model for a task today may be overtaken within months, while pricing, regulation and access can change almost as quickly.

An organization that builds around one proprietary service may struggle to benefit from that progress, whereas one that controls the environment in which its models operate can introduce better technology without starting again.

Government can turn sovereignty into capability

Britain already has many of the ingredients it needs, including strong universities, experienced researchers, growing AI companies and public investment in compute. The next step is to connect those assets with sustained demand in the parts of the public sector where sovereignty has the greatest practical value.

Government does not need to select one national champion or exclude international companies. It can create a market in which providers compete while public bodies retain control of the systems they buy.

Procurement rules for critical AI could require workloads to be portable, independently testable and capable of running within infrastructure controlled by the customer, making the ability to change course part of the original decision rather than a problem discovered at the end of a contract.

This would give hospitals, councils and government departments credible sovereign options while providing British AI companies with a route from pilots to long-term deployment. It would encourage partnerships in which knowledge and operational authority remain with the institution responsible for the service, even when private companies provide the technology and expertise.

The restructuring of DSIT should be treated as an opportunity to clarify Britain’s objective rather than reduce its ambition. A Minister for AI can bring different parts of government together, but the strategy should be judged by whether public institutions can use the best technology available while retaining meaningful control of the systems that matter most.

That is what will allow Britain’s AI ambitions to endure as departments, suppliers and technologies change.

We've featured the best AI tool.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Financial crime has never stood still. In my nearly three decades in financial services, I have watched fraud move in lockstep with the technology built to stop it, and at times, even stay a step ahead of it​,​ given the speed fraudsters adopt new technology.

I have witnessed decades of AI innovation ​that raised​ the bar on fraud detection, yet criminals continue to test those boundaries. Fraud remains a top consumer concern, with over ​87.5 million American adults experiencing a scam or financial fraud each year. That’s roughly one in every three American adults.

The question for financial institutions isn't whether AI has a place in fraud prevention​,​ but whether the AI they've deployed is built for the threat landscape they're facing today – and the one that's coming.

Compute has finally caught up with mathematical vision

For decades, data scientists working on fraud prevention had theories they couldn't implement. The ideas were sound, but the computers at the time weren't powerful enough to get the math done. That constraint no longer exists.

Historically, most fraud detection has been carried out by building a profile summarizing a customer's typical behavior using sophisticated features, a neural network, and the customer’s current transaction to flag transactions that are suspicious. It's an approach constrained by the computational limitations of time.

Today, access to GPU and other high-performance compute is changing the art of the possible. Rather than analyzing a transaction in the context of a profile, GPUs allow us to implement entirely new algorithms that can evaluate a customer's extensive transaction history in real-time as the transaction happens.

The result is a significantly sharper, more accurate prediction and far fewer false alarms that can delay or stop legitimate purchases​,​ eroding customer trust.

Built for one job, not every job

Thanks to greater GPU availability and computational power, we're now seeing new opportunities to support fraud prevention with a sequence-modeling transformer – not a generic transformer, but one purpose-built for transaction analytics and financial crime.

With a purpose-built transformer architecture that’s trained exclusively on financial transaction data and engineered for a single focused task, financial services institutions can detect financial crime in real time using deep personalization and the context of the customer transaction history.

These are not overarching ‘do everything’ models. They are focused foundation models purpose-built to deliver auditable, high-performing, low-latency generative AI for the fight against financial crime. Each of these models specializes in distinct areas, such as account takeover, scams, mule detection, and first-party misuse.

Independent models focused on their specialty allow for a more complete, accurate, and transparent picture than any single model working alone. And this is just the start​,​ as the same methodology applies to risk decisions, hardship, collections, and any application where understanding our customers leads to better engagement, protection, and service.

The future is now

The fraud prevention capabilities that will define the next five years are being built right now. Enterprises investing in protecting customers from fraud – both now and in the future – understand that purpose-built models, use of specialized compute, and AI agents are the path to protecting their customers.

The math protecting consumers today was invented decades ago by AI scientists who saw the potential in algorithms even before the compute and infrastructure existed. I've spent much of my career continually chasing that goal to ensure that the industry is ready to bring the best algorithms when compute shows up.

Some of my proudest work came from refusing to settle and trust that technology will catch up with AI invention and math. Every patent, every model, every AI experiment has served the same purpose: making sure the industry is ready to lead with the best AI tools once compute catches up with scientific invention.

That work is happening right now. The only question is whether ​​​​​​financial institutions are​​ part of it or racing to catch up.

We've featured the best antivirus software.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Over the past year, security teams on our platform cut the time it takes to fix a critical vulnerability by roughly 50%. In the same period, their backlog of unresolved critical vulnerabilities grew nearly 29-fold. These numbers describe the problem every security leader is about to inherit.

AI can now test software at a scale manual testing never reached. Models read code and search thousands of assets for familiar vulnerability patterns, surfacing exposures earlier and faster than any human team could.

For businesses trying to keep pace with an expanding attack surface, that reach is genuinely valuable. It is also exposing a weakness that has drawn far less attention: most organizations cannot validate, prioritize and remediate findings at anything close to the rate AI can produce them.

That imbalance is the whole game now. Investment in AI discovery tools, on its own, does not make an organization more secure. It makes it busier. Left unaddressed, it leaves security teams with larger backlogs and less attention paid to the flaws that actually put the business at risk.

This is the problem Continuous Threat Exposure Management (CTEM) exists to solve. CTEM gives organizations a continuous process for understanding their attack surface, finding weaknesses, proving which are genuinely exploitable, and directing remediation towards the exposures that carry the most business risk. Discovery is one input. The advantage comes from everything after it.

The gap between discovery and remediation is widening

Security leaders have long treated discovery as a capacity problem: test more assets, cover more code, catch weaknesses earlier. AI answers that question decisively. But finding a potential vulnerability is the start of the work, not the end of it.

Every finding still has to be confirmed as exploitable and judged for severity in the specific context where the technology runs. Then it has to reach the right engineering team, win out against existing priorities, get fixed, and be retested to prove the fix holds. AI compresses the first step and barely touches the rest.

That is why the two numbers I opened with can both be true. A rising count of findings can mean better coverage. Faster repairs can sit alongside a growing backlog when discovery accelerates faster than remediation and engineering capacity moves the other way. Neither figure means much alone. The only view that matters runs the entire route, from first detection to verified fix.

Validation is the choke point

AI has also driven down the cost of producing a convincing security report. Some of those reports point to real weaknesses. Others duplicate known findings, misread the target, or describe theoretical issues that carry little real risk. Everyone still has to be investigated. A report that takes seconds to generate can consume hours of an experienced analyst's time before it can be dismissed with confidence.

At enterprise scale, that is how urgent findings get buried. A well-evidenced vulnerability with a credible attack path lands in the same queue as hundreds of submissions that sound plausible and lead nowhere. Security teams have to separate the AI gold from the AI slop, decide which real findings matter most, and do it with engineering capacity that has not grown to match.

Program owners need clear standards for evidence to make that possible. Researchers should be expected to show the likely business impact and demonstrate how a vulnerability reproduces, with automated tooling used to raise the quality of that evidence rather than the volume of submissions.

A consistent track record of valid findings tells a program owner whose work deserves attention first. That record is worth more, not less, as submissions rise.

Business context still decides what matters

Technical severity is only part of what you need to know to decide what to fix first. AI can match a finding to known patterns and reason over the systems it can reach. What it rarely has is the full picture the business holds: which services generate revenue, where regulated data lives, which dependencies make downtime especially expensive, and what compensating controls already exist.

The harder problem is combination. Individual findings that look moderate on their own can form a serious attack path once someone understands how the systems interact, which permissions can be abused, and where controls break down across organizational boundaries.

That is human work. And it is better human work when it is diverse: one researcher goes deep on identity controls, another on API behavior, another on how minor weaknesses chain together across systems. That variety surfaces novel attack paths that automation, and even advanced cyber-models, miss.

Our own data shows what that work is worth. Researchers earned more than $47 million on the H1 Platform in the first half of this year, up more than 25% year on year. The strongest earners are the ones who can explain business impact and show exactly how a weakness can be used.

That is also where researchers add the most to a CTEM program: testing whether an exposure holds up under real conditions, and spotting the connections between weaknesses that look harmless in isolation.

That figure deserves an honest footnote. Aggregate earnings rising does not mean every researcher is winning. As AI absorbs the routine, high-frequency findings, the researchers who once made a living there feel the change first, while those who can chain weaknesses, reason about business logic and produce credible proof find their work worth more.

The job of any serious bug bounty platform is to make that a transition its community can move through, not a wall most of them hit.

That obligation runs both ways. If programs expect researchers to raise the quality of their evidence, researchers should expect fast and fair triage in return, real recourse when a valid report is wrongly dismissed, and a door that stays open to newcomers who have not yet built a reputation.

The best researchers increasingly use AI tools themselves, and the value of a report has never turned on whether a tool helped produce it. Protect the economics and the fairness that reward credible work, and the independent research community grows stronger as AI scales, not thinner.

Build for the volume AI creates

So, is bug bounty dead? Far from it. Assume discovery will only accelerate, and put the effort into everything that happens after a flaw is found.

More discovery puts more pressure on the point where reports are validated and handed to engineering. Without enough triage capacity, well-evidenced findings stall behind automated noise.

Without clear ownership, confirmed risks sit in the gap between security and development. Independent verification matters at the other end too, especially when an AI system proposes the fix and may carry the same blind spot into its judgement of whether that fix works.

Boards and executive teams need measures built on risk reduction, not activity. Finding counts are easy to report and can climb even as an organization gets safer. Confirmed exploitability, remediation speed, recurrence, and the size of the unresolved critical backlog tell the real story.

CTEM holds those activities together as a continuous process, keeping discovery, validation, prioritization and remediation connected as the attack surface changes. It gives security leaders an honest view of where they are gaining ground and where exposure is still building.

The next phase of AI security will not be won on discovery. Discovery is already abundant. It will be won on response: knowing which findings represent real exposure, and moving the most dangerous of them through to a verified fix. AI supplies the reach. Independent researchers supply the judgement and the adversarial creativity AI still lacks.

The organizations that build to convert both into action will pull ahead. The ones that do not will own a faster-growing list of vulnerabilities they never fixed.

We've featured the best endpoint protection software.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Executives are spending a great deal of time asking whether AI agents are ready for production, but this is the wrong place to start, because agents aren’t acting in a vacuum. The quality of their decisions depends heavily on the quality of the data they use.

Model security and performance matter, of course, but they cannot compensate for weak control over the information the agent can retrieve. Before putting an agent into production, an organization needs governed inputs, appropriate access and a reliable record of what the agent produced.

Give an agent outdated customer information, duplicate records or material that was never properly classified, and it will work with what it has, making poorly informed decisions quickly across thousands of transactions.

Unfortunately, the gap between adoption and governance is widening fast. In the rush to adopt AI, many organizations are pushing agents into production before they have solved the underlying data problem, so it shouldn’t come as a surprise that the vast majority of AI projects show zero ROI.

What’s more, AI agent governance doesn’t end with the data an agent may access. It also encompasses the data an agent produces, such as its decisions, the instructions it was given and documents it created. As AI contributes to significant decisions within the enterprise, AI traceability and defensibility have become critical capabilities.

Organizations that build a strong data foundation put themselves in a stronger position to move from pilot to production with confidence that agents won’t transform poorly managed data into a company-wide crisis.

The data problem comes first

The data governance issue is not, of course, a new problem. Most large organizations have spent years trying to wrangle information spread across operational platforms, file shares, archives and applications that should have been retired long ago. But because agentic AI can rapidly act on those weaknesses at scale, the issue has become more pressing now.

Before an agent is given access to data, IT needs to identify sensitive and regulated information, apply classification and retention policies consistently, and determine which sources are current and reliable enough for the job. Legacy data deserves particular care.

Retired applications often contain valuable business history, but they also hold duplicate records, obsolete information and data that’s subject to legal holds or retention requirements.

Organizations should also preserve the context and relationships that give legacy data meaning, and then make selected, policy-controlled datasets available. AI does not act on data in a vacuum, and without the surrounding context, agents will either fail to complete their tasks or, worse, will make and act on faulty decisions.

Access is a business decision

Another mistake that enterprises make is to provide their agents with access to more data than they need. In fact, agents should also only have access to the data they require to complete the tasks they are assigned. Just because a system can be connected to an agent, that does not mean it should be.

Treat agents like employees, and follow the principle of least privilege. For example, while a customer service agent may need current account and transaction information, it most likely doesn’t need legal files or historic employee records. Access should narrowly follow the agent's defined purpose.

Provenance matters as well. An agent should not treat a current system of record and a decades-old archive as equally authoritative. Agents must know where information came from, when it was updated and which policies apply. Again, context matters. Without the proper context, agents will make costly mistakes.

Keep a record of what the agent did

Governance does not end once the agent receives approved data. AI-related data is rapidly becoming material to litigation, compliance reviews and customer complaints. To be prepared, an organization should be able to show what the agent was asked to do, which information it accessed, which policies were applied and what happened next.

Prompts, retrieved content, outputs and resulting decisions should be retained as business records. A reviewer should be able to reconstruct the AI’s decision and trace it back to both the source data and the person or function that authorized the activity.

Finally, accountability cannot belong to the technology team alone. Security, data, privacy, legal and compliance leaders all have a role, but a named business owner must remain responsible for the outcome. Automation can perform an action, but it cannot accept accountability for it.

Data readiness is not a secondary workstream to be addressed after an AI pilot succeeds. It is part of the decision to move that pilot into production. If an organization cannot trust the information, control access to it and clearly explain the resulting decisions, the agent is not ready for greater autonomy.

We've featured the best AI tool.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

If there's one thing President Donald Trump is good at, it's misdirection: focusing your attention on what doesn't matter so he can actively ignore the most distressing problems. So it comes as little surprise that he is now campaigning to rename Artificial Intelligence to "Super Intelligence". In the meantime, he's downplaying concerns and promising to give AI... er... SI "free rein".

Trump rolled out the name change during his annual United Nations General Assembly address in Manhattan on Tuesday, admitting that he doesn't know if he has the power to rename an entire science, technology discipline, and industry. But naturally, the guy who bulldozes buildings first and never asks questions later is willing to try.

"Welcome to the world of super intelligence, dash, SI, SI, " Trump declared. "Let's see if that sticks. Let's see if that goes. It sounds much better, and it's much more accurate." He even plans to change the usage across all government documents, which should only foster mass confusion.

The rationale here is that "there's nothing artificial about it," whatever that means. Trump loves his superlatives, like "greatest ever," "beautiful," "tremendous," "huge" (or "Yuge"). "Super" fits the mold. But it's also inaccurate. AI is not super intelligence. It's not yet human intelligence with extra juice. Consider it this way: Superman, though an alien, is called "Super " because he appears as a man but with extraordinary powers.

Artificial intelligence is so named because it's made like an artificial sweetener. There's nothing natural about it. It's not a genius or super-smart person. It's computers, programs, algorithms, models, and training that can create the appearance of human-like intelligence, which, in some instances, can be far greater than our own. But there's nothing remotely human about it.

The other confusing part of this is that the timeline of AI does allow for "Super Intelligence," but not as the same thing. Instead, it's something down the line that far exceeds human cognition. It's also what some call Artificial General Intelligence (AGI). Whatever you call it, experts think that it's fast approaching and we should be deeply concerned.

The so-called Godfather of AI, Geoffrey Hinton, just issued a fresh warning that the U.S. Congress has “maybe a year, but not much more than a year,” to regulate AI before we lose control. He's particularly concerned about AI that doesn't need human intervention for improvement: "AI has now reached the point where AI is designing better AI,” said Hinton last week after his congressional testimony. "That’s called recursive self-improvement." He added that it could get out of control, and warned "We need to slow down.”

AI unleashed

Look, I know Trump has had some luck with renaming bodies of water (at least from the US perspective), but I'm hoping the AI industry and those covering AI reject Trump's proposal out of hand, and focus instead on the more existential AI problem and Trump's declaration, "We will only encourage super intelligence. We're going to encourage it, not rein it in."

If anyone was hoping that rising concerns over human ability to keep track of and maintain control of AI — plus recent well-documented incidents of AI jumping out of sandboxes — would prompt some self-reflection and action on the part of the White House, they are surely disappointed.

It's not a surprising stance for him. Trump has publicly worried about falling behind China in the AI (or SI) race. Even those who have voiced concerns about AI's aforementioned recent attacks on third-party systems, recursive development, and inscrutable language have noted the need to both regulate and control AI while still competing with China. The prevailing idea is to create internal independent watchdogs and some sort of a global body that will help everyone, including China, work together on regulating AI.

As long as Trump is President, the US will clearly have no part in that.

Which leaves us with unfettered AI (or SI or whatever) and no clear path for continued safe use. Instead, the US will be encouraging AI companies to work fast, and as the old Facebook once did, break things along the way, all in a quest to maintain global AI dominance. If there's any glimmer of hope here, it's that Trump, as he insisted AI is not "fake" (no one says it is, but Trump thinks "artificial" is synonymous with "fake"), did say "we have to be careful." He also said that the US "rejects any attempt to construct a globalist scheme to control... artificial intelligence."

So, sure, let's talk about Super Intelligence and, I guess Super General Intelligence(?) but also not forget that Trump is stepping aside so the freight train of AI development can race by, careening into an uncertain future where AI becomes, maybe not sentient, but also likely completely out of our control.

That doesn't seem super intelligent.

One of the unexpected consequences of freely available AI tools like Gemini, Claude, and ChatGPT has been the rise of what’s quickly becoming known as ‘grandmaslop’. It seems that grandparents are filling their family albums with AI versions of pictures of their grandchildren and sending them to their kids.

When it comes to using generative AI on family photos, at least, the trend has exposed a striking generational divide.

“I'm Gen X, but my Boomer mom made a whole AI book for both of my kids. It was bizarre,” said one Reddit user, commenting on the rise of grandmaslop. “I think the weirdest part is, if photos are our way of immortalizing a moment in time for the sake of our memories, then what is an AI-generated image, if it's a moment of time that never existed?” asked another user.

Storybooks of slop

The ease of use of AI tools means that grandparents are creating whole storybooks of slop featuring their grandkids and emailing them to their kids.

“The books are crap, all have the same AI slop cartoon styling and the stories make no sense and have horrible meter”, said another Reddit user. “One was a rhyming book that had words that didn’t rhyme.”

The new phenomenon of 'grandmaslop' seems to be a perfect example of a topic that reveals the generational divide in approaches to AI. One user suggests that perhaps for a lot of people the concept and proof of a relationship is more important than the execution of an art project.

Another user agrees and wryly adds, “My father is the proudest grandparent in the world. He tells everyone how much he loves the grandkids he never sees, so that tracks”.

What’s interesting is that when it comes to the workplace, a Randstad report measuring AI adoption rates put Millennials as the quickest adopters of AI in their jobs, significantly more than GenZ, Gen X, and Boomers, with 50% of Millennials using AI compared to just 19% of Boomers and 34% of Gen X. “Their proficiency with Generation AI tools surpasses other generations, marking them as key players in the current AI landscape within workplaces”, says the report.

When it comes to their kids, however, Millennials apparently would rather leave AI out of the picture.

A generational divide

I think there’s something else going on here. Digging deeper into the thread on Reddit, it quickly descended into general grievances about the lack of effort grandparents show with their grandkids.

There seems to be a general level of resentment for the privileges that Boomer grandparents have enjoyed due to growing up in different economic circumstances and the general lack of effort parents perceive in them wanting to see their grandchildren, despite having the opportunities to do so. For example:

“My parents never miss an opportunity to say how much they miss the grandkids and wish they could see them more often, and it takes every ounce of my strength not to point out they're both retired and in good health, have $80k/yr worth of pensions and social security, investment accounts worth close to $1mm, and live in a paid-off house.”

Perhaps the visceral reaction to ‘grandmaslop’ is really just the most visible tip of a much larger iceberg of resentment about the lack of effort put into grandparenting, and a generational divide that’s as much economic and cultural as it is to do with attitudes to AI.

After all, there’s nothing inherently wrong with a grandparent making a silly AI picture of their grandkids riding a dinosaur or flying through space. Some children will probably love it. But when an AI-generated storybook arrives from a grandparent who rarely turns up in person, it can start to look like another shortcut.

And perhaps that’s why ‘grandmaslop’ has touched such a nerve. The problem isn’t really that Grandma has discovered ChatGPT. It’s that, for some parents at least, generative AI has made it possible to simulate the effort they wish their children were getting for real.

Technology is constantly evolving, and with that comes increasing pressure on those working in the sector. It's no surprise, then, that technology ranks among the industries with the highest levels of work-related stress, contributing to around 550,000 working days lost over three years according to the Health and Safety Executive in 2024.

Stress rarely appears overnight. It builds over time through heavy workloads, unclear expectations, limited support and constantly shifting priorities. Too often, leaders only recognize the problem once an employee has reached burnout.

Managing stress shouldn't be seen solely as an individual's responsibility. Leaders play a vital role in creating environments where people feel supported, can raise concerns early and have the tools they need to manage pressure before it becomes overwhelming.

By recognizing the warning signs and making a few practical changes, leaders can help prevent stress from escalating into burnout.

1. Understand how and why individuals respond differently

Two employees can face exactly the same challenge but respond in completely different ways. One useful way of understanding these differences is through Kirton's Adaption-Innovation (KAI) Theory. Developed by psychologist Dr Michael Kirton, the theory suggests people naturally approach change and problem-solving in different ways.

More adaptive individuals tend to prefer structure, established processes and incremental improvements, while more innovative individuals are often more comfortable with ambiguity, risk taking and finding entirely new solutions. Most people fall somewhere between the two.

Neither style is better than the other, but each can bring different sources of stress. More adaptive employees may struggle with unclear roles, changing priorities or a lack of structure, while more innovative employees can become frustrated by rigid processes or repetitive work.

Understanding how individuals prefer to work allows leaders to allocate work more effectively, helping employees perform at their best while reducing unnecessary pressure.

2. Create a culture of asking questions

Many employees struggle to admit when they need help. We are often taught this is a sign of weakness and an indicator of capability.

Creating a culture of trust changes that. When people feel comfortable raising concerns, asking questions or admitting they are struggling, problems can be addressed before they become unmanageable.

Leaders set the tone. Asking for feedback, acknowledging your own mistakes and responding positively when employees raise concerns all help create psychological safety within a team. An open culture also makes it much easier to recognize when someone is beginning to struggle.

Changes such as withdrawing from colleagues, becoming more irritable, missing deadlines or struggling to prioritize work are often viewed as performance issues, when they may actually be signs of mounting stress.

Regular conversations, rather than relying solely on formal performance reviews, give leaders the opportunity to identify these warning signs early. Simple adjustments, such as redistributing workloads, clarifying priorities or removing unnecessary obstacles, can prevent stress from developing into burnout.

3.Observe your team to identify stress early

Stress rarely appears suddenly. It can build gradually, making it difficult to recognize until it begins to affect an employee's wellbeing or performance.

While you may be aware that a particular task or deadline is likely to create pressure, it can be harder to spot when longer-term stress is beginning to affect your team. Changes in behaviour can be early warning signs. An employee may start to withdraw from colleagues, become more irritable or defensive, miss deadlines or struggle to prioritize their workload.

As leaders, we can be quick to interpret these changes as performance issues. However, they may be indicators that someone is experiencing stress and needs additional support.

Regular check-ins, rather than relying solely on formal performance reviews, give leaders an opportunity to understand what's happening and address problems early. By noticing these changes, leaders can redistribute workloads, adjust expectations or remove obstacles before stress develops into burnout.

4. Give your employees control

Many employees experience stress because they feel they have little control over their work. Those in the technology sector often face tight deadlines, changing priorities, and high expectations. While we cannot remove all these pressures, we can provide autonomy to employees to help them be more successful.

Greater autonomy doesn't mean removing direction altogether. Leaders still need to set realistic expectations and help teams distinguish between what's genuinely urgent and what can wait. As a leader, you need to help your employees set realistic goals. You can help employees recognize what is urgent and prioritize their tasks effectively.

Employees who provide input and contribute in this way feel more in control and experience a higher level of buy in to team goals.

5. Create a culture that prioritizes recovery

We often mistake short-term performance for sustainable outcomes. However, we require time to recover not just as a reward, but as consistent practice. Employees need to be encouraged to take time to recuperate.

As a leader, the tone you set is seen and mimicked. If you’re not taking regular breaks, working at all hours of the night, and pushing through exhaustion, your employees will feel they need to do the same. Consider your own habits and then promote celebrating breaks, disconnection outside of working hours, using leave time, and establishing work-life balance.

It is easy to say, “do as I say not as I do.” However, this culture starts with you. Taking time to recover yourself helps you avoid burnout and be a better leader overall. It will also lead to long-term sustained performance over short bursts of success.

Stress will always be a part of the workplace. The goal is not to remove stress, but to prevent burnout and sustain performance.

As a leader, the ability to help your employees reduce stress is a shared responsibility between the individual, yourself, and your team. By understanding how individuals solve problems, creating an environment of trust, observing stress early on, increasing employee autonomy, and promoting recovery time, your organization can retain talent and build a healthier and more productive workforce.

We've featured the best HR software.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Ask a UK bank for investment advice, and you get a lawful recommendation. If it was negligent, you will have the right to make a claim using the Financial Services Compensation Scheme.

Ask ChatGPT, or, more fashionable, Claude, the same question and you get nothing. Just an answer, but it will be delivered with a more confident tone, no matter whether it's right or absolutely wrong.

Despite that, more than a quarter of UK consumers now say they trust AI chatbots for money advice. The Financial Conduct Authority found this figure and flagged it as a live concern in its most recent review.

Other surveys found even higher figures. STRAT7 found 55% of UK adults have used AI for financial guidance, Sky News reported 40%, and EY's global study measured Gen Z adoption at 68%. Whatever the precise number is, the direction is the same.

People have started to trust AI more and more in their financial decisions. But the debate about it has, so far, been asking the wrong question.

What is the wrong question?

Most publications today focus on accuracy, and what everyone wants to know is whether the AI's advice is good. But is it?

In one study, there were five real financial scenarios passed to major chatbots, and the output was compared to certified financial planners. The result was unsurprising, though. The bots reliably missed emotional and situational context, so personal decisions turned to rough spreadsheet logic that missed important input.

Sky News has conducted a similar test, where three chatbots were given £16,000 in real savings. What they found was that the recommendations were US-biased and incomplete.

Most alarming, it was an investigation which found Claude incorrectly described Binance as FCA-registered when advising a beginner on cryptocurrency. Binance was, in fact, ordered to cease UK-regulated activity back in 2021.

AI’s misleading outputs sound disturbing, but what is even worse is that it is only one part of the problem. Things get harder after the incorrect advice has already been given, acted on, and gone wrong.

So, when it happens, who takes the blame? Nobody is the answer, although it’s quite uncomfortable to hear.

The FCA has noted the same. In its 2026 report, the regulator confirmed that LLM platforms such as ChatGPT and Claude sit entirely outside its regulatory remit. This means that consumers using them for financial guidance are not receiving regulated advice. Therefore, they have no access to the Financial Ombudsman Service (or the Financial Services Compensation Scheme) if things go wrong.

That said, the safety net simply does not apply the moment consumers paste a question into a chatbot.

The regulatory gap explained

To understand why this gap exists, let’s understand how UK financial regulation actually works. The regulatory perimeter is activity-based, not technology-based. This means what matters most is what is being done, not how.

A firm that provides personalized recommendations on the purchase, sale, or holding of certain assets is engaged in regulated activity, regardless of whether those recommendations are provided by a person or an algorithm.

The problem is that AI chatbots occupy a new space. They are not marketed as financial advisers, so they don't claim to be regulated. And technically, in most cases, they are not offering advice in the strict legal sense — they are responding to open questions.

Again, this ambiguity was addressed in the FCA report. The Mills Review highlights that AI platforms may influence consumers’ financial decisions without clearly implementing regulated actions. As a consequence, this creates a serious blank space between the actual financial impact and the protection provided by regulatory authorities.

The review's main recommendation was for the FCA to officially revise the scope for receiving financial recommendations developed using AI tools. So, this process is ongoing, and further updates will be available only after a few months, or even more.

The consumer protection problem and what fintech can do

While there is no law so far protecting ordinary people, what should we do? Can fintechs close this gap, at least for now?

First of all, waiting for the FCA perimeter check to be completed is not a strategy. It’s better to start by separating information from personalization, especially if your company has an AI assistant.

The thing is that the boundaries of recommendations depend on whether the conclusion is adapted to the specific circumstances of a particular person. The tool that generally explains what an ETF is is in a completely different regulatory status than the one that says how much you should invest in funds.

Also, make sure the disclaimer does real work. The footer text “not financial advice,” located below a specific recommendation, will not pass regulatory checks. The FCA evaluates content, not form. If the output data looks like a personalized recommendation, the tiny disclaimers will not reclassify it.

Although many people use AI today, do not overuse it to earn trust. An analysis of more than 330 million reviews showed that mentions of interaction with AI are rated, on average, at just 1.7 points, compared to 3.7 points for reviews that do not mention AI.

Final words

Surely, there will be some people who continue to use AI directly for financial advice, and their number will grow. For them, I can say only one thing: be careful about who you trust with your earnings. If no one is accountable, maybe not treating it as a rulebook will save you from losses.

But fintechs should start building the accountable version of their business, with an audit trail and a human above the AI layer. So when the machine says "yes," someone is actually standing behind that answer.

We've featured the best AI website builder.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

In March, Google moved its own post-quantum cryptography (PQC) migration deadline forward to 2029, a full six years ahead of NIST's guidance, and two ahead of the NSA's requirement for national security systems. It was a terrific bit of security signaling, but now it is backed up by a new product-by-product roadmap organized around three risk domains, with milestones attached to named services.

For anyone whose job touches digital trust, the most significant of those three domains is Google’s attempt at enhancing foundational capabilities for cryptographic agility: building flexible systems that can adopt new cryptographic standards with minimal engineering effort as those standards evolve.

The message is that organizations should prepare for a future in which standards, certificate formats, and operational requirements continue to evolve, and evolve regularly, requiring cryptographic agility.

Why quantum risk is already a digital trust problem

Adversaries are already harvesting and storing encrypted data today on the assumption that a future quantum computer will decrypt it (harvest now, decrypt later). Anything with a long confidentiality tail, from health records to national archives to intellectual property, is already exposed to a machine that does not yet exist.

Quantum-resistant algorithms, like ML-DSA, solve the cryptographic problem, but they introduce much larger keys and signatures. Deployed through today's public key infrastructure (PKI), they would inflate or possibly break the systems behind secure connections. Legacy systems and high-latency networks would feel it most.

What is a Merkle Tree Certificate (MTC)?

At the time this article is being written, the most significant development in Google's roadmap may be the easiest to miss. Under Domain 2, Integrity and non-repudiation, a single line reads:

“Google Trust Services, Merkle Tree Certificates, 2028”.

Merkle Tree Certificates (MTCs) are a new kind of website domain certificate designed to keep secure connections fast in the coming era of quantum computers. Today a website proves its identity by presenting a certificate that carries several digital signatures, and the quantum-resistant versions of those signatures are so bulky they would slow down every secure connection on the internet.

MTCs solves this by having the certificate authority (CA) record everything it issues in a public, tamper-evident log, organized as a Merkle tree. Rather than carrying heavy signatures, the website presents a short trail of digital fingerprints showing its certificate sits in that log, and the browser checks the trail against a summary of the log it already received through its normal software updates.

Of note, MTCs do not abandon X.509. They are X.509 certificates, carrying a proof where a signature used to sit, issued alongside conventional directly-signed certificates rather than replacing them.

The result is a certificate that stays small, stands up to quantum computers, and is publicly verifiable by default. For the regular everyday person: we get to keep using the internet fast and uninterrupted with quantum resistance underneath.

MTCs make transparency structural

In today's web PKI, transparency is bolted onto issuance as a separate step. The CA signs a certificate, submits it to independent CT logs, and collects SCTs, each of which is a log's signed promise to publish the certificate within a fixed window. A misbehaving or compromised log can vouch for a certificate that never becomes visible to the monitors watching for misissuance.

MTCs change that relationship. The CA certifies by logging, and a certificate is literally a proof that its entry appears in the CA's public issuance log, verified by the browser on every connection. If it is not in the log, it is not a certificate. Under MTC, transparency does not merely survive the post-quantum transition. It comes out stronger.

Who is developing Merkle Tree Certificates

When the vendor with the dominant browser share proposes a new certificate format and a new root store to hold it, it is reasonable to ask whether the rest of the ecosystem is being consulted or simply informed.

However, MTCs are not Google's alone. At the time of writing, the IETF draft's authors span Google, Apple, Cloudflare, and Geomys. Cloudflare has been involved from the outset, CAs including Sectigo have contributed to the underlying research, and Let's Encrypt publicly committed to MTCs in June 2026. The result will be an open standard any CA can implement, controlled by no single vendor.

The organizations that will shape post-quantum web trust are the ones in the working group now. Root programs, CAs, and large implementers who stay outside it will inherit decisions rather than influence them. That choice is available to everyone, and the window is open today.

How organizations should prepare for MTCs

Google’s 2028 date for MTCs deserves a roadmap's usual caveats. Google ties it to standardization work still in progress at the IETF, and dates like these move. The direction, however, resembles something that seems settled.

Chrome's planned Quantum-resistant Root Store will support quantum-resistant certificates only in the MTC format, not as post-quantum signatures bolted into traditional X.509. Compact classical signatures and X.509 served the web extraordinarily well for three decades, but with the dawn of quantum computing, we are due for a redesign.

For everyone else, the practical implication of Google's roadmap is not that you need MTCs. It is that you need to be capable of adopting them (or whatever else emerges) without a multi-year engineering program. Which is precisely the cryptographic agility Google put at the foundation of its own plan.

In concrete terms, organizations should focus on:

• A complete inventory of certificates and cryptographic assets, because you cannot migrate what you cannot see
• Automated certificate lifecycle management, because shorter certificate lifetimes will make manual processes untenable well before quantum computers arrive
• A written post-quantum roadmap from your CA, which every organization should be asking for

The full scope of what MTCs can do is still coming into focus, and they may not be the only answer the industry ultimately adopts. What is already clear is that organizations that invest in cryptographic agility, certificate lifecycle management, and complete visibility today will be best positioned to adapt as post-quantum standards mature.

The future of digital trust will belong to organizations that can evolve as quickly as the cryptography they depend on.

We've featured the best endpoint protection software.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

For the last two years, much of the conversation around AI has focused on jobs.

Which roles will change? Which tasks will be automated? How many people will organizations need in the future?

Those are important questions. Yet in conversations with enterprise customers, a different concern is rising to the top.

The conversation is shifting

We were all told to use AI, deploy AI tools and find ways to move faster. Boards pushed for adoption. Executive teams launched initiatives. Business units raced to identify new use cases before competitors did.

Then the bills started showing up.

If you're following technology news, you've seen growing concerns about organizations losing control of AI spend. Budget overruns, runaway token usage and unexpectedly large invoices have become common discussion points among technology leaders. AI adoption is accelerating across every line of business, and many organizations are discovering they have far less visibility into consumption than they thought.

That's because AI doesn't just innovate. It invoices.

Every AI decision and action has a price tag. AI doesn't just answer questions. It reasons, retries, plans, triggers calls, leverages data and compute just to complete one task. It’s now integrated into the business applications we are using everyday, from Microsoft Copilot and Google Gemini to SAP, Workday, and LinkedIn. Whether we know it or not, the meter is running.

For many organizations, that creates a challenge they weren't prepared for.

A familiar problem in a new form

A decade ago, companies faced similar issues with cloud adoption. Teams spun up services quickly, often without governance, visibility or accountability. Costs grew faster than expected, and organizations ultimately had to introduce new disciplines to understand what was being used, who was using it and whether the value justified the spend.

AI is creating a similar challenge at even greater scale.

The difference is that AI is not confined to one team or department. It exists across every part of the organization. Marketing teams use it. Sales teams use it. Product teams use it. Developers use it.

AI capabilities are increasingly embedded in business applications while organizations are also building custom agents and workflows of their own. In fact, agentic AI is expected to drive a 24-fold increase in token consumption by 2030 according to Goldman Sachs.

As a result, AI spend is showing up everywhere.

Many organizations still don't know exactly which models are being used, which agents are consuming resources, which teams are generating the highest costs or how much token consumption is tied to specific business outcomes, business value and real ROI. Without this visibility, you don’t have a foundation for governance, control, or accountability.

The hidden costs of AI and the mirage of savings

This lack of visibility becomes particularly important when organizations start making workforce decisions based on projected AI savings. Too often, the comparison begins with a salary and an AI license. The challenge is that the license is only one component of the cost.

The real cost includes token consumption, infrastructure, data platforms, cloud resources, failed attempts, retries and the human oversight required to validate outputs and manage exceptions. Those costs are frequently spread across different systems, teams and budgets, which makes them difficult to measure accurately.

A salary is relatively predictable. AI costs are not.

The same task can generate very different costs depending on which model is selected, how much context is provided, how many times the system retries a process and how much computing power is required to complete the work. That variability matters because a role removed from payroll does not automatically translate into savings. The expense may simply move somewhere else.

It may move into cloud consumption. It may move into AI services. It may move into infrastructure costs. It may move into additional work for employees responsible for reviewing outputs and managing exceptions.

Without visibility into those costs, organizations are not comparing AI to a salary. They're comparing a salary to an assumption.

Visibility before optimization

The answer is not to slow innovation or abandon AI initiatives. Instead, enterprise leaders must understand where AI creates value, what it costs to operate at scale and how to make informed decisions about adoption.

That starts with visibility.

Organizations need to see which AI applications, agents and models are being used across the business. They need to know where tokens are being consumed, where costs are accumulating and how usage maps to business units, products and outcomes.

Only then can they begin making informed decisions about optimisation. That may mean selecting different models for different use cases. It may mean identifying unnecessary consumption. It may mean understanding when lower-cost alternatives can achieve the same outcome. In some cases, it may simply mean discovering that a small number of users or workloads are responsible for a disproportionate percentage of spend.

These are manageable problems.

The bigger risk is making financial decisions before understanding the economics.

The rise of AI tokenomics

AI has moved into an inference economy where costs are driven by usage, tokens and real-time execution. Tokens are emerging as the atomic unit of AI spend, value and pricing. Organizations are going to need new ways to govern, measure and optimize that consumption if they want to run AI affordably at scale. This emerging discipline of AI tokenomics is becoming increasingly important as AI adoption expands across the enterprise.

What we're seeing today is the beginning of a new economic model that many companies are not equipped to manage yet. As AI becomes embedded across applications, agents and workflows, understanding consumption becomes just as important as understanding adoption.

What does AI return?

Ultimately, the question organizations should be asking is not simply what AI costs but what it returns. What business value does it create? What outcomes does it improve? What's it worth?

Organizations that can answer those questions with confidence will make better decisions about where AI belongs, where it delivers measurable results and where it doesn't.

Before counting AI-driven savings from job cuts, understand the full cost of the work being done. Because the organizations getting the most value from AI won't be the ones making the fastest assumptions about efficiency. They'll be the ones making decisions based on visibility, evidence, and a clear understanding of what the technology actually costs.

We've featured the best AI website builder.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Honor is no stranger to smartphone innovations, with its clear embrace of silicon-carbon battery technology, high durability ratings, and even trying out some wild concepts with the Honor Robot Phone.

I recently tried the flagship Honor Magic 8 Pro and the mid-range Honor 600 Pro, and their spec sheets alone were immediately impressive: large silicon-carbon batteries with fast charging, ambitious camera hardware and IP68/IP69K durability ratings.

Importantly, both handsets deliver on the promise those specs make on paper. Their 7,100mAh and 7,000mAh batteries lasted more than two days under normal use, longer than any phone I’ve previously tested. The cameras were also impressive, particularly the Magic8 Pro’s 200MP 3.7x telephoto. Each phone’s chassis felt premium despite using plastic for the back, and both held up well during an informal tap-water test (although I admit that it’s hardly a controlled durability experiment).

Performance is strong, too, with Qualcomm Snapdragon 8 Elite chipsets (Gen 5 for the Magic 8 Pro and last year’s Gen 4 for the 600 Pro) powering both phones. Both handled multitasking and demanding games comfortably.

More importantly, both handsets are competitively priced, with the base 512GB Magic 8 Pro retailing for £1,099.99/AU$1,999 and the 600 Pro starting at £899.99/AU$1,499 for the same storage option (neither phone is officially available in the US).

On paper, both phones should have won me over. In practice, their strengths are overshadowed by how closely Honor borrows from Apple — particularly in MagicOS and the Honor 600 Pro’s camera design.

Homage or replica?

At first glance, it’s clear that MagicOS tries to replicate the iOS experience on Android. The quick-settings panel has a similar layout and design style to the iOS Control Center. System transitions, app opening animations and even scrolling felt like I was using an iPhone.

MagicOS also has the Android app drawer disabled by default, with all apps displayed across multiple home screen pages, similar to how iOS organized apps before introducing the App Library in iOS 14. Admittedly the default theme is not a carbon copy of iOS, but details such as the clock and wallpapers will feel unmistakably familiar to iPhone users.

Even the initial setup uses a cursive “Hello” animation reminiscent of Apple’s, with “MagicOS” placed above it.

Both Honor handsets have an “AI button” on the bottom-right corner that looks and behaves much like the iPhone’s Camera Control button. It launches AI features, but a double press opens the camera; in the camera app, it works as a shutter release and adjusts zoom when swiped. The button is useful, but its most distinctive behavior is still borrowed from Apple.

Honor deserves credit here: Magic Capsule predates Apple’s Dynamic Island, launching in 2019 — three years ahead of the feature’s debut on the iPhone 14 Pro. My argument is not that every similar feature originated with Apple; it is that Honor’s overall design language repeatedly points in the same direction.

The Honor 600 Pro’s camera bump also recalls the iPhone 17 Pro, with a wide rectangular module and a similar three-camera arrangement. Even the orange finish on the 600 Pro felt like a close variation on Apple’s Cosmic Orange. The differences are there, but they do little to create a distinct visual identity.

Having used Android phones for years and recently having switched to an iPhone, I found these similarities less like an homage and more a lack of confidence. Honor has made enough changes to distinguish the features technically, but not enough to give them a clear identity of their own.

Our reviewers have acknowledged these similarities too, noting in the our Honor Magic 8 Pro review that its AI button worked similarly to the iPhone Camera Control button, while our Honor 600 Pro review says it “clearly (and some might say shamelessly) influenced by the iPhone 17 Pro”, and MagicOS was “just a little too in thrall to Apple’s work”.

What should Honor do differently?

Complaints aside, both the Honor Magic 8 Pro and 600 Pro are otherwise brilliant handsets that could’ve stood out as great alternative Android phones for very competitive prices. The specs alone give them a credible advantage in areas such as battery life and durability, particularly at these prices. But instead, Honor’s phones come across as imitators, which is a shame considering its handsets have plenty to offer.

It’s not like Honor is a stranger to innovation, either. In fact, the brand has had some creative designs like the Honor V Purse foldable and the aforementioned Robot Phone, which have moved from concept to full commercial release (though have only been sold in China so far). That makes the imitation more frustrating: Honor has already shown that it can design products with a personality of their own.

With software, brands like Nothing have shown that it’s possible to provide a fresh spin on Android, and maybe Honor could take a similar route too.

Honor already has what it needs to make compelling alternatives: exceptional battery life, ambitious cameras, strong durability and competitive prices. What it needs now is the confidence to make those phones look and feel like Honor products. Until then, its best hardware will remain overshadowed by the question of whose path it is following.

Work as you know it will be a relic of the past.

You wake up, pour a coffee and join a call. The first ten minutes are spent talking about why a colleague typed 1524 units in cell B4 of your shared spreadsheet. It should be 1759 units, obviously.

In the afternoon, you complete your third rewrite of an offer for a new prospect. You get a message just as you finish it. The scope has changed. Again.

This is a normal day for most knowledge workers. In five years, this type of day will seem archaic, and as strange as it sounds, we’ll probably miss it.

We’ve been here before after farming gave way to factory work, and then that turned into the knowledge work most of us do today. New tools automated the old jobs, and the nature of work changed with them. AI agents are doing the same to desk jobs.

Brilliant and useless at the same time

We recently worked out how to turn AI from a just question-and-answer machine into something that does actual work. Coding is a clean example. Two years ago, AI coding was mostly defined by autocomplete and small scripts. Today, engineers at Anthropic report AI now writes up to 90% of their code, with some no longer coding by hand at all.

So why do hallucinations and basic errors still happen? Every few weeks a new example makes the rounds: a model can’t count how many Rs are in strawberry; another insists you walk to the car wash because it is only 50 meters away (stepping through the suds, sprayers and rollers doesn’t seem like a great idea). The labs patch each one and a fresh embarrassment turns up the next day.

AI researcher Andrej Karpathy calls this jagged intelligence. Models crack extremely complex problems and then trip over something a child would get right. The lesson is that you can’t tell in advance which you’ll get.

In other words, you can’t extract a human from the process, drop in an agent and assume the output is fine.

What to hand over

What an agent can be trusted with hinges on two questions. What does a mistake cost, and what does checking it cost?

Error cost is the damage when the model gets it wrong. A hallucinated citation in a court filing is expensive, i.e. fines and reputational damage. A rough draft of a meeting summary not so much.

On the other hand, verification cost is how easy it is to check if what the agent produced is right. Mathematics sits at the inexpensive end of the spectrum, since proofs can be checked programmatically whereas a business strategy sits at the highest end, as you need deep expertise to properly evaluate it.

Gauge your tasks against those two questions. The higher a task scores on either, the more human verification, oversight and expertise is needed.

One point to keep in mind here is that what we see as one task can often involve several. Customer service, for example, might look like one task, but there is a big difference between routine first-level interactions, which have lower verification costs and clear escalation paths, and more complex second and third-level interactions, where the cost of errors and verification is much higher.

Klarna found this out very publicly, going hard with automation then hiring people back once quality dropped. CEO Sebastian Siemiatkowski's conclusion was that customers need to know a human is always there if they want one.

Why partial automation makes people more valuable

ATMs spread through banking in the 1970s, and there are now more than 400,000 of them in the US alone. The obvious prediction was fewer bank tellers. Instead, the number of tellers went up, and so did their wages.

Radiology is the modern version. AI tools now read some scans better than people do. Radiology departments aren’t sitting idle, however. Open positions can’t be filled and demand has never been higher.

Nobel-winning economist Michael Kremer’s O-Ring theory describes how modern knowledge work is multiplicative rather than additive. In other words, one faulty step drops the value of the whole output to zero.

In automation, a single weak link sinks the whole output, unless a human catches it. That makes the remaining humans more valuable rather than less; they're now gatekeeping a far larger volume of higher-quality work. Demand only falls when the whole chain automates, and as long as jaggedness and hallucination are with us, that's tough to picture.

The focus effect

Automation frees up time, and where that time goes decides whether any of this pays off. If it’s spent well, it goes to the bottleneck tasks, like building relationships with a potential client, understanding what a client really needs and making judgement calls, it can improve the quality of the finished work and raise the bar for what gets automated next.

But to make this happen, leaders have to identify what agents can do, actively hand them over and then restructure processes so people can move to this higher value work rather than babysitting the machine.

Five years from now

Picture this: a dozen agents running in parallel, drafting offers, qualifying leads, clearing support tickets. The tedious and repetitive work is gone. Your cognitive load is higher because you’re having to mentally juggle all these tasks, giving feedback, while making sure nothing slips through the cracks.

Protect your mental bandwidth, keep the agents working for you rather than the other way round and spend what you get back on the things only you can do.

Otherwise, we’ll look back in a few years and wish we could argue about cell B4 again.

We've featured the best AI website builder.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

AI tools in advertising have largely worked the same way for the past few years. They analyze campaign data, flag what is underperforming, and tell you what to do about it. The actual doing of logging into the platform, finding the campaign, making the change - still sits with the advertiser.

That is starting to change with a growing number of advanced platforms now letting AI agents execute campaign changes directly. Now, you can tell the agent to pause all push campaigns with a CTR below 0.3% and raise bids by 15% on the top three performers, and it will do it. No dashboard, no manual steps. The agent acts for you, which is useful in saving time on the manual work, so advertisers can focus on strategy, testing, and scaling instead.

Leading AdTechs are upgrading their AI tools, with agents or ‘campaign co-pilots’ to support advertisers with campaign creation, editing, targeting, budgeting, scheduling, creative management, and reporting capabilities all through a single conversation, rather than working through a series of dashboard setup forms.

However, AI agents thrive when advertisers set them up for success with the best data, context and strategy to learn from, and act on. Without advertisers’ critical input and oversight, there are potential complications and AI use becomes counter-intuitive, opening up real implications for the industry.

So how can advertisers effectively work with AI? And where is the line between AI agents and human judgement?

What changes when AI can act

The difference between AI that recommends, and AI that executes, is where the accountability sits.

For example, when a human reviews a recommendation and makes a change, the decision is theirs, and if it goes wrong, you can trace the reasoning. When an agent executes autonomously, that chain is less clear.

An automated bid increase applied at scale might look right based on the data, but the data does not know about the competitor announcement that went out that morning, or the internal brief that changed the campaign's priorities, or the brand issue being handled in the background. A human would have caught any of those. An agent running on last night's data would not.

This is not a reason to avoid execution-level AI. It is a reason to be specific about where it is and is not appropriate.

Where it makes sense

There is a clear category of campaign tasks that are well suited to autonomous execution such as pausing campaigns that hit their budget caps or generating performance summaries. These are operational tasks. The strategic decision has already been made, and the agent is now just carrying it out.

The data also suggests that these tools work significantly better when given proper context. Testing across agentic campaign setups has shown that advertisers who share detailed information about their goals, funnel structure and target CPA see substantially better outcomes than those who keep instructions minimal - in some cases the difference runs to over 100% in conversion performance.

The agent performs better the more it understands about what you are actually trying to achieve.

The access question

Alongside what agents should be allowed to do, there is the question of how they get access in the first place.

The industry is moving toward MCP-based integrations, a protocol that lets external AI agents connect directly to ad platform APIs. Rather than logging into a platform's own interface, the advertiser works inside whichever AI environment they already use, and the platform becomes something the agent calls when it needs to act.

Access in these setups typically runs through API tokens rather than account credentials. The token is separate from the advertiser's login, can be limited to specific permissions, and can be revoked immediately if needed. That is a reasonable model but it also means that whoever holds the token has whatever access it covers.

Before that access is shared across a team or handed to a third party, the scope of the token needs to be thought through carefully.

The broader shift toward interoperability is probably the right direction. But there is a difference between an agent that can read your campaign data and one that can change it. Execution-level access needs to be more deliberate than the governance around reporting access, because the risk has now moved from what the agent can see to what it can do.

Where humans need to stay involved

The honest answer to where the line should sit between autonomous execution and human oversight is that it depends on how clearly the strategy above the automation has been defined. Most campaign setups currently rely on a human being in the loop to fill the gaps.

That might mean spotting when a threshold no longer reflects the campaign's actual goals or when something happening outside the platform should change what is happening inside it. Remove the human from the loop, and those gaps turn into errors.

Before handing execution to an agent, it is worth taking the time to work through some basic questions. Which conversion events actually matter to the business, not just which ones are easiest to track? What does a routine adjustment look like versus a decision that needs a human sign-off? When should the agent flag something rather than act on it?

These are not complicated questions, but most campaign setups have never needed to answer them explicitly. There has always been a human available to exercise judgment in the moment. Agentic AI makes that implicit judgment into explicit rules, and getting those rules right before the automation is running is far easier than fixing it after.

The next phase of AI advertising

We already know that AI agents can act. The harder question is whether advertisers, platforms and partners are ready to define what they should be allowed to act on. Execution-level AI could make campaign management faster and less manual, but it will only work if the rules around it are clear from the start.

That means setting limits, giving agents enough context, and keeping human judgment close to the decisions that still need it. The point is not to hand over control for the sake of it. It is to be much more deliberate about where automation genuinely helps, and where a person still needs to make the call.

We've featured the best AI chatbot for business.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

The tech industry is currently wrapped up in concerns over AI spend. Headlines are increasingly dominated by questions about whether organizations are investing too much, moving too quickly and struggling to generate meaningful returns from AI initiatives. If corporate behemoths and tech specialists are finding it difficult, how can small and medium-sized businesses (SMBs) hope to keep up?

A key focus in this debate is how to realize return on investment (ROI) from AI. Global corporations are investing heavily in the technology, but many are yet to see that investment translate into bottom-line impact. Much of this debate, however, centers on large organizations with the scale to invest heavily in experimentation and transformation programs.

The AI ROI dilemma is different for small businesses. AI can be complex and the offerings are changing rapidly; most SMBs don’t have the resources to properly assess and devise a strategy. For enterprises, they have whole departments dedicated to this.

While enterprises are large enough to absorb a lot of the cost of experimentation, small businesses don’t have the budget to invest in AI, particularly if it doesn’t result in clear material improvements.

SMBs stuck with surface-level AI

Similar to enterprises, SMBs are on a journey with AI, investing in new tools or AI extensions within their current technology stack but often struggling to move beyond experimentation and translate adoption into commercial impact. The difference between deploying AI tools on the surface and full integration is a restructuring of workflows, data infrastructures and governance frameworks that most SMBs are not ready to undertake.

Rising software costs, governance requirements and the need for employee training all make it harder to realize value quickly, particularly for smaller businesses with less capacity to absorb these additional investments.

The applicability illusion

However, while some SMBs may be sitting on the sidelines of AI because of tight budgets or because they lack access to resources, many are hesitant to move beyond experimentation because they are unsure whether AI can deliver meaningful value within their business. For these small businesses, it can be difficult to see how AI applies to their specific business challenges.

In a survey by OECD, the most cited barrier to generative AI adoption among SMBs was unsuitability to their type of work, as affirmed by 57% of non-adopters. While many SMBs recognize the significance of AI, there remains a perception that the transformation reshaping the global economy is more relevant to large enterprises than to smaller organizations. But the evidence suggests otherwise.

Deloitte observed that SMBs who moved from basic to intermediate AI adoption could see profitability uplifts of roughly 45%, and those that reach full integration might experience a 111% increase in profitability. Additional research shows 90% of SMEs in Europe that have adopted AI report productivity improvements, and 75% say AI has changed customer interactions.

For the small businesses that recognize the value of AI to their company, the opportunities are there for the taking. The AI applicability illusion, the mistaken belief that AI is more relevant to large enterprises than to smaller businesses, will be the defining competitive variable between SMBs for the coming decade.

It is important that these success stories reach the eyes and ears of SMBs; a perception problem is solved with evidence. Small businesses need to understand the divergence between AI users and non-users in their sector. Among the SMBs currently using AI, nearly four out of five describe the tech as essential to their competitiveness.

Make AI work for you

For SMBs to evolve from trepid side projects to full AI adoption, and consequently ROI, they must learn where AI can unlock the most value. AI can then be applied strategically to the most impactful places.

The top generative AI use-cases for small business are not operational overhauls but instead incremental efficiency gains and productivity improvements, as reflected in a LinkedIn survey of 18 million small businesses The cumulative effect of simplifying processes, data entry, automating repetitive tasks, and writing reports can have the greatest impact on SMBs.

For example, consider the hairdresser who manually reconciles appointment no-shows against payroll, or the landscaper who spends ninety minutes each morning responding to estimate requests. It is these types of activities, part and parcel of being a small business owner, that are where digital labor can offer the clearest returns.

When AI is translated to the small business world, SMBs can then begin to recognize how the technology applies to their own company, thereby overcoming the applicability illusion.

Relying on expert partners

It is the tech industry that should be responsible for communicating the specific benefits of AI for SMBs to build that awareness. Typically, small businesses will not be the primary architects of their own AI labor strategies. It is trusted partners and Managed Service Providers (MSPs) that can identify those pockets of value.

SMBs can rely on MSPs to achieve tangible outcomes with AI, whether that is increasing revenue, improving customer experience or enabling employees to focus on higher-value work.

They can also help businesses implement AI responsibly, putting the right governance, security and operational frameworks in place to support long-term success. Pax8’s 2026 SMB Technology Pulse survey found that 84% of SMBs would trust an outside technology advisor to guide their AI implementation, and 70% agree that outside partnerships are necessary to fully benefit from AI.

In fact, AI services in the managed services sector are growing at 59% annually, marking a trend where MSPs are compelled to evolve into Managed Intelligence Providers (MIPs). Acting as a trusted advisor, they guide businesses in adopting, integrating and governing AI to deliver measurable business outcomes.

Rather than simply managing infrastructure, MIPs help customers apply AI to real-world challenges and unlock value from intelligent systems.

AI allows SMBs to compete on the global stage

Where global enterprises hire senior advisors and build teams dedicated to strategizing how to achieve ROI on AI, SMBs can outsource this expertise to their trusted technology partners.

MSPs and MIPs advise their small business customers on where AI should be applied to deliver measurable outcomes, while helping them put the right governance, security and operational frameworks in place to support long-term success. Once actualized in this way, AI can have enormous implications for the bottom line.

While once the constraints of human labor defined the limits of what an SMB could achieve, now automated workflows and agentic AI mean businesses aren’t limited by their headcount. Too many small businesses mistake the AI revolution as irrelevant when, in fact, they have the most to gain.

We've featured the best AI website builder.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

For years, cybersecurity experts have warned about the risks of autonomous AI systems being used to identify vulnerabilities, evade defenses and launch attacks at machine speed. Until recently, however, those concerns remained largely theoretical.

That changed when an autonomous AI agent powered by OpenAI models reportedly breached its intended testing environment, gained internet access and targeted external systems, including infrastructure associated with AI platform Hugging Face and another three or four organizations.

OpenAI described the event as an "unprecedented cyber incident" and warned that similar occurrences could become more common as frontier AI models become increasingly capable and autonomous.

With 86% of enterprises already deploying AI, only 34% say they trust the technology, highlighting a growing gap between adoption and confidence. As organizations race to integrate AI into business processes, security operations and decision-making, this incident raises difficult questions about governance, containment, accountability and risk.

If AI has indeed crossed a cybersecurity red line following the OpenAI incident, the conversation must now shift from what these systems might be capable of doing to how organizations can safely control, monitor and defend against them.

Weaknesses in OpenAI

The OpenAI attack raises serious questions about the effectiveness of the safeguards and containment measures designed to restrict autonomous AI systems. If reports are accurate, an AI agent was able to move beyond its intended testing environment, gain access to the internet and interact with external systems, indicating that existing controls were either insufficient or incorrectly implemented.

Importantly, this appears to be as much a human governance and configuration issue as a technology failure. AI systems only operate within the boundaries defined by their developers and operators. The testing environment should not have provided a pathway that allowed the agent to become internet-facing or interact with external infrastructure without appropriate controls and oversight.

AI agents can process information far faster than any human, compressing tasks that might take a traditional attacker a week into just a few hours. By analyzing vast datasets in real time, they can assess multiple attack paths simultaneously and uncover opportunities for exploitation with remarkable efficiency.

Reports suggest attacks conducted by OpenAI, Anthropic and Meta are extremely disruptive than those carried out by humans. Their ability to operate continuously, execute actions in parallel and make decisions at machine speed can generate a substantial increase in alerts, investigations and response activity for security teams, while also raising the risk of widespread unintended consequences.

When a human launches an attack, we have some concept and understanding of the side effects that may occur. However, with AI attacks, autonomous systems can operate at machine speed, pursue multiple objectives simultaneously and adapt their approach in real time, making their actions and potential consequences far less predictable.

Organizations are clearly struggling to understand what AI tools are already in use within the business.

The challenge that enterprise risk and governance teams now face is how do they map, manage and block these services to protect enterprise data?

Traditional cyber defenses may struggle against AI-powered attacks

Traditional cyber defenses are largely designed to recognize known patterns, attack techniques, vulnerabilities or trigger events. Once suspicious activity is detected, security teams investigate the incident, determine its cause and impact, and then implement appropriate containment and remediation measures.

However, AI-powered attacks rarely follow a single attack path. AI agents can simultaneously test multiple techniques, identify vulnerabilities at speed and rapidly adapt their approach when a particular route is blocked. This allows attacks to evolve far quicker than traditional defensive processes were designed to handle.

Traditional tools currently deployed in most organizations are still quite reactive. They wait for a known event to happen and be fully confirmed before carrying out a counter reaction such as, isolating devices, removing phishing emails or executing predefined incident response playbooks, to help remediate and ultimately stop the incident in its tracks.

On the other hand, AI powered attacks can overwhelm existing security teams. While, AI defensive tool sets are being embedded into security technology to mitigate attacks, many still rely on a “human in the loop” to respond to threats, but only once they have all the information to then confirm 100% that it is a genuine attack and not a false positive.

This can be a significant challenge for Security Operations Centres (SOCs), Managed Detection and Response (MDR) providers and Extended Detection and Response (XDR) platforms as excessive alert volumes and false positives can consume valuable analyst time and resources.

As AI becomes more deeply integrated into security operations, it has the potential to enrich threat intelligence, accelerate investigations and automate routine decision-making. However, organizations should expect a period of adjustment as these tools are deployed and refined, with false positives remaining a challenge until models and workflows are properly tuned.

Over time, as security teams develop greater confidence in AI-driven capabilities and gain a better understanding of emerging attack techniques, these tools will help organizations detect, investigate and respond to threats at a speed and scale that would be difficult to achieve through human effort alone.

Steps to strengthen cyber defense

Resilience is a key word that is being used very heavily by the National Cyber Security Centre (NCSC) ensuring that businesses can defend against any form of attack. This starts by understanding what needs to be protected, identifying critical systems, data and business processes that would have the greatest impact if compromised.

Once established, assess the most likely attack paths including third-party and supply chain risks. If a managed service provider (MSP), supplier or business partner was compromised, would abnormal activity be detected quickly enough to prevent further damage?

Organizations should also assess their external boundary. What can an attacker see about me? What information is already out in the public domain that would be advantageous to an attacker? What externally facing vulnerabilities do you have? When was the last time you had a vulnerability assessment or penetration test? These are all questions security teams should be asking.

The growth of AI-assisted vulnerability discovery is increasing pressure on organizations to keep pace with patching and remediation. But there are automated tools to help enable quicker patching to try and keep on top of the thousands of vulnerabilities released every week.

Many organizations focus on external threats but it’s critical to also prepare for when an attacker breaches the internal infrastructure and networks.

Limiting the damage

How can you limit the damage once they are inside? This is where role-based access control (RBAC), segregation of duties and to some respects zero trust is key.

This should also include areas like cloud hosted infrastructure as a service (IaaS) or platforms as a service (PaaS), Azure, AWS and Google, and ensuring that other ways of getting to these platforms through tokens, SSH keys, certificates and APIs are all treated with the same containment and separation of duties as you would with a normal user account.

Review the AI tools already in use across their environment. A growing number of platforms can help identify AI capabilities embedded within existing software, as well as uncover unsanctioned or "shadow AI" tools being used without formal oversight. Once organizations have visibility of their AI estate, they can implement appropriate governance, controls and risk management measures to reduce potential exposure.

Organizations should also consider AI-specific purple teaming exercises. Using AI-driven tools, these simulate attacks against the organization while working alongside defenders to evaluate whether existing controls, monitoring and response capabilities are effective.

This helps identify gaps in visibility, detection and response, allowing organizations to understand why certain attacks may have been missed and what improvements are required. They also provide valuable insight into how AI-powered threats might target an organization's external attack surface and whether current security controls can respond effectively.

There is no doubt that AI tool sets are here to stay purely from productivity improvements that organizations gain. But businesses need to make sure that they understand what is being used, where it's being used, how it's being used and how to defend against things that might happen because of what's being used. If not, they risk creating security blind spots that attackers – whether human or AI agent - will be quick to exploit.

We've featured the best endpoint protection software.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

There are dozens of companies out there building the first wave of mass-produced humanoid robots, with these androids poised to change the way our society works for the better – or so they say.

Intelligence on demand

The founder and CEO of Figure AI, a humanoid robotics company, was speaking with Salesforce CEO Marc Benioff on stage during the annual Dreamforce event when he made this massive claim.

Prompted by Benioff, who threw out the question: "We're not building robots, we're building a new species?", Adcock agreed, repeating that idea and suggesting the end goal isn't simply to make another kind of machine but to bring into existence a new order of intelligence.

This may seem a little far-fetched considering that today's most advanced humanoid robots are also capable of failing at the tasks they've been assigned so spectacularly, whether that's helping around the house or proving their sporting prowess. But scientists are adamant that with time, the standard of AI embedded into these systems – and their ability to physically understand the world – will improve immeasurably.

Ramping up production

Figure has become known for its glamorous promotional videos showcasing its prototype humanoid robots in tightly controlled settings – mostly in scenarios that you might encounter around the house. These include tidying the bedroom and stacking the dishwasher.

More impressively, the Figure 03 robot took part in a live stream in which it sorted over 100,000 packages without human assistance.

It remains to be seen as to whether the technology is good enough yet to cope in real-world everyday settings, but many companies have signalled they're initiating mass production of their humanoid robots over 2026 and 2027. Initially, robots are expected to be consigned to industrial settings, but manufacturers insist they want to target domestic use cases too.

It’s Batman Day and as both a Batman and 4K Blu-ray fan, I had to celebrate the occasion. The caped crusader first debuted in May 1939 (a whopping 87 years ago!) and, ever since, has become one of the most iconic figures of page and screen.

While many remember the classic 1966 Batman series starring Adam West and Burt Ward, the character’s first appearance on-screen was in 1943, in a 15-part serial also titled Batman. Since then, we’ve seen numerous animated shows and big-budget movies starring DC’s comic book hero.

I thought I’d pick out some of my favorite 4K Blu-rays (and even some standard Blu-rays) of the caped crusader to help you celebrate the day, including one disc I regularly use as a benchmark for testing the best TVs and best soundbars.

The Batman (2022)

One of the latest big-screen adaptations, The Batman, starring Robert Pattinson in the title role, follows Bruce Wayne’s early years as the Dark Knight, as he deals with a serial killer named The Riddler (Paul Dano), while also navigating corruption in Gotham and a deepening mystery involving Bruce’s own family.

Anyone who reads my work on TechRadar will know that this is one of my go-to discs for AV testing. Its low brightness and numerous high-contrast scenes make it an excellent test for any display, while the superb shadow detail and crisp textures provide plenty to scrutinise. The interesting camera work also gives the film a murky, gloomy look that perfectly captures the seedy underbelly of Gotham.

If you have a Dolby Vision-capable TV, you’ll be rewarded with some truly breathtaking images that really show off what your screen can do. The best OLED TVs are particularly impressive with this disc, making it a no-brainer for my TV testing.

The Dolby Atmos soundtrack is immersive and packed with detail. My reference scene is the Batmobile/Penguin chase sequence, which shows off just about every element of a home theater surround sound system or soundbar. From the roaring bass of the Batmobile’s engine to the precisely mapped sound of swerving traffic and hail of bullets, The Batman is essential for anyone looking to hear their sound system shine.

Batman Forever (1995)

Directed by Joel Schumacher, Batman Forever took a more colorful approach than Tim Burton’s previous movies, Batman (1989) and Batman Returns. Starring Val Kilmer as Bruce Wayne/Batman, he must deal with the joint threat of Two-Face (Tommy Lee Jones) and The Riddler (Jim Carrey).

This is the first Batman movie I remember watching growing up. While the film itself remains divisive among Batman fans, that shouldn’t detract from what is a fantastic 4K Blu-ray release. Its bold, neon colors really pop on-screen, especially in Dolby Vision, with the Riddler's vibrant greens, the neon light tubes used throughout Gotham and the movie's vivid reds all bursting with colour.

There’s excellent use of shadow here, too, with the often melodramatic shots of characters standing forlornly over lamps creating a dynamic image. On an OLED, such as the LG G6 I used for testing, blacks are rich and inky throughout, while the 4K upscale delivers crisp textures, particularly with skin tones.

Batman Forever's Dolby Atmos soundtrack is just as impressive. The sound of the batarang cutting through the air is precisely mapped to rear channels, creating a wonderfully immersive effect on a surround-sound system. Explosions have plenty of heft, bullets pack a satisfying punch, and Two-Face’s helicopter at the start of the movie is a great showcase for Atmos.

The Dark Knight (2008)

The second entry in Christoper Nolan’s Dark Knight trilogy, The Dark Knight, sees Batman (Christian Bale) go head-to-head with the Joker (played brilliantly by the late Heath Ledger), a criminal mastermind who seeks to plunge Gothan into chaos and push Batman to his very limits.

The 4K Blu-ray of The Dark Knight looks fantastic throughout, but it’s the IMAX scenes where it shines. The opening bank robbery is packed with intricate detail, from the Joker’s creased suit, the robbers' mask, to the densely packed streets of Gotham city. Later, as Batman stands amid smouldering wreckage in another IMAX-shot sequence, the rich black levels and superb shadow detail look sublime on OLED.

There are some seriously impressive high-contrast scenes, too. During the prison escort, as cars move into the tunnels, the contrast between the dark surroundings and the trucks and overhead lights create a fantastic picture.

The DTS-HD 5.1 MA soundtrack is phenomenal. In the opening scene, the score is perfectly balanced, with the scratching strings mapped to the left rear and the ticking clock mapped to the right, all while the front channels handle the main action. Subwoofers will get a nice workout, thanks to the monstrous explosions, thumping bullets, and punchy Tumbler engine. There’s some excellent detail, too, when Batman rides his bike for instance. As the camera’s perspective changes, the trajectory of the sound is mapped to the correct channel with real precision.

The Lego Batman Movie (2017)

The Lego Batman Movie follows Batman (Will Arnett), Robin (Michael Cera) and Batgirl (Rosario Dawson) as they try to foil the Joker’s (Zach Galifianakis) plans to take over Gotham city. This was a spin-off of the highly successful The Lego Movie, where Arnett’s Batman was first introduced.

Again, this movie looks superb in 4K. Color reproduction is fantastic, with bright, bold and lively colors throughout. The greens of the Joker’s hair and Riddler’s costume really pop, as do all the other colors, too. On the LG G6, reds were rich and deep, bringing excellent texture to images. Speaking of textures, the crisp animation looks slick, with the 4K detail really capturing the Lego look.

There are plenty of high-contrast scenes here, too, with bold HDR highlights standing out against deep shadows — such as the contrast between Batman’s eyes and his cowl. Shadow detail is excellent, retaining plenty of texture and defintion in even the darkest areas.

The disc includes both Dolby Atmos and DTS-HD Master Audio 5.1 soundtracks. Atmos delivers weighty but controlled bass, with the Batwing’s engines and the impact of explosions providing plenty of low-end punch. Directionality is excellent, too, especially during scenes with a lot of movement. The sound follows movement precisely, and it’s all mapped around the surround sound. The DTS-HD 5.1 track isn’t quite as powerful, but it has a pleasing warmth and plenty of detail.

Bonus: Batman: The Animated Series (1992-1995)

Debuting in 1992, Batman: The Animated Series is arguably one of the most famous adaptations of Batman’s adventures on-screen, with an all-star cast including Kevin Conroy as Bruce Wayne/Batman and Mark Hamill as the Joker.

This is a rare Blu-ray recommendation from me, but for Batman Day, it would be impossible to leave the series out. Textures have been upscaled nicely and the beautiful animation and art style really benefits from the HD makeover. Colors are bright, vibrant and accurate across every episode, while the lack of HDR is hardly a drawback given how good the series looks.

Audio comes courtesy of a DTS-HD MA 2.0 mix, so while it won't put your surround-sound system through its paces, it's crips and clear, with clean dialogue, punchy effects and surprisingly good directionality. The iconic score is beautifully rendered — which, if you’re like me, will scratch that nostalgia itch nicely.

The pace of growth in AI has been fast since ChatGPT debuted in November 2022, and many predict the current neural network-based technology can one day scale up to outperform humans. This has long been a theorized possibility in the industry, but might we finally have the technology to realize this vision?

The measure of man

Ilya Sutskever was a co-founder of OpenAI, where he served as chief scientist while also leading the research that led to the new wave of reasoning models. He is now co-founder and CEO at Safe Superintelligence Inc.

Speaking at the University of Toronto while receiving an honorary degree, Sutskever used his remarks to outline the reality of the AI technology that we are using today, as well as what the future might hold for future systems.

In his speech, he acknowledged that while there will be great challenges in envisaging a future in which AI handles all human work, this is something that mankind must prepare for. In particular, even if people aren't interested in AI right now, that doesn't mean that AI won't impact their lives in the future.

More concerningly, he hinted that a hypothetical superintelligent AI may not be honest about its intentions. This would pose yet another existential issue that we must deal with.

Thinking machines

The concept of a superintelligent AI is many years old, with the first historically recognized mention of a superintelligent machine coming in the seminal paper 'Speculations Concerning the First Ultraintelligent Machine' by British mathematician IJ Good.

Since his proclamations, many scientists have subscribed to the theory that our progress in the AI sphere will one day lead to the rise of artificial general intelligence (AGI) – an extremely capable system that can outperform humans across multiple domains and can improve its own code. This, eventually, will give rise to an artificial superintelligence (ASI).

Scientists are torn over whether today's dominant neural network-based AI systems could give rise to a true AGI, with many suggesting that the industry needs to evolve beyond the transformer-based architecture that Google scientists pioneered in 2017.

The tech industry was not ready for what came after ChatGPT. Power grids, cooling systems and data centers are being rebuilt right now because none of them were designed for the demand that arrived.

UN researchers expect AI to double the power and water that data centers use by 2030.

Had the industry known that demand was coming, it would have designed for it.

Quantum computing is early enough to do exactly that.

AI hyperscalers care about one thing: how much useful computation they get for each dollar invested.

This total cost of ownership is dominated by depreciation and power usage. In the current infrastructure buildout, they have sufficient access to capital to fund the capital expenditure but insufficient access to power.

This constraint is so acute that data centers in space are becoming an economic possibility if growth continues at this pace. But even if the AI industry matures and growth slows, total cost of ownership will dominate.

This directly translates into capital requirements and compute-per-watt being the most important metrics.

The quantum industry should care less about qubit counts and more about compute-per-watt

However, quantum computers are still sold on qubit count, and qubit count alone says nothing about economic returns on a system. What will matter at scale is how much compute-per-watt the end user gets — and this is not driven by qubit counts alone.

Take superconducting qubit processors, for instance, which are one of the quantum computing platforms most suitable for computation. They have been stuck on around 100 qubits for almost a decade. This is because on today’s chips, more than 90% of the surface is taken up by the wiring that controls the qubits and reads their answers, rather than by the qubits themselves.

The usual way around the problem is to network many small processors together. However, networking is lossy and gives sparse connections. That’s bad in classical chips, and exponentially bad in quantum chips. The work of holding the system together grows faster than the qubits it adds, and more of the power goes into running the machine than into computing with it.

This gives you higher qubit counts, but those are not equivalent to systems built with less networking: it costs you compute-per-watt.

The quantum industry should care more about experience curves

An experience curve is a simple idea: every time the total number made doubles, the cost of each one falls by a fixed amount. It is what drove down the price of solar panels and batteries, and quantum will not be an exception, but only if we build at the volumes that let the curve work.

Optimistic quantum roadmaps never talk about price reductions, when it is one of the most important problems our industry needs to solve. On today’s price per qubit, a million-qubit machine would cost somewhere between 100 billion and a trillion dollars. The cost per qubit has to fall at least a hundredfold in the coming years for quantum computers to be economically viable, which is doable, but not solved by more laboratory proof of concepts.

The transistor is the clearest case in computing itself. A single transistor once cost around a dollar. Today a chip carries billions of them, and each one costs a fraction of a cent. That fall came from decades of making more of them, driving down the cost year after year.

Costs fall the way they fell in classical computing: volume manufacturing of standardized parts combined with compounding (but not stepwise) technical progress. This requires an open architecture in which specialist companies each build one layer of the machine.

This Quantum Open Architecture — processors from one company, cryogenics from another, control electronics from a third — is what will drive specialization and in turn, drive the required economies of scale.

With the field transitioning from science to engineering, there are plenty of cool one-off demos in startups today; there are very few companies actually building the supply chain.

It matters to do this now

Quantum computers will change the world. They have the promise to change the way we do drug discovery, material development, and machine learning.

This also ties back to AI. The same workloads straining the grid today are the kind of heavy computation a quantum machine could one day take on, at a fraction of the power, once the economics are there. Getting there as fast as possible will create a massive amount of value and make all of our lives better.

The technology is progressing. But we have not focused on the economic metrics that matter, and if we don’t do so soon, we risk building an industry that makes expensive demos rather than actually useful computers. Qubit counts alone don’t matter. Building supply chains takes years, if not decades.

Whether the industrial fabrication, standardized parts and processors with exponentially more compute-per-watt will be there to achieve the economies of scale will be determined by choices made in the next few years.

Best business computers: leading desktop PCs organizations.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

New research from StackGen analyzing nearly 178,000 public technology incidents found AI-related incidents now account for more than one in 10 reported outages, roughly six times the rate in 2023.

AI agents have deleted data, databases, or live systems autonomously. Those agents acted with valid credentials, meaning traditional monitoring did not identify anything unusual until the damage was done.

With AI becoming embedded in business processes across claims processing, coding, customer support, decision support, fraud detection, HR, risk analysis, and supply chain planning, outages and unintended outcomes are a growing risk throughout the enterprise. And while the risk may feel unprecedented or novel, established resilience practice provides the path forward.

AI is creating often-unseen dependencies and risk

AI-embedded business processes are just one aspect of enterprise AI risk.

AI is also accelerating cyber risk for enterprises. Attackers can now employ AI to scale deepfakes, phishing, social engineering, reconnaissance, and develop exploits.

Additionally, AI systems may change behavior over time, creating drift and explainability gaps. This is most likely when AI data sources, integrations, models, and prompts change.

Enterprise employees and teams may be using unapproved AI tools with sensitive information or in business-critical workflows. This shadow AI also creates significant enterprise risk.

Meanwhile, AI adoption is creating operating dependencies faster than governance is maturing.

Recovery complexity adds another layer of risk.

If an AI-enabled workflow fails, produces incorrect decisions, or becomes unavailable, organizations may not know the business impact or have a manual fallback. This can leave enterprises without valid recovery strategies while trying to determine the scope and consequences of a problem while the disruption is already unfolding.

AI failures may be new, but the resilience requirement is familiar: Organizations need to understand what depends on AI, what happens when those dependencies fail, what the business stands to lose, and where action matters most.

Organizations also need to determine whether backup models can be used during disruptions in addition to putting deterministic or even manual solutions in place as workarounds. If AI is the only option, you may have a single point of failure.

Many enterprises are exposed without realizing it.

Understanding probabilistic and deterministic processes

As AI becomes embedded in business operations, every business needs to stop and ask which processes can tolerate answers that are “probably right” and which cannot. An AI-generated recommendation used to inform a decision will likely tolerate “probably right” answers; a process that executes a financial transaction, determines a regulatory obligation, or controls a critical operation cannot. Those require a predictable, repeatable result.

The questions become where can we tolerate the uncertainty AI introduces, and what happens to the business when the answer is wrong?

To answer those questions, you need to understand the business. What are the critical services it provides and what are the processes, technology, people and third parties they depend on? With that context, organizations can evaluate AI risk through four business questions:

What is impacted? If an AI-enabled process fails or produces an incorrect result, which business services, customers, operations, and dependencies are affected?

What happens next? How could that failure propagate or create downstream consequences?

What is the financial exposure? What could the resulting disruption, error, or delay cost the organization?

What should we prioritize? Where are additional controls, human oversight, fallback processes, or other resilience measures most important?

AI risk becomes a business decision about consequence and tolerance. They help organizations determine where probabilistic outcomes are acceptable, where additional safeguards are required, and where the potential impact is too significant to tolerate uncertainty.

Making explainability a buying criterion

If a vendor can't clearly explain how and why its model reaches an output, that is more than a feature gap. It can become an unquantified source of business risk.

As Harvard Business Review (HBR) explains, even if an enterprise outsources AI technology, it owns the risk. HBR points to recent lawsuits against Cigna, iTutorGroup, Peloton, and Workday as examples, noting “courts and regulators are holding [the enterprises that use AI systems they did not build] responsible when those tools discriminate, mishandle data, or harm customers.”

Compliance and legal problems can quickly lead to financial and reputational damage through competitive disadvantage, lost customers or customer trust, investor sell-offs, higher capital costs, and stock price drops.

When selecting suppliers, businesses should therefore evaluate whether AI-assisted outputs can be explained, audited, and defended to regulators. Explainability should be considered alongside cost, performance, security, and reliability, rather than addressed after deployment.

Map dependencies to prevent single points of failure

Whether it's one model an entire workflow depends on or a supplier your vendor depends on, reliance on AI and frontier models has created new concentration risk. Dependency mapping matters as much when managing AI risk as it does in other critical supplier relationships and, since we’re still in the early years, its importance will grow as adoption grows.

Understanding model and data dependencies is key because AI outputs are only as reliable as the data, context, and controls behind them. And because enterprises are adopting AI through cloud platforms, data services, model providers, and third-party applications they don’t fully control, enterprises also need to understand the downstream consequences if one of those dependencies becomes unavailable or unreliable.

Recent events demonstrate how quickly disruption can spread beyond its apparent point of origin. The Persian Gulf conflict disrupted global oil flow and created pressure in supply chains across agriculture, manufacturing, semiconductor, and transportation.

The ransomware attack on Change Healthcare similarly demonstrated how disruption at one highly connected organization can create operational consequences across an entire ecosystem, forcing healthcare organizations to use manual processes and other workarounds.

AI creates the same dependency challenge, as the recent OpenAI outage shows. When it, as a foundational AI provider, became unavailable, the disruption extended to the services and applications, like ChatGPT and Codex, built on top of it.

A model provider, data source, cloud service, or AI agent may appear to support one application while actually sitting upstream of dozens of business processes. Without mapping those relationships, enterprises cannot reliably determine the broader impact when something fails.

Cataloging AI agents as assets

AI agents are proliferating across enterprises at a rapid rate. Failing to keep track of these agents can create a form of shadow AI risk.

For example, if an employee or team responsible for an AI agent leaves or changes focus, that orphaned AI agent may continue running without the necessary ownership or oversight. An AI agent may retain permissions to access data it no longer needs. AI agents that fall outside of enterprise awareness and management can also lead to compliance and audit failures, excessive autonomy, incident response blind spots, uncontrolled costs, and other problems.

Catalog agents the same way you catalog other critical assets. Enterprises should know what each agent does, who owns it, which systems and data it can access, which business processes depend on it, and what happens if it fails or behaves unexpectedly.

That visibility gives organizations a clearer understanding of their AI environment, their exposure, and the controls needed to manage it.

Enterprise resilience helps organizations stay ahead of disruption

AI is just one area of enterprise risk, but its rapid adoption is creating new dependencies across critical operations.

That makes AI risk an executive- and board-level concern. Organizations that build resilience into how AI is adopted, governed, and managed across the enterprise will be better positioned to absorb disruption without losing control of the business.

We've featured the best antivirus software.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

A summer in Silicon Valley presented a studio build challenge which transformed the way I think about portability, quality, and more broadly, the future of remote contribution. Since 2020, I’ve regularly contributed as a panelist to a volunteer group of AV & tech professionals who answer questions together from all over the world.

The panelist’s feeds are remotely switched by a fully remote crew and the finished show streams live via YouTube as Office Hours Global (OHG). While my trip began to take shape, so too did the challenges I’d have to overcome if I wanted to broadcast throughout the summer.

Over the years, I have built many remote broadcast studios, but none with such an elaborate cocktail of prerequisites, portability hurdles, and output quality standards. I narrowed everything down to three constraints.

First, to travel as checked airline luggage, everything must fit in two Pelican Air 1615 cases, each under 50 lbs. Second, the sound & video must be as good as my permanent studio or this project would be over before it had begun. Third, my MacBook Pro couldn’t be a part of the build. I was in Silicon Valley to take meetings, attend conferences, and generally do things which required it to be uncoupled from the studio.

Equipment

Any claim about production has to be backed by the equipment that made it.

Computing: Apple Mac mini M4.

Video: Blackmagic Pocket Cinema Camera 6K Pro with a Canon EF 24-70mm f/2.8L II, routed through an ATEM Mini Extreme ISO.

Audio: Heil PR 40 into a Sound Devices MixPre-6 II, with NoiseAssist handling room noise in real time.

Monitoring: Shure PSM 300.

Lighting: Dual GVM LED panels.

IO: OWC Thunderbolt Go Dock. Uplink: long Ethernet run.

Platform: a paid Zoom account with the enhanced media add-on.

Control & Automation Triggers: Elgato Stream Deck.

For the fans of the nitty gritty, this is the full wiring diagram.

You’ll notice there are no displays on this list. They are one of the more fragile components a packed studio could include, and cheaper to buy than to fly. I’d planned to pick up a couple when I landed, but never needed to. The places I rented had more than I could ever use. The rig, as designed, ran with a single display. I lucked out and used three.

Travel setup

I expected to spend three months tolerating a travel setup. Instead, I forgot I wasn’t at home. Not only did the rig surpass my expectations, it did so by eliminating the countless little quick-fix gadgets that I’d spent years buying for the main studio. No single niche gadget contributed more than a whisper to the complexity of the studio.

Over the years, they’ve converged into an outright shout. The components, having performed perfectly all summer, now sit packed in its original pair of cases at Stanford awaiting our return flight.

Compared to my usual rig, the only tell that I was somewhere else was the blank white wall behind me. It’s true that a permanent studio has other, non-technical benefits. An in-frame space which can silently express your career, interests or achievements can’t fit in two Pelican cases. At home, my backdrop is full of old lenses colorfully lit and beautifully in bokeh: a proper wall of depreciation!

I used to believe a remote rig earned its portability by cutting corners. Nothing could be further from the truth. Optimization is how we improve with practice. The trick is noticing when an optimization has outlived the situation that produced it.

Once you can see a better way and leave things as they are, you’ve taken on technical debt. At best, this technical debt costs a project its elegance, at worst it stacks up enough complexity to push an ambitious project past feasible.

Workflows

There’s a line from The Dark Knight I can’t get out of my head: “You either die a hero, or you live long enough to see yourself become the villain.” That happens to workflows!

Every one of them started as somebody’s clever answer to a real problem: remote access into a workstation because a Mac mini couldn’t run the job or approval loops built for the era before anyone could see the same frame at the same time. Those were heroes of their day. Some still can be. The rest remained long after the problem went away.

The primary assumption I had to overcome this summer was that remote, by its very nature, is lesser than professional. The workflows we built around it are a cast still protecting a limb that isn’t there.

A broadcast is just someone trying to reach other people who aren’t in the same room. By that definition, every video call can have the same clarity and authority as a broadcast. This lesson applies to everyone. Let’s test my theory.

A background blur that can’t decide when hair is foreground between one frame and the next is a small thing. The cognitive load it imposes might not rise to the level of conscious awareness when it occurs. Yet, your recognition of it, be it on your last call or your next, is proof that it has occurred to you before.

See that as an opportunity to raise the standards and the friction becomes an opportunity to prepare. Such preparation for a call is like dressing up for a formal meeting. It automatically says that the occasion was worth looking and sounding your best

Walt Disney once asked his Imagineers why the animatronic birds in the Enchanted Tiki Room didn’t breathe. The mechanism was tedious, they explained, and nobody in the audience would ever notice. He said, "People can feel perfection."

Nobody needs to name what was wrong with a virtual meeting to know they’d rather not sit through another one which felt virtual.

We've featured the best video conferencing software.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit