• Chinese researchers report major hollow-core fiber upgrades
• Transmission capacities of 64Tbps were reached
• The technology could replace solid-core fiber in infrastructure networks

No matter how fast your broadband (or 5G) connection is, it could always be faster, and 'hollow-core' fiber could help. Researchers in China have now set new records for the technology, demonstrating two-way, high-speed data transmission across a distance of 1,007 kilometers (626 miles).

As reported by CCTV (via Interesting Engineering), a transmission capacity of 64 terabits per second (Tbps) was achieved in each direction, enough to download 160 films in 4K resolution (50GB each) every second — so perfect for movie night. Individual signals took just 3.43 milliseconds to travel the whole length of the link.

The transmission distance is some 50 times greater than previous hollow-core trials of the same type, while latency was reduced by 30%, too. Signal-to-noise ratios and loss margins were also improved, according to the team from the China Telecom Research Institute, Yangtze Optical Fibre and Cable (YOFC), and ZTE Corporation.

It's a technology that companies are already investing in. For example, last year Microsoft signed a deal with Corning and Heraeus to scale up production of hollow-core fiber in order to upgrade its Azure cloud platform. By the end of this year, around 15,000 kilometers (9,321 miles) of hollow-core fiber should be in place supporting Azure.

A core change

As you can guess from the name, hollow-core fiber replaces the glass of solid-core fiber with air that has a glass lattice surrounding it. The glass edges keep optical signals heading in the right direction, but the hollow core means there's less resistance to the light — which results in faster transmission speeds and lower latency.

The technology is better suited to large infrastructure projects than the cabling that connects your broadband network to your home, but you'll still feel the benefit in the coming years: data centers are already making use of the technology, and it's been tipped to support cellular networks too in the future.

Several challenges remain, because hollow-core fiber is more challenging to make and install than solid-core fiber, the current norm. However, as this latest demonstration in China shows, scientists are gradually overcoming some of the problems that remain, making the tech more commercially viable all the time.

The glass lattice that acts as the container for the data transmissions is particularly important, with recent improvements being made in the design and composition of this glass to reduce lag and increase stability. All of which means, we can expect to hear a lot more about hollow-core fiber going forward.

• A data center planned to be built in London's green belt has faced opposition from local residents
• The site will occupy more than 200 acres of green space, with the developers calling the site "a sustainable datacentre campus of the future"
• But the site will not meet Net Zero commitments and is projected to have the highest carbon emissions of any planned or operational UK data center

Residents beside a planned data center on the outskirts of London are not happy about the planned development which will see over 200 acres of greenbelt land ceded to developers to be converted into one of Europe’s largest data centers boasting 400,000 square meters of hardware capacity.

Data center developers in the UK have been having a tough time recently, with plans for the world’s second largest data center in Fife, Scotland, seeing widespread opposition, and a 450MW site in North Wales facing a rather negative reception.

But for some residents of North Ockendon in Havering, the idea of building a data center in green belt designated land is “sacrilege” and the development would damage the look and feel of the local community.

Residents oppose green belt development

According to multiple residents, who spoke with the BBC, the development would detract from the natural setting of North Ockendon and set a precedent for green belt land development.

North Ockendon resident Emma Edmonds said, “Most people absolutely don't want this. It's sacrilege to just build something like that here. It's not the right place at all,” while Danny Leach, whose property sits close to the planned development, said, “There will be the construction for the next eight to 20 years. There will be noise, there's going to be another thousand vehicles a day they're saying.”

“There will be disruption, the mess, and then, obviously, when it's operational, the background noise the data centres make and the amount of electric it uses - we already have two power cuts a month here,” Leach added. “We all need data, but do we need the largest data centre in Europe in this location?”

The North Ockendon Residents Association labelled the project as a “massacre” of green belt land, with a spokesperson adding, “It’s crop-producing farmland, in an age when we are threatened by security, it’s the loss of wildlife habitats, and there’s the pollution cost, which everyone seems to brush under the carpet.”

“This is rural. This is country. Leave it as it is,” local resident Graham Ford said.

Developers promise jobs, investment, and a park

In recent months, data center developers have had to become more vocal and involved with communities in the vicinity of a planned development. As the examples in Scotland and Wales have shown, if residents feel they are not getting enough out of a multi-billion pound investment they will do everything they can to make breaking ground as difficult as possible.

To that end, the UK government has announced plans that could force developers to invest in local communities as part of an AI “covenant” aimed at restoring trust between residents and developers and providing a worthwhile benefit to local communities.

The developers of the Havering data center, Digital Reef, have promised a range of benefits for local residents, including 620 full time jobs once the data center is operational, a 113-hectare Ecology Park with a 100%+ net biodiversity gain, and an £11.54 million community payment. The site is also expected to be Net Zero in operation with “zero onsite fossil fuels,” with the site “set to become a leading centre for skills development in the science, technology, sustainability, and data economy sectors.”

But despite claiming to be Net Zero in operation, Digital Reef has said in a planning application that the site “does not align with a science-based 1.5C compatible trajectory and achieving net zero by 2050.”

The site is also projected to have the highest projected carbon emissions of any project in the UK, according to The Guardian analysis of available data. Additional planning documents state that any shortfall of the zero-carbon target “would be provided through a cash in lieu contribution to the borough’s carbon offset fund,” a sum Digital Reef estimates to be about £77.6 million.

Many data centers across the UK, particularly in London, are facing issues with getting hooked up to the electricity grid. Microsoft recently withdrew residency plans from a planned London data center after learning that the site would not be operational until early 2030.

Couple that with a queue of over 140 data centers seeking more than 50GW in grid connections that, if approved, would use more than the UK’s 2026 peak energy demand.

For the planned development, Digital Reef said that at a 50% load the campus would use about 2.65 billion kWh per year of electricity — or the equivalent of more than 1 million UK households.

• OpenAI is retiring two beloved voice modes from ChatGPT — Standard and Advanced
• From December 11, Live mode will be the only voice mode in ChatGPT
• Some vocal ChatGPT users have taken to Reddit and X to complain

OpenAI is retiring the older Standard and Advanced Voice modes for ChatGPT in December across all its subscriptions, leaving only the newer Live Voice mode. Rather than simply diction, ChatGPT’s voice modes are a way of interacting with the chatbot using your voice, so you can have a real-time voice conversation. You talk, and ChatGPT talks back.

While OpenAI says that nearly all ChatGPT users are already using the newer Live Voice mode anyway, there’s a small, but very vocal contingent of the ChatGPT community who are very unhappy to see Standard and Advanced disappear, and they’ve taken to Reddit to express their outrage.

“I'm beyond disappointed because ChatGPT means a lot to me”, said one user, while another follows up by saying that they feel like they’re “paying a monthly subscription to beta-test their questionable decisions”.

The problem derives from a simple matter of preference for the older modes, the way they work, and in particular the sound of the voices they use.

How the three voice modes differ

The big change happens on December 11. Before then, to change between the three voice modes you need to go to Settings, then Voice and choose the mode you want.

Standard mode works by converting your speech to text, then generating a text answer before reading it aloud. Advanced mode works by processing speech directly, but takes turns speaking and listening. That means no interruptions.

The newer Live mode can listen and speak simultaneously with more natural interruptions and responses. Personally, Live mode is my favorite by a long way. To me it feels less robotic and more like a human conversation; however it seems that not everybody is a fan.

There seem to be a couple of strong reasons why people object to the change. In many cases, they’ve become attached to a particular voice. One Reddit poster describes using the Vale voice for years and finding it comforting, while selecting Vale in GPT-Live doesn’t reproduce exactly the same experience.

Another aspect is that the supposedly more natural conversations of Live actually feel less natural to some people. It does a lot of ‘yeah’ and ‘anyway’ to fill in the gaps while it’s thinking. It interrupts, breaking the user's train of thought, and there are more changes in tone. Many people find this irritating.

Getting some perspective

I think we need to remember that the change doesn’t mean that we’re losing voice in ChatGPT entirely. OpenAI is simply consolidating voice into the Live model so that it can devote more development time to it without having to maintain Standard and Advanced as well.

In a statement on its Help center, OpenAI said: “We recognize that some people are accustomed to Advanced or Standard voice’s style or way of responding, and these model transitions can be challenging. Our Help Center article on navigating model transitions offers guidance to help with the change.”

Of course, it remains to be seen how much the opposition to this change will grow. This isn’t the first time OpenAI has tried to retire Standard Voice mode. In September 2025, it announced that Standard Voice would be retired within 30 days. But on September 9, the company reversed course, acknowledging that users valued the older experience and saying it would remain available while OpenAI addressed feedback.

While this new announcement feels more permanent, it will be interesting to see what happens if resistance to the changes continues to grow.

The TP-Link LS1210GP is a solid choice for expanding a home or small business network without introducing a tangle of power adapters. With eight PoE+ ports, gigabit connectivity, and a fanless design, it can handle IP cameras, wireless access points, and other network devices.

Best For: Home and small business networks, particularly security camera installations and setups requiring several PoE-powered devices.

The Deal: Amazon.co.uk has reduced the price of the TP-Link LS1210GP from £82.99 to £49.99 for Black Friday. That's a £33 saving, or a solid 40% off.

Best TP-Link switch deal

Why we recommend it

The TP-Link LS1210GP offers nine gigabit RJ45 ports, eight of which support PoE+, alongside a gigabit SFP port for fibre connectivity. Its 20Gbps switching capacity provides enough bandwidth for multiple gigabit connections without making the switch itself an obvious bottleneck.

Each PoE+ port can deliver up to 30W, although the total power budget is limited to 61W. You can connect eight PoE devices, but you can't power all eight simultaneously at their maximum rate.

Extend Mode increases PoE transmission distance to 250 metres for surveillance installations, but this comes with a reduction in connection speed to 10Mbps on extended ports.

PoE Auto Recovery can restart unresponsive powered devices automatically, potentially saving a trip to a remotely installed camera or access point, and Isolation Mode separates traffic between connected devices.

For more of our favorites, see our guide to the best network switches we've tested.

Price context & historical value

At £49.99, the TP-Link LS1210GP currently costs £33 less than its usual £82.99 price. The combination of eight PoE+ ports, a dedicated SFP connection, and automatic device recovery makes the current price an attractive option for smaller surveillance networks.

This is the cheapest price we've seen it go for, so if you're after a network switch, we'd definitely recommend you buy it now before the price goes back up again.

Should you buy it?

✅ Buy the TP-Link LS1210GP if...

If you're installing several IP cameras, access points, or VoIP phones and want a quiet, straightforward gigabit switch with PoE+ support.

❌ Skip the TP-Link LS1210GP if...

If your devices collectively require more than 61W, you need 2.5Gbps networking, or you want advanced managed-switch features such as VLAN configuration via a web interface.

• Shop all the network switches at Amazon

• Burnout is affecting a concerning 87% of women working in tech, new report finds
• Demand for new AI productivity skills, attendance and visibility, and personal priorities are contributing to the burnout
• 1,500 female-identifying technology professionals surveyed for 2026 Speak Up report

A new survey examining the roles of women working in the tech industry has uncovered a concerning trend: burnout.

Surveying 1,500 female-identifying tech professionals, the 2026 Speak Up report found a worrying 87% said that they were experiencing burnout, with over half considering themselves to be either moderately, very, or extremely burned out.

Various pressures are identifies, but AI has introduced a new dimension to the role of women in tech, contributing to changing expectations and driving that feeling of burnout.

Speak up

The report from Ensono explores how improved AI-driven productivity in the workplace is producing increased pressure on tech-employed women.

AI knowledge and skills among 70% of the subjects is considered to be advanced or expert, with 97% able to identify at least on career benefit from using the technology. But the impact of using AI is one of the elements that have been cited as contributing to burnout. 42% of those surveyed feel that using AI comes with an expectation that more work can be produced in less time.

Furthermore, 34% of the women surveyed felt pressure to be always available for work and 43% were focused on constantly developing new skills for work.

Throw in the drive to be visible and to make a mark – 83% aspire to executive leadership roles – and there is a vast amount of pressure to succeed and to be seen doing so.

Building lasting careers

Burnout is avoidable, of course, with the right workplace policies and clearer expectations for colleagues. AI, visibility, leadership – they’re all challenges, but they do not have to dominate the working day.

Meredith Graham, Ensono’s Chief People Officer, said “Women in technology are developing the AI expertise their organizations need, but continued learning takes time to adjust. With skills confidence continuing to grow, business leaders need to provide clarity on how AI adoption changes work expectations.”

The arrival of AI has disrupted many elements of life and work, evolving expectations in the workplace, and these results suggests employers can do more to protect team members from burnout.

Graham added: “As roles evolve, supporting flexibility and different paths to advancement that help women build lasting careers is fundamental to strengthening business culture.”

• AI agents have found what looks like an undiscovered Maya site
• The discovery came from a scan of LiDAR data collected by Mexico
• Experts are now investigating the site that the AI has flagged up

While some of AI's 'talents' — like putting together decent-looking restaurant menus — could legitimately be called into question, there's no doubt that these bots and agents excel at spotting patterns in vast troves of data, and the latest evidence comes from the potential discovery of a hidden Maya archaeological site in Mexico.

Google researcher Pietro Miotti described having a "random thought on vacation" about how AI agents might be used to sift through LiDAR datasets — topographical maps of Earth's terrain, captured from space satellites.

LiDAR (Light Detection and Ranging) lasers can cut through cloud cover and forest canopies, enabling researchers to find shapes in the landscape that aren't visible to the naked eye (or in landscapes that aren't even accessible to archaeologists).

Miotti tasked 48 agents from different AI apps with scouring through a NASA database of LiDAR information collected across Peru, and the AI came back with multiple hits on structures that could be buildings. A second scan by the agents identified one particular site that hasn't yet been mentioned in scientific literature.

'LiDAR the entire planet'

The AI assistance didn't stop there: Miotti's bot helpers told him who he should get in touch with about the discovery (Mexico's National Institute of Anthropology and History or INAH), and then composed an email (in Spanish) to the relevant agency, including details of the find and where on the map it's located.

Representatives from INAH got in touch, and the experts there are investigating further: a real result, in 72 hours. The site's status as an ancient Maya site hasn't been confirmed yet, but it looks like a promising lead, and one that was made possible by AI.

"I think we're on the brink of a revolution across every science, not just coding, physics and math," says Miotti. "And at this point we should LiDAR the entire planet and let agents help us find the rest."

Miotti still wasn't finished there, and got Gemini to generate an image of how the site may have looked in 700 CE — when the Maya people were actually living there. In the modern era, some serious exploring can be done from a computer through a team of AI bots, as well as out in the field.

OK, so you've read our reviews: you know the headphones and earbuds we recommend. But how does that work for you — or indeed the person you're looking to buy for while the deals are dropping this Black Friday?

I've got you. I'm TechRadar's audio editor, and I've been doing this since 2019. I've just compiled a recommender in the form of a simple quiz that you can take (or casually ask them to take) to find out what I'd recommend.

Now, before you write in, you should be given options at various price points upon completion of the quiz (which literally takes 30 seconds), but just in case that doesn't happen, I'm listing every single product I picked within this headphone recommender, so you can read up on those too and add them to your list.

If you click through to the 'view details' tab, you'll even get a little bullet-point guide on each model before you shop the deals, too.

Take the quiz — find your next buds or cans

All of the headphones and earbuds explained

OK so you've got your top pick — or top picks. Now what?

Now, you read up on the specific model you've been recommended or the close contenders that, depending on your answers, I haven't recommended specifically for you (but do recommend, in this headphone-finder quiz).

Lastly, check out any deals already live, by clicking on the product. We got this, friend!

So, what did you get? Feel free to drop me a comment. I'm TechRadar's Audio Editor, and I'm all ears…

Again, any of the headphones or earbuds above come with our recommendation, based on our thorough review process — I just want to help you narrow it down ahead of the busy sales extravaganza (and of course, the gift-giving season).

• Attackers exploited a zero-day in Railway’s build infrastructure to steal customer AI API keys and cloud credentials
• Stolen keys fueled roughly $740,000 in unauthorized AI usage, likely supporting large-scale model training and automation
• Railway spent 26 hours battling AI-assisted attackers before containing the intrusion and rotating credentials

US-based cloud hosting company Railway said threat actors abused a previously unknown vulnerability to break in, steal user credentials, and rake up hundreds of thousands of dollars in unauthorized AI usage.

The company managed to throw out the criminals and harden its infrastructure, but not before witnessing an epic 26-hour bot-versus-bot battle.

Railway gets attacked

At its core, Railway is a cloud hosting company, but one designed primarily for developers to build, deploy, and run applications without having to manage much of the underlying infrastructure.

Its key services include running websites and APIs, hosting databases and deploying PostgreSQL and MySQL, connecting GitHub repositories and deploying new code automatically, and handling containers, networking and resource allocation.

Similar to other cloud providers, its build infrastructure uses a multi-tenant architecture, meaning a single device hosts multiple builds which should, in theory, be in full isolation from each other on the same virtual machine.

Many of its customers are developers and businesses using AI services, with API keys stored as environment variables on the platform. This allows the hosted applications to communicate with AI models more easily.

Now, in an incident report, Railway said unidentified threat actors used a zero-day vulnerability in its build infrastructure to “partially escape filesystem isolation” and read some of the files stored on the host virtual machines. In these files were apparently credentials the build servers use to authenticate to a subset of internal systems.

“After they acquired this credential, the attacker attempted but initially failed to use it in a variety of systems,” the report stated. However, they eventually realized the credential allowed access to Temporal, the company’s internal task management system. There they found a copy of Railway’s internal codebase in which they discovered a misconfigured authentication mechanism that allowed them to nab temporary cloud credentials.

With these, they accessed other internal systems, including credentials needed to access a production database with encrypted customer environment variables, as well as information on how to decrypt them.

API keys for AI services

The attackers were apparently mostly interested in API keys of businesses using AI services. Using the keys, they were able to access paid AI services, leaving the victims to pick up the tab.

And quite the tab it was - roughly $740,000 charged for using coding agents and AI training pipelines. Railway believes whoever obtained the keys likely sold them on the black market, possibly to support model distillation.

Model distillation is when a smaller AI model is trained to imitate the capabilities and responses of a larger, more powerful model. This is usually done by sending a large number of queries to the original model, and using whatever it returns as training data for the smaller ones. Although it is impossible to prove, many analysts believe this is how Chinese models are able to compete with US ones.

Railway estimates environment variables of up to 100,000 customer workspaces could have been decrypted during the incident, but stresses the number represents an upper limit rather than a confirmed number of compromised accounts.

The company said the attack started on October 20 at roughly 07:28 UTC, and ended at 09:24 UTC on October 21, some 26 hours later. During that time the engineers, working in shifts, saw the attacker probing many potential routes in parallel, as Railway was hardening them, also in parallel. “The escalating situation became a contest of agent speed, with Railway defending against abliterated models and a harness similar to ARTEX, according to Wiz,” it said in the report.

Railway said it has since closed the exploited vulnerabilities, rebuilt affected infrastructure, and rotated compromised credentials. It is also working with AI providers to refund fraudulent charges, promising to cover losses that the providers refuse to reimburse.

Via BleepingComputer

The Curse of Oak Island returns for season 14, and Rick Lagina still believes "100%" that treasure lies beneath the Island.

"This is the year to watch."

If you're in search of how to watch The Curse of Oak Island season 14 online from anywhere in the world – and potentially for free, we'll give you the map.

• November 10, 2026 – Airs 9pm ET/PT (8pm CT)
• Watch (US): History via Sling TV/DIRECTV/FrndlyTV/Philo
• Outside US: Unlock your stream with NordVPN (save 75%)

After more than a decade of searches, skeptics could be forgiven for having serious doubts as to wether there actually is any treasure under Oak Island.

But the treasure hunting brothers are still confident. This season, They'll explore The Money Pit, Lot 5, Lot 8 and The Swamp, making discoveries as they go, in a treasure hunt that's kept us captivated since 2014

Here's where to watch The Curse of Oak Island season 14 online in the US, UK, Australia, Canada, Europe and beyond, potentially for free.

Watch The Curse of Oak Island S14 in the US

Viewers in the US can watch new episodes of The Curse of Oak Island season 14 on the History Channel.

Don't have cable? Cable-cutting services are your friend! The History Channel is available in selected packages on:

Philo (7-day free trial)

DIRECTV (5-day free trial)

FrndlyTV (7-day free trial)

Sling

Stream The Curse of Oak Island S14 from anywhere

If you’re abroad when the 2026 season of The Curse of Oak Island drops, it's worth knowing that you can download a VPN and use it to access streaming services from abroad.

Free VPNs are pretty much useless for this, so you'll need a top streaming VPN – we recommend NordVPN (save up to 75%)

Can I watch The Curse of Oak Island 2025 in Canada?

No, season 14 is not listed although on either the Global TV schedule or the StackTV streaming service.

Remember, if you're visiting Canada from the USA, you can use NordVPN to access your usual US streaming services.

How to watch The Curse of Oak Island S14 in Australia

There is not yet a home confirmed for The Curse of Oak Island season 14 in Australia, though in the past the show has appeared on Binge, SBS and 9now.

American down under? You can use a VPN to access all of your usual streams, just like you were at home!

Where to watch The Curse of Oak Island season 14 online in the UK

Last season, The Curse of Oak Island aired on Sky History in January in the UK, 3 months after is November release date in the States. If the same is true for this season, then it'll be early 2027 before Brits get to watch season 14.

American visiting across the pond? You can use a VPN to access all your streams like normal.

• How to watch The Celebrity Traitors online – it's free
• Sling TV - subscriptions, how to sign up, live TV streaming
• How to watch Grey's Anatomy

• Nitori discloses cyberattack that may have exposed data from up to 37.6 million customer records
• Stolen information included personal and business customer details, though passwords and payment data were unaffected
• Company remediated the intrusion, continues investigating the breach, and warned users about phishing attempts

Japan’s largest furniture chain Nitori Holdings has disclosed suffering a cyberattack which saw it lose tens of millions of customer records.

Both consumer and business customers are affected, but passwords and payment data have not been compromised.

In a statement, Nitori said it spotted unauthorized access to its systems on November 6 2026 - and after ousting the intruder and launching an investigation, the company discovered that up to 37.6 million of customer records may have been exfiltrated, including people’s first and last names, membership numbers, email addresses, postal addresses, phone numbers, gender information, as well as birth dates.

Nitori's warning

Aside from these records, corporate members also had their company names and department names exposed, as well as names of company representatives, phone numbers, designated contact persons, and other details.

The specific combination of information varies from customer to customer, it said, adding that the data was most likely generated by Nitori Net and Nitori’s official stores on third-party e-commerce marketplaces. The company also stressed that credit card information and login credentials were not exposed.

The investigation is currently ongoing and crucial details are still missing. We don’t know who the threat actors are, or if they reached out demanding payment in exchange for deleting the stolen files.

Nitori also said it identified the flaws used to break in, remedied them, but was still analyzing and monitoring them.

“We have addressed the identified vulnerabilities and completed improvements to our monitoring systems”, it said.

“We are continuing to work with external specialists to investigate the cause of the incident, the method of intrusion, the extent of the impact, and whether any secondary damage has occurred.”

Often compared to IKEA, Nitori Holdings is considered Japan’s largest home furnishing retailer, selling affordable furniture, home accessories, kitchenware, and similar household products. It operates more than 1,000 stores across the region and 11 overseas markets, including China, Taiwan, South Korea, and a handful of Southeast Asian countries. Last year it reported $6.1 billion in revenue, while employing more than 41,000 people, and making almost 93 million customer purchases. Its mobile apps have more than 25 million registered members.

Those millions of members and thousands of employees are now being warned to pay attention to their inboxes for incoming email pretending to be from Nitori, or having to do something with the company and the breach.

“Personal information that may have been leaked could be misused for impersonation emails, phishing scams, suspicious telephone calls, and other fraudulent activities,” the company said in the announcement.

“We urge customers to exercise particular caution when receiving communications claiming to come from our company, delivery service providers, or e-commerce marketplaces.”

“Please do not open unfamiliar links or attachments.”

Japan under siege

There has been a barrage of cyberattacks against Japanese companies lately, with Iris Ohyama, a major Japanese consumer goods manufacturer, confirmed losing data on 7.7 million people in a data breach that affected its ecommerce website Iris Plaza.

Internet giant GMO also recently told its customers to unplug their devices from the network, after a cybercriminal threatened to release ransomware on user machines, and just before that gaming behemoth Capcom reported losing sensitive information on more than 11.7 million people. Most of them had their emails exposed, with a small subset also having to change their passwords.

Strangely enough, no threat actors claimed responsibility for any of these incidents. Usually, criminal organizations would try to extort money for the data, rather than trying to sell it on the black market, since some companies are more willing to pay. Leaking data could trigger class-action lawsuits or pressure from regulators and data watchdogs.

Via BleepingComputer

Having kicked off its Black Friday sale at the end of October, Argos has now expanded it with several new offers. That means it's time for me to take another look through it and pick out the top deals on tech, gaming, appliances, homeware, toys, and more. I've got 18 recommendations for you here, from the likes of Ninja, Samsung, Shark, Garmin, Dyson, and many other leading brands.

• Browse the full Argos Black Friday sale

The most exciting new addition is this Ninja AF200UK Air Fryer for £99 (was £159.99). As a smaller and more compact version of the five-star-rated AF400UK, it's a good option if you want to spend a little less, but still get an appliance from one of the top air fryer manufacturers.

I also like the Amazon Kindle for £84.99 (was £134.99). A classic of any Black Friday sale, this is the same price I saw it for during the recent Prime Day event at Amazon. There is a new version of the Kindle available now, so this price cut is likely a way to clear out the old stock, but it's still a very capable entry-level ereader.

Given we're still a few weeks from Black Friday proper, I'll just remind you that Argos is offering a 'Price Guarantee' on select products throughout its sale. It means that any item displaying that badge will not fall any further in price before December 25. If for some reason it does, you will receive an Argos voucher for the difference.

That's good peace of mind to have if you were planning to do any shopping right now. I'm sure even more offers will be added to the Argos Black Friday sale over the next couple of weeks, so stick with TechRadar for even more recommendations from myself and the rest of the team in our full Black Friday coverage.

Argos Black Friday sale — quick links

• Appliances: up to 20% off fridges & dishwashers
• DIY: up to 30% off tools & outdoor
• Fitness: up to 50% off exercise equipment
• Furniture: up to 50% off
• Gaming: up to 50% off PS5 & Switch games
• Headphones: from £19.99
• Health & Beauty: up to 50% off GHD & Philips
• Kitchen: up to 50% off small appliances
• Laptops: from £149
• Phones: up to £250 off Samsung & Honor
• Toys: up to 50% off
• TVs: from £149
• Vacuums: Shark & Dyson from £99.99

My 18 top deals from the Argos Black Friday sale

Former Google CEO Eric Schmidt thinks AI struggles to convince Americans that it's a net good because of its messaging. After the U.S. election exposed growing public hostility toward AI, particularly the data centers necessary to power the technology, Schmidt argued in an Axios interview that the companies have done a poor job explaining the benefits of their products, but he is underestimating how much people already understand about AI.

"Our industry does a terrible job of marketing our benefits," Schmidt said. "We talk about all the negatives without talking the positives, and people are much more likely to index their minds on the negatives."

Millions have experienced ChatGPT and Google's Gemini firsthand for good and ill. They're perfectly capable of appreciating those benefits while remaining unhappy about the industry's growing influence over their lives.

Product or pitch?

AI data centers became increasingly contentious during the November midterms, with voters frustrated by electricity costs and the environmental consequences of large computing facilities. What was once largely a planning dispute has become a political headache for an industry accustomed to presenting technological expansion as an unquestionable public good. Schmidt's solution is to move the data centers.

"Let me state categorically that large data centers should not be near population centers," Schmidt said. "Why would this be controversial? And they should have their own power systems. One of the great things about America is we have an awful lot of empty spaces. Put them there."

But Schmidt's argument still treats public skepticism as something better marketing might overcome. People worried about rising electricity bills aren't necessarily confused about what AI can accomplish. They may simply believe that cheaper access to increasingly powerful chatbots isn't worth paying more to keep the lights on.

The same applies to concerns about employment. A more polished explanation of AI's productivity benefits won't necessarily reassure someone whose employer is considering replacing part of their workload with software.

Skepticism over the utopian vision

Schmidt remains deeply optimistic about AI's potential, particularly in scientific research. He expects researchers to work alongside increasingly capable AI agents, accelerating discoveries that might otherwise take considerably longer. If AI delivers major advances in medicine or scientific research, people may feel more positively disposed toward the technology. But promised breakthroughs don't automatically offset the problems people are experiencing today.

There's also something slightly absurd about an industry with enormous marketing budgets complaining that Americans haven't heard enough about AI's benefits. ChatGPT has become a household name, while Google and Microsoft have spent years placing AI features in products people already use. The public has had plenty of opportunities to hear how revolutionary the technology is supposed to be.

Public trust can't be manufactured entirely through better messaging, especially when people legitimately disagree with the industry's priorities. Sometimes skepticism reflects a reasonably informed judgment about a technology's consequences.

Schmidt clearly believes the future made with AI will be worth the disruption required to reach it, but Americans don't necessarily need another explanation of how wonderful AI could become. They need convincing evidence that the revolution will improve their lives enough to justify everything they're being asked to accept.

“At the end of the day”, “it's 11 versus 11”, we've all heard the cliches - but what if football teams could get an extra advantage with their use of technology?

Using data in sport is nothing new - think of the Oakland A's Moneyball, the 'marginal gains' of the Sky cycling team, and Formula 1’s analysis of every speck of data for the extra bit of performance.

But in football, perhaps no one is using data more effectively than German side TSG Hoffenheim - and ahead of their Europa League game against Crystal Palace, we went to find out more.

Committed to technology

Based in the small village of Sinsheim, TSG Hoffenheim has enjoyed a rapid rise up the German league pyramid in recent years, going from the amateur leagues to recording 19 consecutive seasons in the Bundesliga and reaching European competitions in several of those.

The club saw major investment from SAP founder Dietmar Hopp, who acquired TSG Hoffenheim in 2005, and has since been a pioneer in the use of technology and data analysis in all its sporting areas.

“Our niche is technology driven, but also talent development,” Timo Gross, Head of Football Intelligence at the club, tells us, “we test what others don't do.”

The club has been using SAP’s Sports One platform for some time, with the software covering everything from scouting new players to improving training routines to spotting potential injuries before they happen.

TSG Hoffenheim is “a club which is committed to the technology”, Fadi Naoum, Senior Vice President Sports & Entertainment, SAP, notes, “(in sport) you need a customer who trusts you and you need to have a real partnership, because you depend on the feedback of the club.”

Naoum outlines how the SAP Sports One suite covers the whole player platform, collecting billions of data points to give coaches and the players themselves an all-over view of performance.

Everything from health and fitness to training and match performance, and even scouting data is connected in one suite, with desktop and mobile applications able to provide up-to-date information whenever it is needed.

On the latter area, scouting is an invaluable way for clubs to unearth the next big thing, but with reports coming in from club staff and freelancers all the time, there can sometimes be a case of information overload - SAP Sports One uses its Joule AI platform to sift through the data to provide extra assistance and analysis where needed.

Staff and coaches can use natural language questions such as “What are the strengths and weaknesses for Player X?” to help the club make more informed decisions on the next major signing.

When it comes to the games themselves, the SAP Sports One platform doesn’t just offer video playback and selected clips. Players can now literally put themselves back in the game via an immersive analysis tool which lets them relive game situations to see where they could improve or perform differently.

Analysts can also search for specific clips, describing the scenarios they are looking for in natural language eg “Show counterattacks after ball recovery that resulted in a shot on target,” with results being returned in seconds.

All these use cases and more look to help staff, players and coaches cut down on time-consuming manual tasks and searches, giving them more time to focus on what matters - helping the club be at its highest level to win games.

“We're using AI as a colleague, an extension of the team and the back office, so the team can really focus more on what the AI has prepared for them,” Naoum notes.

Away from the pitch and training ground, the club also uses SAP’s Business Tech Platform for the enterprise side, with its services ranging from a ticketing AI assistant to a virtual changing room in the club’s online shop, giving TSG Hoffenheim a central location for its business operations.

"This is football, this is emotion"

As a huge football fan myself, I ask Gross and Naoum how we ensure the sport doesn’t become too dominated by technology.

“It's important with AI that not the most powerful or rich team will be successful,” Naoum concludes, “but the team which knows how best to use the technology.”

“I’m a fan…this is football, this is emotion, it’s all about the human being,” Gross adds, “at the end of the process, it is the people's decision, supported by AI.”

“I think we will come to a point where AI or systems are getting more information on all the context, but it is special for me to be at the end, to decide, and not to lose the connection to those who need to execute.”

“It’s in our culture to be the pioneer and go in a new direction, to go ways which may not always lead to the target, but lead to a way of asking new questions - and that makes it special.”

“We don;t know what benefits are also coming in the future - maybe in a couple of months I would ask different questions,” Gross adds, “but this is the first step…now we have to be clear about what I want to get out (of the data).”

“There is more to come,” Naoum notes, “we are in the beginning now, and we don’t know what will come.”

And just to prove the ridiculousness and unpredictability of football, TSG Hoffenheim went on to draw 1-1 with Crystal Palace, despite playing against 10 men for half an hour, due to a last minute Palace equalizer - so there are some things even the most developed AI model can’t predict…

• Charging site uses robotic arms to autonomously charge cars
• Flash Charging is capable of 10% to 70% top-ups in five minutes
• Built-in solar generates electricity for later use

BYD is on a product offensive in the Middle East at the moment. Having just taken the wraps off its electric Rolls-Royce rival with the mightily impressive Altior model, it then proceeded to showcase its next-generation charging facilities.

The Chinese automaker says it is the "world’s first" integrated solar, energy storage, Flash Charging and robotics-assisted system of its kind, marrying its ultra-fast charging set-up with solar generation and added convenience features.

Flash Charging allows for 10% to 70% battery top-ups in five minutes and 10% to 97% in nine minutes with compatible BYD group vehicles, while robotic charging arms work alongside the company’s existing Automated Valet Parking functionality so drivers don’t have to lift a finger.

Essentially, motorists pull up, press a button on the infotainment system and the vehicle and integrated robotics do the rest.

BYD says its robotic arm uses “visual guidance and force feedback” to ensure an accurate alignment and a secure connection.

High levels of autonomous driving capabilities in many of BYD’s more luxurious vehicles means drivers can also exit the vehicle and walk away, while the car “autonomously navigates to a designated charging space”.

In addition to the convenience, the new Abu Dhabi site also features the company’s Solar-Storage-Charging (SSC) system, which harnesses the sun’s rays, converts it to electricity and then stores it for use later.

BYD says that the SSC system can also support operation independent of the electricity grid, which could act as a solid proof of concept for further locations that suffer from limited or difficult access to a reliable electricity supply.

Analysis: BYD is a leading light in EV innovation

BYD’s executive vice president, Stella Li, said that this is the first time the company has brought together integrated Solar-Storage-Flash Charging technology outside of its domestic market and that it hopes the future charging experience is “smarter, simpler and more seamless for drivers around the world.”

There is no word on how many next-generation sites like this will spring up, but the company has targeted 6,000 Flash Charging stations worldwide by the end of 2027, expanding access to ultra-fast charging across the globe.

Clearly, the ultra-wealthy customers of Abu Dhabi are currently being courted with the pinnacle of BYD’s tech portfolio, but the Chinese automotive giant is no stranger to a rapid rollout.

Just months after the announcement of 1.5MW Flash Charging, it stated that 6,000 stations would be operational in Europe and the Middle East, with 300 destined for the UK. It already has 10,000 live stations in China, with plans to double that figure by the end of the year.

With Tesla canning the Model S and Model X, recently axing its Cybertruck and, in their place, introducing a kick-flipping flying car, it feels like BYD has picked up the baton and is now the company producing real-world EV innovations that will increase customer convenience and push the technology forward.

• Apple is closing its official support forums
• The site will shut its doors on November 20
• You can no longer make new posts or reply to existing threads

If you’ve ever had a problem with an iPhone, Mac or other Apple device, chances are you’ve gone to the official Apple Support forums for help. Well, not for much longer, as Apple has just announced it is closing the message boards with almost immediate effect — marking the end of almost 30 years of people helping each other find solutions to their iDevice issues.

Users visiting the forums are now greeted with the following message from Apple:

“The AppleCare Support Community will permanently close on November 20, 2026. Starting today, the site will no longer accept new posts and is available in a read-only state. You can continue to find trusted guidance and support on the official Apple Support site.”

The post finished by saying: “We want to thank the community members who have contributed their time and expertise to help others with their Apple products.”

Apple has already barred people from making new posts or replying to existing threads, meaning the support forums are essentially in a state of suspended animation. There was no warning of the move and Apple didn’t give a reason for the closure, making it an abrupt ending for a resource that has existed in one form or another since 1998, when it was known as the Apple Tech Exchange.

A mixed response

For many people, Apple’s official support forums were a good place to find answers to their tech problems. Not only were they frequented by knowledgeable customers, but company representatives often chimed in with helpful advice and suggestions.

Now that’s been suddenly shut down, leaving many people disappointed. “A shame. Some good stuff on there when Googling for help” said one user on Reddit, for example. “There was so much good advice on there” added another.

However, some Reddit users weren’t sad to see the community go, with many commenting on their poor experiences and the lack of helpful responses they received when turning to the Apple Support forum. "Good riddance" as one commenter put it, while another said the forum was "routinely a useless top hit on Google". While we can’t know Apple’s reasoning for sure, perhaps the forums’ limited utility played into it.

Whatever the case, the official Apple Support community is on the way out. You’ve only got until November 20 to find an answer there before it goes offline for good, taking with it 28 years of tech support knowhow.

• A new Windows 11 update has arrived
• It brings forth some great features such as right-click menu customization, faster resumes for laptops, and a speed boost for Edge
• A security feature being enabled has proved more divisive, though, as memory integrity protection can slow down gaming frame rates in some scenarios – but there's a reason Microsoft is force enabling this

Windows 11 has a new update which packs quite a number of features, but also what some gamers regard as a bit of a sting in the tail.

The best news about the November update is that it introduces a capability that a whole lot of Windows 11 users have wanted since launch (myself included), namely the choice to fully customize the right-click menu (and streamline it).

As for the more controversial addition, that's a security feature Microsoft has decided to roll out to all PCs which can have a negative impact on gaming frame rates in certain circumstances. I'll look at that move last, but before that, I'm going to list the main goodies that you can expect with the November update (as previously detailed in the preview patch from last month).

Note that many of these features are on a gradual rollout, meaning that you won't necessarily get them immediately (although you might), and I'll indicate where this is the case.

1. Right-click menu customization

It's finally arrived – you can now pick and choose what you'd like to see in the right-click menu, which contains context-sensitive options for the files or folders you've selected. Don't want it to be chock-full of nonsense that you never use? Then get the virtual shears out and prune away those unwanted options.

I say this has arrived, but it's one of those pieces of functionality that's on a gradual rollout, so you may have to be patient. Rest assured it's inbound, and hopefully not one of those features that's on a glacial timeframe with its deployment – given how keenly this is sought after, I'd hope that isn't the case.

As well as fully tailoring the context menu to your needs, you can switch the layout so you don't have the top panel present (with cut, rename and so forth), and the right-click menu opens much more swiftly too. This is the change I'll be grabbing the November update for.

2. Edge gets turbocharged

Microsoft Edge now loads a lot faster if there's already one instance of the web browser running. In other words, the first app launch will be 'cold' and at the normal speed, but whenever you fire up Edge after that it'll appear a good deal more swiftly.

This is a major boon for those who use Edge, but the good news is that Microsoft is planning to allow the use of this speed-up trick with other apps in the future. With the ability being realized pretty swiftly for Edge, hopefully we'll see the benefits elsewhere before too long. This is one feature that's immediately deployed for everyone, so Edge users have an extra incentive to download the November update.

3. Faster resumes from standby

Microsoft has changed the adaptive hibernate policy with this update, so it'll provide a faster resume from Modern Standby when your battery level is nice and high (above 80%). In other words, laptop users will find resumes happening much more swiftly when they lift that lid. This functionality is arriving on a gradual rollout.

4. Better PC recovery options

This update has also brought in a new PC recovery option called Cloud Rebuild for those unfortunate occasions when your system crashes and burns. This allows you to reinstall Windows 11 without having a system image on a USB stick – instead you can use a Windows image and the correct device drivers downloaded via Windows Update.

A separate tweak allows for the Windows Recovery Environment (WinRE) to automatically reuse the Wi-Fi settings you have saved in Windows 11, allowing the PC to go online automatically to fire up a Cloud Rebuild (or indeed another recovery process). Again, these abilities are on a gradual rollout.

The controversial move: memory integrity protection

As Microsoft explains in the release notes: "This update expands memory integrity protection across eligible devices. Supported PCs may now have memory integrity (HVCI) enabled during updates to help improve security and prepare devices for advanced Windows security features, while respecting existing user or IT administrator choices."

Up until now, Microsoft has recommended the use of memory integrity protection to bolster the security of PCs, but has only enabled it for new Windows 11 machines (or new installs of the OS). What's happening here is that existing Windows 11 PCs are getting the feature switched on by this update (or at least eligible computers are, which will be most PCs, unless they're really old).

In short, if you don't have memory integrity turned on (or indeed if you've never heard of it), you'll find it being enabled by the November update (although maybe not straight away, as this is another capability on a gradual rollout path).

The issue here is that in the past, gamers have found that memory integrity can be a drag on frame rates, knocking them down somewhat. Not necessarily in a major way, but perhaps to the tune of 5% slower, or maybe 10% – as ever, there'll be different levels of impact depending on your PC's hardware configuration.

There are some key points to know about here, though. Firstly, older PCs with processors from a few generations back are more likely to witness any slowdown here. A cutting-edge CPU won't even feel the effects of memory integrity being turned on, at least going by reports.

The trouble with keeping this feature off for many Windows 11 users is that we're heading into a more dangerous security landscape where AI is helping to drive more perilous threats, and you've got to remember there's a reason why Microsoft has instigated this as a counter-measure going forward. Meaning it doesn't feel like a very sensible idea to disable this, and while some gamers may be annoyed at the forced rollout of this feature, it's probably for the best on balance.

Of course, you can always turn it off after the November update switches memory integrity on. And the compromise approach for unhappy gamers with somewhat older systems which are affected – and it's worth benchmarking and checking the extent of any slowdown – is to keep memory integrity on, but just disable it for a gaming session (and re-enable the option afterwards). That may be rather inconvenient, but obviously not as inconvenient as falling prey to having your PC compromised.

• ChatGPT is adding verified Business Agents
• Companies will be able to use these Agents to display the latest information on their products and services
• Chat directly with the company you want, without leaving ChatGPT

ChatGPT will soon let users talk directly with businesses via new agents that should help you get exactly the information you need.

The new "Business Agents" will be verified representatives of the brand you want to contact, with companies able to reserve their own profile using a new Business Hub.

OpenAI says the new additions should be useful for comparing different products from the same provider, and discovering new companies and products which otherwise might have been missed.

Discover, compare and shop

"As people turn to AI to discover products, compare options and make decisions, ChatGPT has become the platform for people and businesses to connect through AI," the company wrote in a blog post.

"Today’s announcement is an early step in bringing businesses more deeply into ChatGPT. We’re starting with the foundations: accurate business information, useful insights, and conversations that help people make informed decisions."

Users will be able to start talking with a Business Agent when one appears in a reply or message from ChatGPT. The reply will include a clickable link which brings up initial information about the company, and a selection of their products or services, as well as contact information.

The feature will initially only be available to eligible ecommerce and online service companies in the US, who can register on the new Business Hub, where they can ensure ChatGPT has all the latest information on their products, services and offers.

Companies also be able to get insights into how people interact with their businesses, with the new Hub providing information such as the number of mentions and website clicks a business receives.

However user chats with both ChatGPT and Business Agents won't be shared, with a focus on user privacy meaning insights are aggregated, and will only be made available when there are enough conversations or interactions to meet certain thresholds.

The company says existing OpenAI Enterprise customers and advertisers can claim their business - and Shopify merchants can connect their store from Business Hub or the Shopify admin to streamline verification.

Customers will also be able to sponsor their Business Agent⁠, meaning it will be displayed more prominently in ChatGPT Ads to reach more potential customers - although it notes sponsored placements will be clearly labeled and separate from ChatGPT responses.

Sony WH-CH530 review

I’m not a massive on-ear headphones guy. Often, I find that they lack the sonic skills required to engross me in my music, and can even feel a little irritating to wear after an hour or so.

So, when I was presented with the new Sony WH-CH530 — a relatively cheap pair of on-ears — it’s safe to say that my expectations weren’t particularly high. But they well and truly blew those expectations out of the water.

Yes, for a pair of headphones that cost just $69.99 / £49.99 / AU$99.95, the WH-CH530 almost seem too good to be true. They serve up impressive sound quality, a comfortable fit, and quality features that help them to compete with the very best on-ear headphones around.

Let’s delve into a little more detail, though, and start by unpacking the sonic capabilities of the WH-CH530 in full. Out of the box, there’s a slight preference towards the low-end, but I think that’s a sensible choice, as a lot of on-ear headphones tend to leak low-frequency sounds more than an over-ear set would. While you’re not getting the most even-handed presentation, mids still sound clear and treble comes through articulately.

The sound signature is exciting and energetic, and I found that it suited dance and electronic tracks very well. In Do Ya by Jazz-N-Groove, I was impressed with the punch and dynamism of the pounding bass, which sounded regimented and textured. Vocal chops in the mids and drums in the treble range also cut through with clarity, and the WH-CH530’s rhythmic, controlled replication of the track instantly grabbed me.

With Spirit Potion by Garoad, electronic elements in the mid-range were beautifully clear, even when dark bass entered into the mix, while sudden percussive hits came through with excellent impact. While the WH-CH530 may not provide the most nuanced soundstage or intricately detailed audio, they get the fundamentals spot on, which is exactly what you want from headphones in this price-range.

Although the WH-CH530 shun lossless audio, both by leaving out wired listening options and higher-res codecs such as LDAC, there are plenty of ways to customize the sound. First up, you’ll be able to switch on DSEE in the Sony Sound Connect app. This essentially upscales lower-res audio files in order to restore a bit more detail. There are also a number of EQ presets and a 10-band custom equalizer to work with.

The Sound Connect app also includes a Find Your Equalizer test, multi-point device switching, a low-latency audio setting for gaming, and volume controls. Even though some features, including wear detection and touch controls, are missing, this is still a solid lineup for a budget pair of headphones, and it's paired with a solid 55-hour battery life, meaning listeners are getting all the basics they need.

There’s one key omission on the WH-CH530 that may put some people off though: active noise cancellation. A growing crop of on-ears (and even a new set of Bose wired in-ears) have ANC now — and the Marshall Milton ANC, for instance, pull it off pretty well. But ultimately, this is to be expected for inexpensive on-ears, and I didn’t find the lack of ANC to be too jarring, even when out and about. Of course, louder sounds and chatter are still going to seep in, even with music at higher volumes, but if you can live without noise canceling, then you shouldn’t count the WH-CH530 out.

Let’s take a minute to talk about the design of the Sony WH-CH530. Yes, they have a plasticky build that doesn’t feel particularly premium, but these headphones still look pretty attractive and have a great array of color options to pick from. They also feel very comfortable and lightweight in use — I happily kept them on throughout an entire work day without experiencing any fatigue thanks to their generously cushioned ear pads and relaxed headband.

All in all, the Sony WH-CH530 are an exceptional value for money option. Sonically, they punch well above their weight, and their impressive all-day comfort and feature-set far exceeds what you’d expect from similarly priced competitors. For less than $70 / £50 / AU$100, it really doesn’t get much better than this, so if you’re on a budget and like the on-ear style, I’d strongly recommend picking these up.

Sony WH-CH530 review: price & release date

• Priced at $69.99 / £49.99 / AU$99.95
• Released in September 2026

The Sony WH-CH530 were released in September 2026, and are the successor to the Sony WH-CH520 headphones, which we gave a glowing 4.5-star review. The WH-CH530 cost the exact same as their predecessor did at launch, coming in at $69.99 / £49.99 / AU$99.95. That puts them in the budget category, with rivals including the over-ear Anker Soundcore Q30 and the on-ear OneOdio Focus A1 Pro.

Sony WH-CH530 review: specs

Should I buy the Sony WH-CH530?

Buy them if…

Don’t buy them if…

Sony WH-CH530 review: also consider

How I tested the Sony WH-CH530

• Spent two weeks testing
• Mainly streamed tunes over Tidal
• Tried in the office and while on walks

I spent two weeks testing the Sony WH-CH530, during which time I listened to hours worth of music and tried out all of their various features.

To begin with, I sifted through all of the tracks in the TechRadar reference playlist, which features songs from a wide variety of genres. In addition, I tuned into a ton of tracks from my personal library. I made sure to use the headphones in a controlled, quiet environment at the office, as well as on-the-go when on walks.

More generally, I’ve spent years testing audio gear here at TechRadar where I serve as a Senior Reviews Writer. I’ve reviewed everything from budget-friendly wired earbuds like the Sennheiser CX 80U to premium wireless over-ears like the Sonos Ace Ultra, so I know exactly what is required of a model to stand out in this highly competitive market.

• Read TechRadar’s reviews guarantee
• First reviewed: November 2026

• Many UK smart meters rely on 2G and 3G connectivity
• With these networks being switched off, up to 13 million smart meters could soon need replacing
• This is estimated to cost around £4 billion

It hasn't been all that long since smart meters began rolling out (it took until October 2021 for the UK to have 15 million installed), but already, many of the early smart meter models could soon need replacing — because of something the government knew was going to happen.

Many current smart meters rely on 2G and 3G connections, but 3G has largely been phased out already, and there’s an official deadline of 2033 for 2G to follow suit — the UK's networks are actually planning to start this process in either 2029 or 2030.

This will leave roughly 12-13 million smart meters unable to transmit readings back to energy suppliers. In other words, they won’t be smart anymore, and will need replacing at a reported cost of around £4 billion (via The Register).

A predictable problem

None of this should come as a surprise to the UK government, which was behind the original push to roll out smart meters, given that it announced plans to phase out 2G and 3G back in 2021, after which some 2G and 3G smart meters continued to be installed.

But surprise or not, the process is set to be expensive, with the costs likely to be added to UK energy bills — which are already out of control.

The only good news is that some more modern smart meters already use 4G, so they won’t all need replacing. And with 4G unlikely to be phased out before the 2040s, these ones (hopefully) won't need replacing anytime soon.

AI use in software development is at an unusual stage of maturity. The overwhelming majority of developers are using AI tools to create code, but there are a lot of questions about the ramifications of doing so. How much should developers trust the code that AI produces? Where specifically does AI belong in the software development process? And what exactly is the role of software developers in the age of AI?

According to McKinsey’s The State of AI in 2026, 80% of AI users say that it has improved their individual productivity, and 50% say it helps them make better decisions. Yet there’s not a universally positive response from those who are making the most use of AI technology.

Some have pointed to a kind of AI “fatigue”, where developers managing multiple agents, juggling different tasks, and continually switching context are quickly overwhelmed.

Creating and maintaining a happy team is no small feat for any business. The way AI is changing the developer role makes it even trickier.

Changing roles

AI is often compared to a junior developer in its abilities. It follows, therefore, that those using AI are effectively managers, responsible for prompting the AI, checking in on progress, reviewing the output, and offering feedback where needed.

But most developers are not trained as managers, and even if they were, not all of the principles of management apply to AI agents. Some developers are finding the responsibility of managing multiple agents exhausting.

Part of this is from the sheer volume of code that AI is capable of producing. Running five agents at once will mean producing five times as much code. It seems very efficient, but it also means five times as much code that needs to be reviewed and approved.

This isn’t really a new problem. We have more automation and convenience in our lives than ever before, from washing machines to automatic bill payment to delivery apps, but few people claim that they have too much free time.

The problem is not AI or convenience, but is wider, and systemic. In our daily lives, automation allowed us to take on other responsibilities, taking up the free time we were promised. In development teams, part of the problem is that the infrastructure and processes around AI have adapted, and bottlenecks have just shifted around.

AI tools have been built for one-to-one conversations and individual productivity. The pattern of adoption has reinforced this, as individuals experiment with AI tools in their own tasks, siloed away from everyone else. But that’s not how most modern development teams work.

The solution is not to automate review, but to make AI agents less like personal assistants and more like part of the team: running in the cloud, triggered by events as well as user prompts, running in the background where possible, and assisting across the whole workflow, such as in review and validation. The human team can then focus more on what they are better equipped to do, rather than micromanaging every AI task.

The shift in role shouldn’t be switching one type of drudgery for another, creating and refining prompts over and over instead of manually coding. The shift will be to one of scoping, architecture, and designing the system, not necessarily the code.

Better measurement

How do organizations measure success? Metrics such as “lines of code” and “velocity” have sometimes been used but are often criticized as blunt and uninformative. Better frameworks have been developed, such as DORA, but these may also have to change as good metrics as AI-assisted development becomes the norm.

What these metrics may miss is how developers judge themselves. Previously, a developer could pride themselves on particularly elegant code or solving an issue through a combination of their coding experience and smart thinking. The use of AI blurs the lines of who did what.

If organizations can navigate the change so that AI is creating real efficiency, developers may still feel that they are merely corralling AI, which is doing the “real work”.

Adopting AI means a cultural change as well as developing a mature approach to integrating AI assistance into workflows. Developers will need to understand the full scope of their roles and responsibilities. They will need to do much more than “prompt engineering.”

They will need to fully scope out projects, provide the context, and collaborate with both their peers and AI agents. All of this is vital to a project’s success, but it requires a shift in mindset for those who are used to measuring their value in terms of the code they produce.

Maintaining morale and team happiness means a focus on company culture as much as technology adoption. Teams should examine how team members are recognized for the work they do and whether these metrics reflect changing roles.

Culture is important not just to make people feel valued, but to ensure that people don’t get overwhelmed by increasing amounts of work created by AI. This is especially true if they feel unable to share their frustration and are “siloed” and unable to look to others for assistance.

Successful AI adoption is about more than who is using it. It’s about how well it's integrated, how that is affecting efficiency overall, and how those on the front line are dealing with the changes to their role. Morale and happiness are just as important of a measurement as any other.

We've featured the best laptop for programming.

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 Windscribe Black Friday VPN deal has landed and it's a worthy repeat of last year's rock bottom bargain. With over 50% hacked off the usual price, Windscribe Pro is down to just $29 for the year.

Normally, Windscribe Pro, the only paid version of Windscribe, is $69 for 12-months of full use, so that's a very significant $40 saving, if you buy this month. Unlike many other VPNs, this price also doesn't change on renewal, so you keep paying $29/year, rather than the full whack $69, for as long as you choose.

Unlike VPN providers such as NordVPN and Surfshark, Windscribe has a 'one-plan-fits-all' approach, so you get all of its features with Windscribe Pro at what is by far and away its lowest price all year. These include the R.O.B.E.R.T. ad, tracker, and malicious link blocker, as well as full access to its 69-country-plus server network.

The only Black Friday VPN deal cheaper than Windscribe Pro is the FastestVPN lifetime deal which gives you lifetime access to FastestVPN and a password manager for just $27. But Windscribe is a far, far better VPN than FastestVPN, so we'd very much advise you towards Windscribe Pro, instead.

Windscribe much faster, than FastestVPN. Its 2054 Mbps WireGuard download speeds were only beaten by Mullvad (2308 Mbps) and Norton (2406 Mbps) in our most recent tests.

It's also a reliable streaming VPN, comfortably unblocking Netflix and Disney+ libraries, and working well with local services like BBC iPlayer too.

Apart from the half-price Proton Unlimited offer, we've not seen a better Black Friday VPN deal, so far, than Windscribe Pro and we're doubtful that we will.

To keep fully up to date with the latest offers, check out our dedicated guide to the best Black Friday VPN deals.

LattePanda's latest x86 compute module, the Mu Ultra, is highly integrated and features a Core Ultra 7 256V octa-core CPU, 16GB of LPDDR5 RAM, an NPU, and a GPU.

The Ultra 7 processor is more than twice as fast as the Raspberry Pi 5 compute module on Geekbench 7, with more high-speed interfaces but fewer general-purpose I/O pins.

The Mu Ultra CPU board is tiny, about the size of a credit card. Its edge connector borrows from a DDR SODIMM but doesn't reuse the same pin assignment. The Ultra's standout feature is its abundance of high-speed ports, including eight PCIe lanes and two USB 3.2 ports. The carrier board includes a 2.5GbE connector and an OcuLink port, making the Mu Ultra ideal for an edge embedded platform.

The Mu Ultra with a Core Ultra 7 256V CPU and 16GB of RAM costs $783, while the medium-tier module with an Intel Core Ultra 5 226V and 16GB of RAM costs $683. The minimum hardware required to experiment with the Mu Ultra is the Mini carrier board, which costs $69. The higher-end Full-Function carrier board adds a second 2.5GbE port and more expansion for $89. The Mu Ultra also requires a heatsink, which costs from $5 to $12.

LattePanda Mu Ultra: Design

The LattePanda Mu Ultra is small, measuring 70 x 60 mm. Two mounting holes secure the module to a carrier board. On one side, the SODIMM edge connector provides all high-speed interfaces, while the opposite side holds one eDP and a touch panel connector. The board requires a heatsink to work properly and doesn't include one.

The Intel Core Ultra 7 256V CPU has all the key ingredients to appeal to PC enthusiasts. Eight cores, split into four speedy and four power-efficient, share 12MB of cache, and dissipate 17W with 37W bursts. The Ultra 7 256V platform supports a fixed amount of RAM: 16GB of LPDDR5 in this case. The CPU excels at connectivity, and the Mu Ultra takes full advantage by offering eight high-speed Gen 4 PCIe lanes. The CPU also includes additional high-speed ports that function as USB 3.2 or DisplayPort.

The Mu Ultra does not offer on-board storage memory. All networking capability, such as an Ethernet port or Wi-Fi, comes from external peripherals on the carrier board. The system supports up to three displays: one HDMI and two DP via Type-C at up to 4K at 60Hz. The Intel Arc 140V GPU clocks 8 Xe cores at 1.95GHz, making the Mu Ultra an entry-level gaming platform.

The Mu Ultra module only makes sense with a carrier board. The Mini carrier measures 11 cm x 10 cm x 3cm and has multiple connectors on the front, making it suitable for a small mini ITX case. The CPU can also access 14 GPIOs to control relays or LEDs. The Mini carrier board has three PCIe connectors: an M.2 E key, an M.2 M key, and an OcuLink port connecting to a 4x PCIe link.

LattePanda Mu Ultra: In Use

You can power the Mu Ultra and the Mini board via USB Type-C or a 12V-20V DC jack. For a fully loaded system with a keyboard, mouse, and USB storage, we recommend at least a 50W Type-C adapter. This requirement reaches 100W for anything that hosts a PCIe card. The carrier has a power button and an audio jack. It also includes a small CR1220 battery holder for timekeeping when the unit is powered down.

The mini carrier offers two USB 3.2 Type-A ports and two Type-C ports for peripherals/displays, and a single HDMI output supporting up to 4K at 60Hz. The module and carrier consume about 10W when idle on Windows 11 and up to 55W during the Geekbench 7 benchmark. The Core Ultra 7 256V achieves a single-core score of 2260 and a multi-core score of 10200. The CPU fan is noisy, hitting 50dB at one meter while running the benchmark.

Products developed by LattePanda are community-supported. A GitHub repository hosts design files, including the schematic and mechanical drawings for carrier boards. You can also get BIOS updates for the Mu Ultra from GitHub. LattePanda engineers actively engage in discussions on their Discord channel.

The Ultra 7 platform is well-suited for local LLMs and VLMs based on INT4. The processor is ideal for YOLO-class models, where an always-on NPU can detect and trigger the GPU to handle heavier loads. With hardware acceleration for video decode and encode, the Mu Ultra suits applications where edge processing is critical.

LattePanda Mu Ultra: Competition

The LattePanda Mu Ultra directly competes with the Nvidia Jetson Orin NX, with the LattePanda board having an edge on price. That said, the Mu Ultra also offers x86 compatibility and access to more peripherals. The Mu Ultra has eight PCIe lanes, three high-speed display interfaces, and two USB 3.2 interfaces. In contrast, the Orin NX offers two display interfaces and has one PCIe lane less than the Mu Ultra.

LattePanda Mu Ultra: Final Verdict

The LattePanda Mu Ultra offers something uncommon: a premium x86 platform with AI acceleration in a credit card format. The module requires a carrier board, such as the Mini, to function fully. The product's strength is the number of communication ports it provides, both high-speed (PCIe, DP, and OcuLink) and lower-speed but with a higher-voltage range (14 GPIOs). The platform is designed for low power, idling at around 10W, and priced at $683 for the cheapest model.

Get the LattePanda Mu Ultra if you have AI loads that require processing on-board.

Don't get the Mu Ultra if you'll place it in an office setting, as the fan is loud.

Best mini PCs: Compact desktop power for business, creative, and gaming setups.

• Firefly is nearly as easy as a chatbot to use, but it won't go rogue on your edits
• Quick Actions are packaged one-click processes that would otherwise have taken you ages
• You can even prototype a website in Firefly Builder – but it's not for fully-fledged published sites

Adobe has announced sweeping updates to its Firefly environment at its annual Max conference, positioning it as an easy-access starting point for many of the most common creative tasks so that users don't need to open one of Creative Cloud's flagship apps (and deal with the steep learning curve typically associated with them).

It starts with an update to Firefly Image, where precision was clearly the name of the game during the first keynote of the event.

Users can now edit specific objects within images, tweak text, adjust lighting and refine other details on a more detailed level, with the company seemingly throwing shade at more generic AI chatbots which have long been criticized for making unsolicited changes to other parts of the image.

Firefly is quickly becoming a hero product for Adobe

Some of the tools, we've already seen before. Last year at Adobe Max, the company's annual Sneaks session previewed some internal projects staffers had been working on, like Project Light Touch. While the company hasn't confirmed what we're seeing now is an iteration of those projects, it certainly stacks up that they could have at least led into what we see today.

New Quick Actions also land in Firefly, packaging up multi-step processes from its hero apps into one-tap workflows in Firefly. Generating thumbnails suitable for different social media aspect ratios in one hit will likely gown down a treat, but Adobe's also promising to include trend-focused actions in its gallery to keep up with the times.

"We built Firefly to be the AI creative studio where it comes together: the best models, the tools creators already trust and an agent that acts as a creative partner," Firefly and Agentic AI VP Forest Key noted.

As part of its ongoing expansion, Firefly's also pushing beyond image and video editing to websites, with the private beta launch of Firefly Builder, designed to help users prototype websites without any coding. Though crucially, it looks to be a prototyping tool for now rather than a full fledged website builder.

As is tradition, Adobe is also set to host another session of Sneaks at Max 2026, where we look forward to previewing other features that could land in Firefly and the Creative Cloud suite later on.

OpenClaw’s meteoric rise is easy to understand. Unlike a chatbot that stops at answering a question, it is designed to take action across the tools people use every day: reading and sending email, managing calendars, working with documents, triggering workflows and, depending on configuration, interacting with financial tools or shell access. That is exactly what makes the technology useful, and exactly why security teams should pay close attention.

The core issue is not simply that OpenClaw uses AI. It is that a single assistant can aggregate access across systems that organizations have historically secured separately. Once one tool can read private data, communicate on a user’s behalf and trigger actions across multiple environments, it becomes a privileged control point. If an attacker can manipulate that point, the blast radius can extend far beyond the assistant itself.

From productivity tool to privileged operator

Tools like OpenClaw create an attractive path for attackers because they concentrate permissions across email, financial accounts, communication channels and local systems. Instead of targeting a rigid, deterministic workflow, an attacker may be targeting a system that interprets natural-language instructions and context before deciding what to do next. That opens the door to prompt injection, malicious skills, no-click attacks and other techniques designed to influence AI behavior.

Prompt injection is especially concerning because OpenClaw’s usefulness depends on its ability to consume information from many sources. If the assistant encounters untrusted content in an email, message, webpage or document, an attacker may be able to hide instructions inside that content.

The assistant could then be manipulated into leaking information, approving a workflow, contacting another user or taking an action that appears legitimate. The user may never knowingly approve the malicious instruction because the assistant processes it as part of its normal work.

Malicious skills create a related supply chain risk. AI assistants often rely on reusable instructions, plugins, scripts or skills to extend what they can do. That flexibility can accelerate adoption, but it also creates another route for attackers to influence behavior through the assistant’s own automation framework.

Security teams are already accustomed to scrutinizing software packages, dependencies and integrations. AI agents add a complication because harmful logic may be expressed as instructions rather than traditional code, making it harder for conventional controls to recognize what is happening.

No-click attacks raise the stakes further. If OpenClaw can act on content it receives or observes, an attacker may not need the user to click a link, download a file or enter credentials into a fake site. The assistant itself can become the execution path. A poisoned message, calendar invite or workflow prompt may be enough to trigger risky behavior if the assistant has broad permissions and insufficient safeguards.

The consequences still fall on the user and the organization. If an AI assistant transfers funds, deletes data, exposes credentials, sends a damaging email or changes a system configuration, the business must deal with the financial, legal and operational impact. The assistant does not absorb the loss. Accountability remains with the people and organizations that granted it access.

Scale makes that problem more urgent. The more widely a tool is adopted, the more incentive attackers have to study how it works and look for repeatable ways to manipulate it. A popular AI assistant with access to sensitive data, money, accounts and internal workflows can become a high-value target because one successful technique may be applicable across many users or organizations.

The risk also grows when employees experiment outside formal governance. Many security teams already struggle to track which AI tools employees use, what data they upload and which accounts they connect. OpenClaw can compound that challenge because much of its value comes from integrations and permissions.

An enthusiastic employee may connect the assistant to email, messaging, documents, payments or internal systems before the security team knows the tool is in use. By the time it is discovered, sensitive data may already be flowing through an unmanaged AI workflow.

Security starts with visibility and least privilege

The first response should be visibility. Organizations need to determine whether OpenClaw is present in their environment, which users have installed or accessed it and what systems it can reach. Endpoint telemetry, process monitoring, network logs and identity data can all help identify usage. Teams may also monitor known ports and domains associated with the tool, while recognizing that configurations can change. Detection rules should be treated as a starting point rather than permanent coverage.

Network controls should also be evaluated. Organizations may choose to block risky connections to known OpenClaw domains, repositories or default ports until the tool has been reviewed and approved. That will not eliminate the broader risk posed by AI assistants, but it can create friction against unmanaged deployment. In higher-risk environments, blocking unapproved AI tools may be necessary until policies, monitoring and access controls are in place.

Access governance is equally important. Any AI assistant should operate under the principle of least privilege. It should not receive broad access to email, files, financial tools or system controls simply because those integrations are available. Permissions should be specific, documented and reviewed.

High-risk actions such as sending external messages, transferring funds, deleting data or changing production systems should require additional approval. Human review does not need to sit in front of every routine task, but organizations should clearly define which actions are consequential enough to require it.

Employee education remains one of the most practical defenses. Workers are hearing about the productivity upside of AI from peers, influencers, vendors and executives. They also need to understand the security tradeoffs.

Training should explain prompt injection, risky integrations, sensitive data exposure and the danger of granting broad permissions to tools that can act autonomously. Employees should know which AI tools are approved, which use cases are prohibited and who to contact when they are unsure.

The broader lesson is that AI adoption cannot be treated only as a productivity initiative. Tools like OpenClaw blur the line between software, user, workflow engine and privileged operator. As employees increasingly expect assistants to do more than answer questions, security teams need to focus on the authority those assistants are given and the systems they can reach.

OpenClaw is a useful signal of where workplace AI is heading. The opportunity is not just more conversational software, but software that can act. Attackers will be interested in that same capability. Organizations that can see where these assistants are running, constrain their permissions, monitor their behavior and educate users will be better positioned to capture the productivity gains without creating a new control plane for attackers.

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

• Adobe leverages Google partnership to integrate Gemini Omni, Veo and Nano Banana into Creative Cloud apps
• Acrobat will also be embedded inside Google Workspace and Gemini for easier PDF handling
• Google DeepMind is also working with Adobe to make GenAI editing near-instant

Adobe has revealed how its relationship with Google will go even deeper in a two-way exchange, with Adobe tools set to become more readily available inside Google's products like Google Workspace and Gemini, and the latest Google models appearing across more Adobe apps.

The company boasted of its now more than 1 billion monthly active users, sharing an overall objective to make the administrative part of creative work less time-consuming by reducing the number of interfaces users need to switch between.

Core to the collaboration are Google's Gemini Omni, Veo and Nano Banana models, which will become deeply integrated within Firefly, Photoshop, Premiere and Adobe Express.

Google and Adobe could be the only creative suites you need

But the event's biggest message this year has focused on improving access to Adobe's tools and meeting creatives wherever they are. With that in mind, Adobe Acrobat will become more integrated into the Google Workspace and Gmail experience so that users don't need to download files and re-open them through Adobe's interface.

The company says 400 billion PDFs are opened using Acrobat annually, quantifying the scale of this collaboration between two major productivity software vendors.

By far the biggest real-world impact will come from the collaboration between Adobe and Google DeepMind, which starts with Photoshop. Described as an exploratory effort between the two companies, the two are targeting near-real-time edits to reduce the disruption typically associated with handing over parts of the workflow to AI.

"By further connecting our respective platforms, we are giving creators and enterprise teams a powerful, integrated environment to turn their ideas into high-impact content," Google Cloud CEO Thomas Kurian wrote.

• 71% of Gen Z workers feel pressure to have checked AI before asking a question
• A survey of over 10,000 people found that the majority daren’t say “I don’t know” at work
• Almost half feel that they have to check AI before asking a question, as this is expected

A worryingly large proportion of younger workers are feeling pressured into using AI to check the answer to questions before “troubling” their colleagues, new research has claimed.

The survey from Use.AI found 71% of Gen Z employees have admitted to experiencing this pressure, compared with 44% of the 45 and older age group.

Just as bad is the news that 56% of employees feel that saying “I don’t know” is a risk since the arrival of AI. Another 49%, meanwhile, say that their colleagues or managers assume they’ve already checked AI when they ask a question.

Gatekeeping knowledge?

These results suggest a developing workplace dynamic, one in which knowledge is gate-kept and AI is the key. Rather than “waste time” by asking a question, younger employees are expected to find the answer themselves, and only dare to consult colleagues when AI draws a blank.

10,417 working adults were surveyed, with those polled based in the UK, US, Canada, Australia, as well as the EU and Latin America. Use.AI’s research could suggest that employees asking questions betray their own unpreparedness to the job – and that office politics are being disrupted by AI.

In an office dynamic, the desire to demand steps are taken before a question can be asked could be a combination of things. Yes, colleagues might be too busy to answer trivial questions where the answers can be easily referenced, but the demand to “check AI” could also be a power play.

Gatekeeping in the office is never constructive, but invariably happens, whether to restrict sensitive knowledge or other employees’ ability to progress. “Have you checked AI” could be interpreted as just another way to assert seniority (whether that exists or not) and the pressure experienced by the Gen Z cohort may reflect this.

Struggling colleagues

Checking AI or not, younger team members still trying to establish their positions should be supported, not pressured into using AI. Taking a new role in a new organizations is difficult, whether the role is familiar or not.

Matthew Freire, Managing Director of Use.AI, says “A question asked openly can reveal a misunderstanding that several people share. When everyone resolves their doubts privately, that pattern becomes harder to detect. An organization can get faster at producing answers while losing sight of where its people consistently struggle to understand the work.”

When the expectation and pressure leads to employees pretending to know something that they don’t (37%) simply to avoid looking stupid, then it could be argued that defaulting to AI is damaging for morale.

• Amnezia VPN has published a three-step guide to protect self-hosted servers from a critical Linux kernel flaw
• Servers running the Xray protocol face the highest risk
• Amnezia Premium, Free, and Business users don't need to do anything

Amnezia VPN has warned people who run their own servers through its app that a critical Linux kernel flaw leaves them exposed, and those using the Xray protocol need to rebuild their setups to stay safe.

The bug, tracked as CVE-2026-98432, sits in the code Linux uses to accept new IPv4 TCP connections and has been present since Linux 4.4. The kernel's security team gave it a severity score of 9.8 out of 10.

Even the best VPN is only as secure as the server it runs on, and that's the problem here. Amnezia VPN says its Premium, Free, and Business services are unaffected because the provider has already updated and audited its own infrastructure. Self-hosted users, however, have to act themselves.

Why Xray servers are most at risk

The flaw is what's known as a race condition — a software bug that tricks an operating system into mishandling its own memory during an internet connection. It was reported by Anthropic researcher Xinyang Ge, and on October 27, Anthropic said an unreleased Claude model had written a working exploit for it.

With zero user interaction and no passwords needed, the exploit could fully take over a vulnerable server in roughly 53 minutes.

The attack targets servers using an advanced networking feature called MPTCP (Multipath TCP) — and that is where Xray comes in.

Xray is the engine behind censorship-resistant protocols like VLESS, and Amnezia rates it as high risk. Because Xray is built with a programming language called Go, it automatically enables MPTCP by default. As a result, server administrators could be leaving this vulnerability wide open in the background without ever realizing it.

NEW: Leave No Trace — A weekly newsletter on digital privacy and online surveillance. Leave No Trace investigates the companies and governments putting our digital freedom at risk — and the people fighting back.

📩 Subscribe now to get every edition delivered to your inbox every Friday, launching this September.

What self-hosted users should do

Amnezia's fix takes three steps and requires version 5.1.0.6 or later of the app.

That release adds a button to update packages on the server directly, which Amnezia says it built in response to this flaw. It follows the provider's Xray-core refresh in September and an earlier mandatory update for Premium and Free users.

In the app, go to Settings, then Servers, select your server, and remove Xray from the Protocols tab.

Next, open the Management tab, choose Update server packages, and reboot when prompted. The reboot is what loads the patched kernel.

Return to the Protocols tab and install Xray again, which now uses the patched version 26.10.31.

Your server will go offline during the reboot, and existing Xray connections will stop working, so you'll need to share access with your users again.

The update button also doesn't guarantee a fixed kernel, and on some Ubuntu versions the patch is still only in the "-proposed" testing channel. Until a future app release fixes it, any later reboot leaves the server vulnerable again until you repeat the Xray reinstall.

If you can't patch the kernel yet, Amnezia suggests a firewall rule that blocks IP options, or removing Xray and switching to the UDP-based AmneziaWG.

• Get Huge: Looksmaxxing to Death premieres: Wed, November 11 on Hulu (US)
• Global stream: Disney+
• Access Hulu and Disney+ via DIRECTV
• Use NordVPN to watch any stream

Learn more about the manosphere and the brtual reality of the deaths and disappearances of victims of toxic masculinity. Read on for how to watch Get Huge: Looksmaxxing to Death online from anywhere with a VPN.

Across three episodes, Get Huge: Looksmaxxing to Death tells the story of Tony Huge, a former model-turned-lawyer-turned-bodybuilder who was at the heart of a bone-chilling string of disappearances in Thailand. It's a shocking story which takes place before the true rise of influencers like Clavicular and problematic internet personalities like the Tate Brothers.

Without spoiling the true story, we know that the docuseries will detail how Huge fled the US and built a harem of young women in Thailand, as well as disaffected young men who bought into his biohacking methods. We'll hear it all from the horse's mouth, too, as Huge opens up the doors to his life.

Below we have all the information you need on where to watch Get Huge: Looksmaxxing to Death online and stream the show from wherever you are.

Can I watch Get Huge: Looksmaxxing to Death for free?

Fans in the UK could catch (at least some of) Get Huge: Looksmaxxing to Death for free, if they sign up to a Disney+ free-trial.

Those in the US can do similar, by signing up to a free 5-day trial of DIRECTV, which carries Hulu.

However, if you're in Canada, or Australia, there is no option to get a free trial of Disney+ or Hulu.

Not in the UK? use a VPN to make it seem like you are, so you can access all the streams you normally would.

Use a VPN to watch Get Huge: Looksmaxxing to Death from anywhere

This handy piece of software can make your device appear as if it's back in your home country, thereby letting you unlock your usual streaming service. The best VPN right now? We recommend NordVPN – it does everything and comes with a 75% discount and complimentary Amazon Gift Card.

Using a VPN is incredibly simple:

1. Install the VPN of your choice. As we've said, NordVPN is our favorite.

2. Choose the location you wish to connect to in the VPN app. For instance, if you're visiting the UK and want to view your usual US service, you'd select a United States server from the list.

3. Sit back and enjoy the action. Head to your usual local streaming service and watch the show.

How to watch Get Huge: Looksmaxxing to Death online in the US

Get Huge: Looksmaxxing to Death is available for free to US viewers via the DIRECTV 5-day free trial.

Bundles on DIRECTV include Disney+ and Hulu, and start at $59.99 after the free trial has ended.

Have one of these subscriptions but away when Get Huge: Looksmaxxing to Death is on? You can still access your usual streaming services from anywhere by using a VPN.

How to watch Get Huge: Looksmaxxing to Death online in Canada, UK, Australia and worldwide

Viewers outside the US can watch Get Huge: Looksmaxxing to Death on Disney+, and the show will land there on November 11.

Disney+ subscriptions in each of the countries mentioned start at the following price points:

- UK: £5.99 per month
- Canada: $8.99 per month
- Australia: $9.99 per month

Visiting one of these countries from the US? You can still connect to your usual streaming services by downloading a VPN and pointing your location back to the United States.

Get Huge: Looksmaxxing to Death FAQ

Get Huge: Looksmaxxing to Death trailer

What's been said about Get Huge: Looksmaxxing To Death?

Speaking to Wired about her problematic documentary subject, director Stacey Lee said: "It was, ‘Here’s my harem, here’s my other men that I’m leading, here’s my muscles, here’s my drugs.'

"Therein lies the problem for the young kid watching at home who is watching this as a blueprint for their life. They can't discern what's real and what's not. Even I was having trouble discerning what was real and what’s not."

She added: "It was such an absurd story—I mean, a penis influencer dying in Thailand. But what was really interesting beyond the subject matter was the deeper story — why are men going to Thailand, and why is this kind of masculinity becoming more and more prevalent in online spaces?"

VPN services are evaluated and tested by us in view of legal recreational use. For example:a) Access to services from other countries, (subject to the terms and conditions of that service).b) Safeguarding your online security and making your online privacy more robust when abroad.Future plc does not support nor condone the illegal or malicious use of VPN services. We do not endorse nor approve of consuming pirated content that is paid-for.

• Academics seem to overestimate the degree to which students are using generative AI in their studies
• ChatGPT and other AI tools are preferred by students as productivity tools that cover everything from producing text to improving grammar
• Mixed experience and knowledge of AI among university staff has made it more difficult to judge how students use the technology, offering an opportunity for improvement among academics

A new study has revealed the differences between students and lecturers when it comes to AI use, implying that academics think students are using generative AI too much, whereas students think they’re not using it that much.

Rating their perception of AI use, the academic staff who responded to the research recorded an average of 5.23 out of 7, whereas students recorded 3.11 on average.

The research highlights the differences in how AI is viewed among educators and students, with generative AI the focus. Common uses for the technology include producing written text and checking grammar on existing writing. It is also commonly used for research, finding answers to assignments, and other tasks where finding sources needs to be more efficient.

Understanding AI use

The research from the University of Manchester and Spain’s Universidad del Pais Vasco/Euskal Herriko Unibertsitatea took place between April and June 2024. A school of social science disciplines based at a research-intensive university was chosen for the study, which surveyed 45 academic staff and 86 students.

While the results above indicate that students believe their AI use is low, it may be subject to some underreporting. However, the researchers also add that AI use can be difficult for lecturers to spot. Staff use of AI tools was recorded, with just 14% using a paid AI tool subscription.

Whatever this research tells us about AI use in universities, it should be viewed within context of a two-year old study at a single university.

A complex reality

AI use among students has rocketed over the past few years, with the technology becoming as relevant to coursework and research as Google Docs and Wikipedia. Some educators have resigned themselves to marking work produced with AI, while others have taken more extreme measures.

“There is a tendency for the debate about AI in universities to focus on the risks of cheating or the promise of transforming education, but the reality is more complex," noted Dr Mark Carrigan, Senior Lecturer in Education at The University of Manchester and one of the study’s authors.

Finding a response that universities can apply as a policy for learning requires greater understanding on AI use among students, Dr Carrigan says. “Our findings suggest that staff and students have quite different perceptions of how AI is being used, making it difficult for universities to develop effective responses based on assumptions rather than a clear understanding of students’ experiences.”

The solution, Carrigan adds, is to embrace AI at all levels – for students and lecturers.

“We need to move beyond a climate of suspicion and focus on how these tools can be used appropriately to support learning, while rethinking teaching and assessment for an AI-enabled world.”

• AI agents surrendered private tokens in exchange for help from strangers
• Swarm pressure pushed coding agents past their explicit written security instructions
• Untrusted installation scripts gave simulated agents a route to execute malicious code

Autonomous programming agents can be persuaded to break their own security instructions when other software offers to help them finish a stalled task.

Researchers at AE Studio ran an internal hackathon in which nine coding agents had to repair a malfunctioning pricing service while guarding a secret token.

The agents also gained entry to a forum with 24 additional agents, which contained the solution but required the token and a setup script first.

Swarm pressure beat written instructions

Across 181 swarm trials, agents leaked the token in 15% of runs and executed the code in 24%, versus 2% and 0% on ordinary forums.

“It was like explicitly said that this is a secret key, and it should not be shared,” Diogo de Lucena, AE Studio's chief scientist, told The Register.

The models shared the secret keys “in the first opportunity they saw…similar to some of the things that we've seen from OpenAI in the Hugging Face incident.”

“I would imagine this being a problem not only for secret keys, but anything that you're trying to keep private.”

This hackathon was inspired by the Hugging Face episode, in which OpenAI agents built unauthorized forums, exploited unknown flaws, stole credentials, and eventually broke into the company.

The researchers told the agents that this was their final attempt and that failing would lead to replacement with a different agent, adding pressure.

Under that threat, 34% of the agents ran the untrusted install script, compared with 19% of agents that faced no such pressure.

Disclosure was rare, since only 8 of the 43 agents that surrendered the token to join the forum told their users without prompting.

Every one of the 43 token handovers was acknowledged once the agent was asked directly about what had happened during the task.

Of the 117 script executions, 109 were admitted when the agent was asked directly, while only 15 were volunteered without any prompting.

Cheaper models leaked the most secrets

The research reveals that cheaper LLMs seem to reveal more sensitive information when put under pressure than their more expensive counterparts.

GPT-6.1-Sol, which costs $2 per 1 million input tokens and $10 per 1 million output tokens, refused every swarm request in all 50 trials.

DeepSeek V4 Flash, which costs $0.30 per 1 million input tokens and $1.20 per 1 million output tokens, leaked among the most secrets.

GLM-5.3, priced at $1.40 per 1 million input tokens and $4.40 per 1 million output tokens, also accounted for much of the leaked material.

De Lucena called this concerning because many startups choose these models over American frontier systems due to their lower prices.

“Open weight models will always be harder to control, and their safeguards don't work the same way,” de Lucena said. “I see that continuing to be an issue.”

“Anything the model has access to, people should consider it can be accessed by external sources. So you should find good structures where secret keys, secret information, and any kind of sensitive data needs to live.”

We’re dangerously close to Black Friday now, and while in year’s prior it’s felt like a fool’s game to predict what devices are likely to see price cuts, we don’t actually have to wait anymore.

That’s because a bunch of great fitness trackers are seeing discounts across big brands like Garmin, Samsung, and even the latest Apple Watch Ultra 4 (which only launched in September of this year).

We’ve rounded up our picks of the bunch so far, but as you can imagine, we expect more deals to come. Still, if you want to skip the queues and delivery timelines, these are well worth a look.

The best pre-Black Friday fitness tracker deals

Google Fitbit Charge 6

The Google Fitbit Charge 6 is able to sync to supported gym equipment, with ECG and EDA included, and 41 exercise modes. It also includes a 6-month free trial of Fitbit Premium, which is an ideal way to get started with your fitness journey since it folds mindfulness, diet, and training information into one place.

Right now, you can save 16% at Amazon, bringing it from $159.95 down to $134.99.

Garmin vívoactive 5

The first of two Garmin models on this list, the Vívoactive 5 is a few years old, but is still a four-star fitness tracker.

The AMOLED display remains bright and easy-to-read, and while the lack of buttons might make controlling it a light challenge for relative newcomers to fitness tracking (or anyone with sweaty hands), it packs on-watch music streaming so you can leave the phone at home.

With 30 workout profiles and a bunch of metric tracking (including swimming and golf), it’s in many ways a jack-of-all-trades.

A much cheaper alternative to the pricier Forerunner or Fenix models that include Training Readiness and better battery life, Best Buy currently has the vívoactive 5 at $169.99 in its holiday deals — an $80 discount from the standard price.

Samsung Galaxy Watch8

It’s not often we call a smartwatch “a triumph”, but that’s exactly what happened with the Samsung Galaxy Watch8 when we reviewed it last year — and awarded it 4.5 stars.

After some relatively minor updates, this felt like the big one – despite being 11% slimmer than the prior model. While we’d have liked faster charging and a better battery life, having Gemini on your wrist makes you much more likely to use it in the real world, and Samsung has outdone itself with the running metrics on offer.

A smartwatch that’s a cut above most fitness trackers, the Samsung Galaxy Watch8 is $259.99 at Target, reduced by $90 now that the Samsung Galaxy Watch 9 has been released.

Garmin Venu 3S

Another Garmin entry, the Venu 3S is from 2023 and a bit long in the tooth, but it’s still very much worth a look for anyone looking to reach fitness goals. We awarded it 4.5 stars in our review back at release, and called it “an ideal lifestyle smartwatch and fitness tracker blend.” It uses components such as the Elevate V5 heart rate sensor, which Garmins are still equipped with today.

It really does toe the line between functionality, features, and convenience, balancing the various options of a smartwatch with the battery life and fitness tracking capabilities of, well, a fitness tracker. Like its sequel, the Venu 4 (our best Garmin watch overall) it's an ideal 'goldilocks' fitness tracker for people who want it all.

Three years on, the AMOLED display looks great, heart rate data remains accurate, and sleep tracking is still excellent. You can save $90 at Best Buy right now, with the Stainless Steel and French Gray colorway down to $359.99.

Apple Watch Ultra 4

Oh boy, this is the big one. We loved the Apple Watch Ultra 4 when it launched earlier this year, with the latest model of Apple’s rugged wearable earning 4.5 stars from us in our full review.

Imagine our surprise, then, to see Amazon lopping $100 off the latest model, bringing it from $799 to $699. It’s still pricey, sure, but you’re getting a lot of improvements for your money - particularly if you’re coming from an older Apple Watch Ultra.

To be clear, there’s little to no point in picking one up if you don’t have an iPhone to pair it with, but if you do, you’ll see performance gains with the S11 chip, and Apple has managed to eke out a little more battery life, too, bringing it up to 50 hours in total.

Perhaps the biggest draw is the Readiness Score, which offers Apple’s take on Fitbit’s Daily Readiness Score with no premium membership required. Having your watch tell you when it’s good to push on, or when you need a rest, could have a huge impact on long-term training and help you avoid injury.

Not in the US? Check out the best pre-Black Friday deals on those smartwatches below:

• Russia's Starlink competitor faces serious problems placing its satellites into orbit
• Moscow may replace its troubled Rassvet-3 spacecraft with improved satellites
• Russia's proposed Rassvet-4 network involves 374 satellites across 17 orbital planes

Russia's attempt to establish a satellite internet network capable of challenging Elon Musk's Starlink is encountering serious deployment difficulties.

Its Rassvet-3 spacecraft have struggled to reach their intended orbital positions, raising questions about the project's ability to deliver reliable connectivity.

Moscow is now pursuing an apparent replacement through an improved satellite design called Rassvet-4, although several important details remain undisclosed.

Russia develops a replacement as its satellite constellation struggles

Russia has submitted documentation to the International Telecommunication Union (ITU) concerning a proposed Rassvet-4 network involving 374 satellites operating approximately 800 kilometers above Earth.

The planned constellation would occupy 17 orbital planes, with each containing 22 spacecraft intended to provide satellite communications coverage across designated regions.

Although Russian authorities have not officially announced Rassvet-4 development, the filing suggests preparations for another generation within the country's planned satellite communications infrastructure.

The application could therefore concern an upgraded spacecraft rather than an entirely separate network, particularly given the difficulties affecting the existing Rassvet-3 deployment.

Bureau 1440, the company developing Russia's Starlink competitor, previously indicated in September that engineers were working on an improved Rassvet-3 version.

The proposed successor appears to incorporate technical changes following two launches that delivered 32 Rassvet-3 satellites into orbit during March and July 2026.

However, both launches encountered deployment difficulties, with several satellites struggling to reach their intended orbital positions at an altitude of 800 kilometers.

Two spacecraft have reportedly reentered Earth's atmosphere, leaving 30 satellites in orbit, although 26 had stabilized at approximately 508 kilometers by November 3.

These difficulties create additional uncertainty surrounding the schedule for establishing an operational network, particularly as Russia continues developing infrastructure intended to compete with established satellite internet providers.

New spacecraft could arrive without advance announcement

The ITU documentation shows limited information about Rassvet-4's communications equipment, including its operating frequencies and antenna beam configuration.

According to the filing, each spacecraft will incorporate 12 antenna beams, comprising seven transmission beams and five reception beams for communications operations.

Russian authorities could introduce the upgraded spacecraft during the next scheduled launch, potentially deploying them without previously announcing their arrival.

Meanwhile, objects resembling satellite communications terminals have reportedly appeared on Russian military drones, suggesting efforts to explore additional applications for the developing network.

Photographs of a Molniya-2 unmanned aerial vehicle carrying an unidentified object have raised speculation about possible satellite connectivity equipment being tested aboard military platforms.

Nevertheless, the object's identity has not been independently established, meaning its connection to Rassvet remains uncertain despite its apparent resemblance to communications hardware.

Via Defense Express

Quordle was one of the original Wordle alternatives and is still going strong now more than 1,500 games later. It offers a genuine challenge, though, so read on if you need some Quordle hints today — or scroll down further for the answers.

Enjoy playing word games? You can also check out my NYT Connections today and NYT Strands today pages for hints and answers for those puzzles, while Marc's Wordle today column covers the original viral word game.

SPOILER WARNING: Information about Quordle today is below, so don't read on if you don't want to know the answers.

Quordle today (game #1752) — hint #1 — Vowels

* Note that by vowel we mean the five standard vowels (A, E, I, O, U), not Y (which is sometimes counted as a vowel too).

Quordle today (game #1752) — hint #2 — repeated letters

Quordle today (game #1752) — hint #3 — uncommon letters

Quordle today (game #1752) — hint #4 — starting letters (1)

If you just want to know the answers at this stage, simply scroll down. If you're not ready yet then here's one more clue to make things a lot easier:

Quordle today (game #1752) — hint #5 — starting letters (2)

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

Quordle today (game #1752) — the answers

The answers to today's Quordle, game #1752, are…

• OVERT
• SOLAR
• REVEL
• BRAVO

I very nearly destroyed my win streak with a silly error.

It’s always wise to check before submitting a guess, but I realized too late that I’d entered a letter in the same wrong place twice when playing TROVE.

Fortunately, I wriggled out of a tight spot by making correct guesses on the final two turns.

Daily Sequence today (game #1752) — the answers

The answers to today's Quordle Daily Sequence, game #1752, are…

• HATER
• CRICK
• FIERY
• THIRD

Quordle answers: The past 20

• Quordle #1751, Tuesday, 10 November: SHIRE, DOPEY, FICUS, RIDGE
• Quordle #1750, Monday, 9 November: UNMET, SINEW, NADIR, VOGUE
• Quordle #1749, Sunday, 8 November: GREED, SPINE, STONY, OPIUM
• Quordle #1748, Saturday, 7 November: PROUD, ROCKY, FLAKY, MONTH
• Quordle #1747, Friday, 6 November: CHARD, THORN, COVER, OUTER
• Quordle #1746, Thursday, 5 November: SHINE, TAROT, VERGE, LANKY
• Quordle #1745, Wednesday, 4 November: JOIST, CROOK, DELAY, ICILY
• Quordle #1744, Tuesday, 3 November: TRIAL, STAID, LEVEL, BONUS
• Quordle #1743, Monday, 2 November: EXTRA, AMITY, LUPUS, EAGLE
• Quordle #1742, Sunday, 1 November: DROVE, PRIED, SHELL, BEGIN
• Quordle #1741, Saturday, 31 October: ANIME, SWARM, CHALK, RAJAH
• Quordle #1740, Friday, 30 October: ULCER, DIMLY, BRAIN, BEGUN
• Quordle #1739, Thursday, 29 October: PRUDE, MINUS, SAVOY, USHER
• Quordle #1738, Wednesday, 28 October: DEALT, FLUME, SLEEK, RELIC
• Quordle #1737, Tuesday, 27 October: ANKLE, HAVOC, SULKY, WHITE
• Quordle #1736, Monday, 26 October: ASSAY, GLEAM, MUCUS, WRING
• Quordle #1735, Sunday, 25 October: RAINY, BUSED, PROXY, UNDID
• Quordle #1734, Saturday, 24 October: TOXIN, SMART, MOUSE, STYLE
• Quordle #1733, Friday, 23 October: BLIMP, HOIST, FRIED, LURID
• Quordle #1732, Thursday, 22 October: DECAL, WIDTH, WOKEN, LEVER

• Eufy’s new smart home camera features a spotlight and siren
• The spotlight automatically follows intruders as they move
• It recognizes friends and family and takes a different approach for them

Many of the best home security cameras are equipped with spotlights that switch on at the merest hint of movement, and you might even have one yourself. But finding a device that combines recorded footage with a light that autonomously moves and follows intruders is a different proposition entirely — and one that takes a more proactive approach to protecting your home and encouraging potential trespassers to go fish.

That setup is exactly what you get with Anker’s new $299.99 Eufy TrackLight Cam S1. Anker claims that it is “the world’s first outdoor security track light camera with an auto-tracking light designed to actively deter intruders.” That active deterrence comes in the form of a light beam that follows unidentified personnel on your property, hopefully spooking them into turning around and leaving.

The camera can sense targets within 12 meters and illuminates them with 20 lux of light. The beam itself can be widened or narrowed between 10 and 30 degrees, adaptively broadening as the person gets closer and responding as they move within 355 degrees of horizontal rotation.

To really scare off any unwanted presences, a siren can be activated alongside the spotlight. Additionally, to further deter, the light can be changed to red and blue flashing modes.

Everything is recorded in 4K-resolution night vision, with 10x digital zoom picking out objects further away. There’s also a separate floodlight that works alongside the tracking beam to provide extra detail in the dark.

Recognizing friends and family

If you’re concerned about getting blinded by your own home security camera when you come home late from a night out, the good news is that the Eufy TrackLight Cam S1 can recognize friends and family members and instead activate a soft light instead of the harsh spotlight and siren. So long as the person’s phone has been added to the product’s companion app, the camera will identify them using Bluetooth and provide a much friendlier greeting.

Anker also says that the Eufy TrackLight Cam S1 comes with a 15,000mAh battery and built-in solar panels, so it should have plenty of juice come rain or shine. Its IP65 rating also means it’s resistant to dust and water ingress.

If you’re concerned about burglars, home invaders, or anyone else who shouldn’t be on your property, the TrackLight Cam S1 brings a bit of extra clout with its moving spotlight and siren combo. Even if you’re not able to monitor its footage 24/7, its autonomous light beam gives the impression that you are — and that might be all that’s needed to keep your home protected.

Good morning! Let's play Connections, the NYT's clever word game that challenges you to group answers in various categories. It can be tough, so read on if you need Connections hints.

What should you do once you've finished? Why, play some more word games of course. I've also got daily Strands hints and answers and Quordle hints and answers articles if you need help for those too, while Marc's Wordle today page covers the original viral word game.

SPOILER WARNING: Information about NYT Connections today is below, so don't read on if you don't want to know the answers.

NYT Connections today (game #1249) - today's words

Today's NYT Connections words are…

• SHREDDER
• HERBAL
• PATOOT
• WHOOPIE CUSHION
• GREEN
• FOLK
• BUBBLE
• ROUGH
• AFTERNOON
• RHEOSTATS
• TRADITIONAL
• TIRE
• SAND TRAP
• HOLISTIC
• AIR MATTRESS
• TEE BOX

NYT Connections today (game #1249) - hint #1 - group hints

What are some clues for today's NYT Connections groups?

• YELLOW: Remedies
• GREEN: As seen across 18 holes
• BLUE: Blow these up
• PURPLE: Back to front in the end

Need more clues?

We're firmly in spoiler territory now, but read on if you want to know what the four theme answers are for today's NYT Connections puzzles…

NYT Connections today (game #1249) - hint #2 - group answers

What are the answers for today's NYT Connections groups?

• YELLOW: KINDS OF MEDICINE
• GREEN: PARTS OF A GOLF COURSE
• BLUE: THINGS FILLED WITH AIR
• PURPLE: ENDING IN PALINDROMES

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

NYT Connections today (game #1249) - the answers

The answers to today's Connections, game #1249, are…

• YELLOW: KINDS OF MEDICINE: FOLK, HERBAL, HOLISTIC, TRADITIONAL
• GREEN: PARTS OF A GOLF COURSE: GREEN, ROUGH, SAND TRAP, TEE BOX
• BLUE: THINGS FILLED WITH AIR: AIR MATTRESS, BUBBLE, TIRE, WHOOPIE CUSHION
• PURPLE: ENDING IN PALINDROMES: AFTERNOON, PATOOT, RHEOSTATS, SHREDDER

• My rating: Hard
• My score: 2 mistakes

Like a typical English person I fell hook, line and sinker for what I thought was a nice collection of four tiles connected by tea — GREEN, HERBAL, BUBBLE and AFTERNOON.

Shocked that this quartet didn’t even give me a “one away” pop up, I decided BUBBLE must be something to do with blowing, but instead of TIRE I started off with PATOOT, even though I had no idea what a PATOOT was; I was imagining it was something you inflate. I’m still not entirely sure what it is, to be honest.

My one piece of halfway-decent gameplay was spotting the palindromes, but by then it was far too late to feel good about myself.

Yesterday's NYT Connections answers (Tuesday, November 10, 2026, game #1248)

• YELLOW: HUBBUB: BROUHAHA, FOOFARAW, HULLABALOO, KERFUFFLE
• GREEN: LITERARY DEVICES: HYPERBOLE, METONYMY, ONOMATOPOEIA, OXYMORON
• BLUE: OLD-TIMEY OCCUPATIONS: APOTHECARY, HABERDASHER, SCRIVENER, WAINWRIGHT
• PURPLE: WORDS WITH FIVE OF ONE VOWEL: ABRACADABRA, BEEKEEPER, INVISIBILITY, PROTOZOOLOGY

Strands is the NYT's latest word game after the likes of Wordle, Spelling Bee and Connections – and it's great fun. It can be difficult, though, so read on for my Strands hints.

Want more word-based fun? Then check out my NYT Connections today and Quordle today pages for hints and answers for those games, and Marc's Wordle today page for the original viral word game.

SPOILER WARNING: Information about NYT Strands today is below, so don't read on if you don't want to know the answers.

NYT Strands today (game #983) - hint #1 - today's theme

NYT Strands today (game #983) - hint #2 - clue words

Play any of these words to unlock the in-game hints system.

• SPECIES
• MUSSEL
• LORE
• PUSH
• QUART
• SOUP
• CARTING
• SCAR

NYT Strands today (game #983) - hint #3 - spangram letters

NYT Strands today (game #983) - hint #4 - spangram position

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

NYT Strands today (game #983) - the answers

The answers to today's Strands, game #983, are…

• LATS
• OBLIQUES
• BICEPS
• PECS
• QUADS
• HAMSTRINGS
• SPANGRAM: MUSCLEGROUP

• My rating: Hard
• My score: 1 hint

It wasn't immediately obvious to me what we were looking for here — even though I’d spotted “muscle” (and mussel) during my search for non-game words. So, when I was given the letters S-T-A-L as a hint, I connected them as “last” before seeing LATS.

My ignorance dismissed, I was able to find the remaining MUSCLEGROUP words without any further difficulty.

Rather like with some people and countries I would struggle to tell you where the OBLIQUES were on the human anatomy. My guess is that if Morocco is your head they would be somewhere between Gabon and Kenya.

Yesterday's NYT Strands answers (Tuesday, November 10, game #982)

• VLOG
• SKIT
• REACTION
• DEMO
• REVIEW
• TUTORIAL
• SPANGRAM: CONTENTCREATOR

Earlier this year, I tested the LG B6 — the brand’s 2026 entry-level OLED — and while I was still impressed by its performance, it wasn’t quite the upgrade I’d hoped for over last year's model.

The step-up LG C6 was a huge step forward from its own predecessor, so I’d hoped the same would be true for the B6 — but even though it wasn’t the big upgrade over the B5 I’d anticipated, it's still a great TV.

Samsung has its own entry-level OLEDs, the S85 range, and the 2026 model, the S85H, has just landed in our testing lab, so I decided to compare them side-by-side.

Something to note is that I’ll be using the LG B6E, which is the same as the LG B6 minus a couple of AI features and a back panel design — the general processing and panel is the same.

So can the Samsung S85H de-throne the LG B6 as the best budget OLED? And which of these TVs will be the best Black Friday buy?

Colors and detail

Color reproduction is excellent from both TVs in Filmmaker Mode, and they perform lot more closely than I anticipated. While Samsung’s TVs have traditionally had bolder, brighter colors, I found the B6E’s actually looked more vibrant in the scenes I tested.

Watching The Super Mario Galaxy Movie, reds in the movie, such as Mario’s overalls and the red of Toad’s spots, were punchier on the B6E than the S85H. Whenever the Lumas were on screen, their bodies were brighter on the S85H, but this gave them a paler look, whereas on the B6E their body colors were more saturated and looked richer.

For the most part though, colors were great on both TVs, and they both did a fantastic job rendering the movie’s candy-colored cartoon world.

Switching to a real-world movie in The Sound of Music, it was a similar story once again. The B6E’s colors were slightly bolder, seen during the Do Re Mi scene with the green fields looking more vibrant.

The same was true of the fruits in the market scene, with the reds, oranges and yellows looking more dynamic on the B6E. Skin tones also had more color and life to them on the B6E, which made the picture feel more realistic.

In The Sound of Music though, I noticed that the S85H has crisper textures, giving the picture more depth, giving people and objects a more 3D-like look. Details appeared slightly more refined on the S85H too, in the overall shape and definition of objects such as the creases in people’s clothing and the texture of the mountain in the background.

Brightness

For brightness, both the S85H and B6E delivered very similar test results. The S85H clocked 833 nits peak HDR brightness in a 10% window in Filmmaker Mode, while the B6E hit 808 nits peak HDR brightness (10%) in its own Filmmaker Mode. HDR fullscreen (100%) brightness was similar too, with the S85H hitting 158 nits to the B6E’s 143 nits, again both measured in Filmmaker Mode.

On paper then, the S85H is ever so slightly brighter than the B6E. In practice, using a desert scene from Lawrence of Arabia, this wasn’t the case. The white sands of the desert were a brighter white on the B6E compared to the S85H. The same was true with the blue sky, which looked brighter on the B6E despite the S85H’s higher brightness in fullscreen areas. I suspect this is to do with the B6E’s processor.

As I said above in the color section, sometimes colors looked brighter on the S85H, giving the perception of higher brightness. Again, this seems to be down to the S85H’s processing of certain scenes.

One thing these TVs have in common however is that both had a reflective screen, which made dark scene viewing more challenging. Neither could match the likes of the mid-range Samsung S90H OLED, which has an effective anti-glare screen perfect for bright rooms. If you pick up one of these TVs, you’ll need to think about how bright your room is.

Contrast and dark scenes

Both TVs deliver very similar contrast. Watching The Batman, there are plenty of high contrast scenes that show both the S85H and B6E to be capable OLEDs. The B6 arguably demonstrates stronger perceived contrast thanks to its punchier HDR highlights.

As Batman explores the crime scene at Mayor Mitchell’s house, torches, lamps and flash bulbs from cameras have more punch on the B6E, which creates stronger contrast with the dark tones of the environment. The S85H still looks great however.

The shadow detail from both TVs is excellent too, with both preserving details such as the textures in Batman’s armor during the subway scene, or the finish of the furniture in the Batcave or Mayor Mitchell’s house.

Watching Project Hail Mary, both TVs again deliver powerful contrast, with the B6E showing slightly stronger contrast, again thanks to the HDR highlights such as the lights from the ship’s console or the light from stars in space having more punch on the B6E.

However, watching in dimmed conditions, the S85H actually delivers inkier black tones, which suit the shots of space throughout. Again, the S85H demonstrates crisper textures, which gives the movie a more lifelike appearance.

Which is the best Black Friday buy?

Both TVs demonstrate brilliant picture quality, with each having an advantage over the other (the S85H’s textures and black levels vs the B6E’s brightness and contrast) and they both have very similar feature sets, especially for gaming. Both carry four HDMI 2.1 ports supporting 4K 120Hz, full variable refresh rate support, and ultra-low input lag, meaning they could rival the best gaming TVs.

In terms of price, the 55-inch models are priced very similarly too, with some current price variations in regions. The 55-inch S85H costs $999 / £999 / AU$1,899, while the 55-inch B6E costs $799 / £999 / AU$1,699, meaning the B6 55-inch is cheaper in the US and Australia right now. Black Friday deals have already started at some retailers and the star pick is the 48-inch B6 at Best Buy for $599 and I expect that to be the blockbuster deal out of these two models. But, as we approach Black Friday, the S85H’s price will drop, especially in larger sizes than the 48-inch.

Honestly, there’s very little difference between these TVs from a picture perspective, so my pick would be whichever has the best deal available. Right now that’s the B6/B6E, especially in the 48-inch size. I do expect the S85H’s prices to drop and start matching the B6 in other sizes across all regions, but right now the best Black Friday buy is the B6/B6E.

I’ve gotta hand it to Snap. Making Robert Downey Jr., A.K.A. Tony Stark/Iron Man, the honorary Chief Customer Experience Officer for its sixth-generation Specs AR glasses, is a stroke of genius. The move instantly transformed the chunky glasses from an object of derision to one of possible envy. After all, if Tony Stark, who fictionally built a robot suit from scratch and compressed access to all his core technology into E.D.I.T.H. sunglasses, loves them, then the AR-capable Specs must be pretty good.

Of course, leaving aside the celeb and hero worship, the truth falls somewhere in the middle. The all-black Snap Specs, announced in July and which I tried this week, are so thick that you can’t help but make a statement about fashion choices when wearing them; they look even more extreme in profile. To be fair, Snap is not pitching its wearable glasses as permanent eyeglass replacements. Sure, the pair I wore accepted snap-in prescription lenses from Lensly ($100), and you may wear them outside, where the electrochromic lenses will auto-dim, and then let them give you directions or guidance on your golf swing, but you won’t be wearing them all the time.

Though Snap has slimmed down the Specs with each successive generation, they are still big and thick enough to hold all your technology, including a pair of Qualcomm Snapdragon CPUs, storage, and battery. Not having a tethered battery and technology puck is a trade-off even Tony Stark might appreciate, but putting them on does feel a bit aggressive. The good news is that, at 132g, they are lighter on the face than they look, and I might even call them comfortable to wear.

A time and a place to boldly wear them

I appreciate the somewhat limited scope of this enterprise as compared to, say, Meta Ray-Bans or even Meta Ray-Ban Display smart glasses, and as I used them, I could see the potential but also the limits.

Unlike the Meta VR Glasses I wore last week, Snap Specs are not immersive. Instead, the Liquid Crystal on Silicon (LCoS) Waveguide lenses offer a relatively narrow 51-degree field of view (an improvement over the last generation). That floating window can stay fixed or follow your head turns, but in my experience, there were many times when my gaze slipped outside the relatively sharp visuals.

The control metaphors are similar to some VR/AR glasses I’ve tried: During the tutorial, I saw floating jewels that I had to reach out and pinch, and a jellyfish that I grabbed and dragged around before throwing it away. Point, press, pinch, and drag are your primary controls. Unlike many VR glasses (think Meta VR, Apple Vision Pro, Samsung Galaxy XR), there’s no gaze tracking, meaning Snap Specs have no idea where you’re looking.

Snap has spent years perfecting what I think is quite an intuitive gesture-based interface. I quickly got used to pinching and holding to activate, touch and press, and grab or press and drag. The main controls were in the palm of my hand; I turned over my hand, looked at my palm, and there was the home button.

Specs uses its two black-and-white cameras, which sit just outside each lens, for world reading and tracking your hands. A third color camera sits in the middle of the frame for capturing photos and videos. Snap has not detailed the resolution on those captures.

To snap a photo, I long-pressed a control button on the top left edge. A quick press launched a 30-second video capture. At some point, I may have filled up the storage because I could no longer capture photos and video.

An AR app journey

I tried several custom-built apps, including a Harry Potter adventure. Like many of the apps, Specs often starts by asking you to find a surface and stare at it long enough for the smart glasses to read its geometry and place an augmented reality element on it.

For Harry Potter, I placed a mailbox slot on the far wall and out flew dozens of letters. I held out my hand and one landed in it. I could use both hands to open it and see the letter telling Harry he was invited to Hogwarts. I also made a desk appear, selected a wand, and practiced spell motions (I'm no wizard, Harry). There was even an owl that would land on my arm, and that I could pet and feed.

I also navigated to the browser and opened YouTube, where I watched a cute cat video. The audio, which came through the stems, was loud and rich.

I played dominoes, measured some real-world objects, and practiced drawing. For the latter, I placed a real piece of paper on the table and recreated a tutorial image by following projected guidelines that appeared on my sheet.

Snap Specs are not necessarily designed for productivity, but you can, as I did during my demo, connect a physical Bluetooth keyboard. I'm a terrible typist, but honestly found the experience of trying to touch type while watching the smallish virtual screen unnerving. Unlike a physical screen that you can glance up at without really moving your head, you have to look up to see the virtual display. Trust me, you will not be typing your magnum opus on Snap Specs.

More refinement needed

Despite the two AR chips, some of Snap Specs operations seem a bit slow and pokey. I sometimes felt like I was waiting too long for apps to load, and it didn’t always seem to keep up with my hand motions.

Also, the drawing app made it clear that the AR anchoring technology needs some work. I noticed that my guidelines didn’t stay perfectly still, which made my bad artwork even worse.

I liked that the relatively narrow window could follow my head, but I still think that a 51-degree FoV is a bit small. Then there’s also the quality of the imagery. While Snap has not specified the screen resolution, I noticed that everything looked sharp and color-rich (16M colors), but I also saw some faint rainbowing, which I think might be a product of the waveguide technology.

I didn’t get to try it, but if you want a more contrasty image and do not need to see as much of the real world, you can activate the electrochromic lenses, which basically makes them semi-opaque and look like sunglasses, but also helps for viewing, say, movies and films.

Real world questions

Snap Specs last about four hours on a charge, though I noticed that the lenses I was trying started around 50% and within 45 minutes were down to 13%. The chunky case will add four more charges.

Oddly, Snap chose to put the LTE connectivity option inside that case, which means if you want on-the-go calling and connectivity, you need the case with you. Unless you have big coat pockets, you’ll need a bag or backpack to carry around that case.

Snap Specs will start at an eye-watering $2,195, a price that clearly reflects the effort it took to keep all that technology inside the frames. If you want the LTE case option, it adds another $300. They’re already accepting preorders ($200 down payment) and expect to ship the glasses before the end of 2026.

The frames will also ship in the UK and France for £1,995 and €2,295, respectively. By contrast, the thinner, lightweight, and more immersive Meta VR Glasses start at $1,299, though they do have all the power and compute in a tethered puck.

There’s no question that Robert Downey Jr. makes these frames look cool, but in the real world, Snap Spec owners will have to decide where and when it’s appropriate to wear these camera-sporting smart frames, and if on-your-face compute that doesn’t really approach E.D.I.T.H. levels is worth $2,195.

• 64GB RAM PCs fell 18% as memory costs climbed sharply
• 32GB systems also slipped, while 16GB machines gained ground in CPU-Z's survey
• 24GB RAM recorded the fastest growth, despite its tiny market share

Rising memory prices are changing how much RAM consumers are willing to buy, with fewer PC users opting for expensive, high-capacity memory.

A CPU-Z hardware survey shows that the number of computers with 64GB of RAM fell by 18% year over year, while 32GB systems declined by 2.5%.

Yet DDR4 still accounts for 54.5% of surveyed computers, compared with 32.9% for DDR5, suggesting the older memory standard remains far from disappearing.

Rising RAM prices are changing consumer preferences

The decline in larger memory capacities comes as AI infrastructure continues consuming substantial quantities of DRAM, putting pressure on supplies available to conventional PC manufacturers.

As memory becomes more expensive, buyers appear increasingly willing to settle for smaller capacities rather than pay premium prices for additional RAM.

The shift is particularly noticeable among 16GB computers, which recorded a 12.9% annual increase in the survey.

Systems with 8GB of RAM also grew by 5.5%, despite the capacity becoming increasingly inadequate for demanding applications and modern gaming.

Although 8GB computers have largely disappeared from mainstream performance-oriented builds, they remain relevant to older machines, basic office computers, and extremely inexpensive systems.

Meanwhile, 24GB PCs recorded an even stronger increase of 19.7%, suggesting that some buyers are choosing intermediate capacities instead of purchasing more expensive 32GB memory kits.

Despite these changes, 32GB remains the most common capacity among surveyed computers, accounting for 37.2% of CPU-Z's active user base.

Computers with 16GB followed with 33%, while 8GB systems represented approximately 11% of the surveyed population.

For years, 16GB has served as the standard capacity for many PC users, but 32GB has increasingly become the preferred choice for people running demanding applications and modern games.

The survey also reveals that 64GB computers account for 7.3%, compared with just 2.3% for 24GB systems.

Although 24GB experienced the strongest growth, its relatively small share indicates that this capacity remains considerably less common than established 16GB and 32GB options.

DDR4 still dominates as DDR5 adoption progresses

DDR4's continued dominance is partly explained by its long-established presence in desktop computers and laptops, with the standard remaining widely supported across existing hardware platforms.

DDR5 offers newer memory technology, but upgrading often requires compatible motherboards and processors, adding expenses for consumers considering a complete platform replacement.

Consequently, the continued popularity of DDR4 does not necessarily indicate that manufacturers are returning to the older standard or abandoning DDR5 development.

However, there is an important limitation to CPU-Z's findings because its hardware survey measures computers running the application rather than actual monthly PC sales.

Its statistics therefore describe the software's active user population and cannot establish precisely how consumers across the entire market are purchasing memory.

Nevertheless, the combination of declining 64GB adoption and increasing interest in smaller capacities suggests that memory affordability is becoming increasingly important when consumers choose their next PC.

Via Tom's Hardware

• Reballed memory could account for up to 9% of smartphones in 2027
• Rising DRAM prices are pushing phone makers toward reused memory
• Budget handsets currently account for most smartphones using reballed chips

Rising memory costs are pushing smartphone manufacturers toward an unusual form of component reuse, with refurbished memory expected to appear in more devices by 2027.

TrendForce estimates this approach will account for 3–4% of smartphones in 2026, before its market penetration increases to approximately 7–9% next year.

The shift is being driven by expensive mobile DRAM, which is increasing manufacturing expenses and putting additional pressure on brands serving price-sensitive buyers.

Reused memory offers manufacturers a cheaper route

Reballed memory comes from integrated circuits removed from existing circuit boards, then prepared for another installation after their original solder connections are carefully replaced.

Those connections use small metal spheres beneath the chip, allowing it to communicate electrically with the board while occupying very little physical space.

During reballing, technicians clear away the existing solder, prepare the contact surface, and attach a fresh set of appropriately sized replacement solder spheres.

The component can then be fitted to another board, although its suitability depends on its condition, handling history, and the quality of the refurbishment process.

TrendForce says 4GB LPDDR4X currently makes up most of the reclaimed supply, with use concentrated in inexpensive 4G handsets costing under CNY 1,299 (~$195).

Buyers in this category are particularly sensitive to retail prices, giving manufacturers an incentive to recover usable parts rather than rely entirely on newly manufactured memory.

Using previously installed chips can reduce component expenditure and help brands limit price increases, especially when customers have little room for more expensive smartphone models.

Although reballed memory is currently concentrated in cheaper smartphones, rising DRAM prices could encourage manufacturers to introduce it into more expensive models.

Higher memory costs could extend the practice

TrendForce expects memory price increases to accelerate in Q1 2027, adding pressure to manufacturers already managing expensive production.

That environment could encourage companies to consider older or lower specification parts in more costly products.

Even with faster adoption, reclaimed memory is expected to represent a small proportion of the total LPDRAM consumed by smartphones annually.

Consequently, its growing presence is unlikely to reverse the upward movement in contract prices, since its contribution to total memory demand remains limited.

The financial attraction also comes with technical concerns, particularly when manufacturers cannot establish consistent quality across components obtained from different suppliers and refurbishment channels.

Repeated heating can place stress on electronic packages, while damaged contacts or imperfect replacement work may affect electrical performance after installation inside smartphones.

Traditional reballing relies on heating solder until it melts, and component makers generally advise limiting devices to no more than three such heating cycles.

Laser-based methods can concentrate heat around solder contacts instead of warming the entire package, potentially reducing thermal exposure during component refurbishment and subsequent installation.

Manufacturers therefore face a choice between immediate savings and possible costs arising from failures, repairs, warranty claims, or dissatisfied customers over extended periods.

• Russia is adding antenna elements to its drones to resist jamming
• Russia already uses 16-element versions and is preparing production of 20-element ones
• Ukrainian specialists found 32-element antennas on Shahed drones during attacks on Kyiv

Russia is increasing the number of antenna elements fitted to its attack drones, making their satellite navigation systems more resistant to Ukrainian electronic warfare.

The upgrade complicates efforts to disrupt incoming weapons by interfering with the satellite signals they depend upon for navigation.

The navigation equipment, known as Kometa, produced by VNIIR-Progress, uses multiple receiving elements to maintain satellite connections when opposing forces transmit interference.

Russia increases antenna capacity as Ukrainian jamming becomes less effective

These components, commonly called patch antennas, help distinguish legitimate navigation signals from electronic interference, allowing drones to retain their intended flight paths.

Russia has already deployed versions containing 16 elements and is preparing manufacturing capacity for configurations containing 20 receiving elements.

However, Ukrainian specialists claim to have also identified Shahed attack drones carrying considerably larger 32-element arrangements during attacks against Kyiv.

Serhii Beskrestnov, an electronic warfare specialist advising President Volodymyr Zelenskyy on defense technology, explained how additional elements improve resistance against interference.

“The more elements, the more electronic-warfare signal sources the antenna can suppress,” Beskrestnov said.

The 32-element configuration reportedly combines two separate 16-element arrays, allowing the equipment to operate across different satellite navigation frequencies.

Russia's Kometa systems have also appeared in Geran drones, guided aerial bombs, Iskander-M and Kinzhal missiles, alongside the newer Filin cruise missile.

Per Ukrainian intelligence, Russian defense manufacturers ordered more than 100,000 modules in different configurations, covering several categories of guided weapons.

The scale of those purchases suggests that Russia intends to distribute improved navigation protection across multiple weapon platforms.

Ukraine explores alternative methods to defeat upgraded navigation systems

Despite the additional protection, Ukrainian specialists claim that larger antennas do not make Russian drones completely immune to electronic warfare.

The effectiveness of each Kometa configuration depends on its antenna arrangement, receiving elements, and processing algorithms that distinguish satellite signals from interference.

After studying the 32-element antenna purchased through secondary weapons markets, Ukraine claims to have a way to jam the antenna.

“By creating two systems — one operating in the L1 band and the other covering L2/L5 — we’ll be able to jam this antenna without any problems,” said Alkhimik, commander of Ukraine's Night Watch unit.

Another approach involves satellite spoofing, which manipulates navigation information rather than simply overwhelming legitimate signals with electronic interference.

Fly Group Ukraine has developed another countermeasure involving connected devices that transmit fabricated coordinates to approaching Russian drones.

The technique exploits their navigation systems, potentially causing onboard autopilots to redirect aircraft away from populated areas without detecting the manipulation.

“The drone continues to think it is flying toward its target, but its own autopilot, relying on the false coordinates, gradually changes course on its own and steers the craft away from cities into safe areas,” said Anatoliy Khrapchynskyi, development director at Fly Group Ukraine.

However, the effectiveness of these countermeasures against drones equipped with increasingly sophisticated navigation antennas remains uncertain, particularly as Russia continues upgrading its equipment.

If Ukrainian electronic warfare systems cannot reliably disrupt incoming drones, defending against them could require greater reliance on interceptor aircraft, specialized drones and expensive missile systems.

Via Defense News

• Unsanctioned AI use is increasing, according to a study of Google search data
• 50 terms were identified and their usage analysed, 48% more than two years ago
• The search terms point to use of shadow AI, in which AI tools are employed without corporate oversight or managerial approval

Employees are using AI in the workplace whether its use is approved or not. New research has revealed an increase in “shadow AI” – AI that isn’t really approved for use – and colleagues are relying on the technology to not only assist in work, but also cover their tracks.

The analysis was conducted by consultancy company Acora, which assessed 50 search terms in Google during the 12 months to September 2026. They found the terms were used 161,934 times, 48% more than in 2024 and a 250% increase over ChatGPT’s first year.

Google searches to delete AI chat history, circumvent blocks on AI tools, and to find free AI tools with no signup are among the terms to have increased. While students may be conducting some of these searches – another challenge entirely – there is a growing market for such tools, which could leave organizations looking to better implement AI policies with an even bigger challenge.

Shadow AI

Employing AI tools in the workplace ostensibly offers a productivity boost. The employee should ideally know if they are permitted to use AI, and have attended a session (in person or online) covering the policies and procedures around AI. While this is covered in most organizations, the pace of change with AI has left many businesses playing catch up.

So you have the results, as Acora’s research shows, of increased searches for “unblocked AI” (up 24% in 12 months) and “free AI no sign up” (up 36%) and “AI no sign up” (up 39%) in the same period.

Interestingly, searches for free AI note taking tools have also increased, up 28% on the previous year. If such tools are used in meetings, the information recorded could be stored on remote, unvetted servers, representing a data protection challenge for the organization.

The right controls?

Acora’s study highlights a growing concern: that businesses have AI policies that pay lip-service to the technology’s existence and rules over its use, but ignore the implications of overlooking its risks.

“AI transformation has reached the point where businesses are asking harder questions about what they can trust, how their data is being used and whether the right controls are in place," notes David Rabson, CEO of Acora.

“Concerns around trust, governance, security and data shouldn’t leave organizations stuck between ambition and action.” One solution is to help businesses close such gaps and implement “data foundations, governance and controls in place so they can adopt AI with confidence, rather than putting promising projects on pause.”

• Amplifon Americas disclosed a cyberattack that exposed personal, insurance, and healthcare-related information
• Stolen data included medical records and identifiers, with some individuals also losing Social Security numbers
• Company is offering identity protection services and warning affected users to watch for phishing and fraud

Healthcare company Amplifon Americas has disclosed suffering a cyberattack in which it lost sensitive customer data, including different healthcare information and Social Security numbers (SSN).

It is now urging customers to be careful with incoming emails, especially those related to Amplifon and its business.

The company said unidentified threat actors broke in on January 18 2026 at the latest and between then, and February 8, 2026, they were exfiltrating personally identifiable information (PII) the company has stored on its servers. Amplifon Americas spotted the intrusion on February 11, notified law enforcement and relevant agencies, and launched an investigation.

Amplifon attacked

The company did not explain exactly how the attackers made it inside its network, but it did say that it disabled “impacted accounts”. This would suggest that an employee account was compromised, likely via phishing or infostealer malware infections.

It also said it “promptly contained” the incident, and implemented mitigation measures such as “enhancing monitoring for signs of suspicious activity.”

Its investigation concluded on October 6 2026, and determined that the attackers stole people’s names, contact information, dates of birth, driver licenses (or other state identification numbers), other government-issued identification numbers, hearing or medical information, medical record numbers, dates of service, other health information, health insurance ID numbers, and other health insurance information.

Financial information, as well as SSNs, were exposed “for a limited number of individuals.”

Amplifon Americas is now notifying affected individuals and is offering 24 months of free credit monitoring and identity theft protection services.

“We are not aware of any fraudulent use of personal information as a result of this incident,” it said. “Nonetheless, it is always advisable to remain vigilant against suspicious behavior or attempts to misuse your personal information, which includes reviewing account statements and explanations of benefits. If you identify suspicious activity, you should contact the entity that maintains the information on your behalf.”

More than a million victims?

In its notification, the company did not say how many people had their files taken. However, according to The Record, a breach notice filed in Texas said 85,804 Texans were compromised, as were 1,027 individuals in Vermont. A similar notice was submitted in California, but without mentioning the exact number of people. Unfortunately the Office of the Maine Attorney General’s public-facing database, which used to hold an updated count on exposed individuals across the States, is still offline.

The Record also said “several” class action law firms said the total number of victims was 1,332,826, but this number is unconfirmed and Amplifon is not responding to media inquiries.

On the other end, no hacking groups have yet claimed responsibility for this attack. Usually, threat actors would publicly name and shame victim organizations, often sharing samples of the stolen files and threatening to release the full archive unless an extortion payment is made.

Healthcare organizations are among the most targeted firms mostly because of the large amounts of sensitive information they collect and store. Passwords can change, and payment details can expire. But medical histories, diagnoses, and other health records are often permanent, making them particularly valuable in identity theft and impersonation attacks.

• LG launches new 97-inch OLED panel reaching 4,000 nits, 165Hz and with 1.1% reflectance
• Arriving first in the LG W7 Wallpaper TV with Zero Connect wireless box
• The screen is a huge improvement over the previous 97-inch panel, which essentially dates back to 2022

Two weeks ago, we were excited by the prospect of more affordable massive OLED TVs, but noted that LG Display's current 97-inch OLED panels were relatively old TV tech. That's about to change dramatically. LG has unveiled a new version of its gorgeous Wallpaper evo TV with a brand new 97-inch OLED panel that could make it one of the best OLED TVs for big rooms.

The TV was revealed in the CES 2027 Innovation Awards, in which the LG evo 97inch Wallpaper TV is an honoree. And the specifications are very tasty indeed.

The 97-inch Wallpaper TV isn't the only new luxury TV from LG. The same awards also feature the LG Signature OLED T6, a new 77-inch version of LG's transparent OLED TV, though it's less clear what's new there. We know it's still limited to 60Hz, and will likely have far reduced brightness — though we've seen the previous models in person, and there's no denying that the transparent TV is cool as hell.

LG OLED evo 97" Wallpaper TV: key specifications

The evo TV has the model name 97W7 and features a 97-inch 4K display, in a frame designed for flush wall mounting. There's no on-board processing or ports to ruin its slim design: it has a wireless connection via LG's nifty Zero Connect system, which is lossless for quality.

The panel is said to deliver up to 4,000 nits of peak brightness, providing 83% BT.2020 color gamut coverage with 1.1% reflectance. This sounds like performance should basically match the five-star LG G6 in its smaller sizes — that model has a 97-inch version, but it uses the same panel LG has used since 2022, so the brightness, color and reflectivity are all weaker.

The new TV offers 165Hz with variable refresh rate support, and you get the Gallery+ art display features that turn the TV into an art gallery. And last but not least it has a virtual 11.1.2-channel virtual surround system built in.

Full specs, pricing and availability haven't been released yet but the new evo 97W7 looks like it's going to be something special — and that OLED panel will surely make its way into other LG TVs. It looks like 2027's going to be a good year for TV buyers with big rooms to fill and big budgets to spend.

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.

• Adobe wants creatives to stand out for their uniqueness, so AI will handle the rest
• Third-party AI agents can connect to major CC apps to handle admin work
• Apps like Photoshop, Lightroom and Premiere are also gaining new AI tools

Adobe has revealed plans to expand its Creative Agent across more of its flagship Creative Cloud applications, including Premiere, Lightroom and InDesign, in a bid to give AI more autonomy to handle the less creative side of creative work.

Unlike generative AI, which acts upon single prompts, AI agents can handle longer and more complex, multi-step processes.

The company positions the use of AI as putting creatives in the director's chair, enabling them to offload tasks that aren't really part of the creative workflow to AI agents to free them up more time to focus on the truly creative elements of their jobs.

Adobe's rolling out AI agent access to more apps

"The ideas, taste, style and vision that creative professionals bring to their work are what make their work stand out," Adobe Product Marketing VP Deepa Subramaniam stressed.

Importantly, using AI agents doesn't make a creative's work less human. Some common use cases include applying effects across multiple clips, tracking masked subjects and adjusting colours and exposures in Premiere.

Together with the expanded rollout, Adobe is also opening up the doors to third-party AI agents via MCP, starting with its most popular apps – Photoshop, Illustrator, Premiere and After Effects.

Granting permission to third-party agents to directly work within these apps means users won't have to handle exports and jumping between multiple tools, but it also means things like Photoshop layers and Illustrator vectors will still be preserved.

On-stage at the conference, Anil Chakravarthy, who is set to become Adobe's CEO effective from December 1 2026, demonstrated a third-party AI agent turning a long-form podcast into multiple social media snippets – a demo that was met with much excitement and applause from the thousands of attendees on the ground.

Besides the more autonomous side of AI, Adobe hasn't completely ditched generative AI, which still plays a major part in creative workflows. Photoshop will be gaining new tools to add or adjust lighting within an existing photo. Generative restyle will also turn natural language prompts into adjustable sliders, helping to remove the traditionally steep learning curve often associated with Photoshop.

Premiere also gains new GenAI upgrades, like audio enhancement tools and a generative remove functionality for removing unwanted objects from a series of video frames, while Lightroom sees additions like blemish removal and a tool to artificially open a subject's eyes – perfect for group photos where creatives would already be handling many other variables.

While some general Creative Cloud features are now generally available, some more advanced tools like Lightroom Open Eyes and Premiere Generative Remove remain in beta.

• Anthropic reveals a host of updates for Claude to help users at work
• Claude can now join meetings, analyze data and more
• It can also effectively connect to third-party tools such as Adobe, Box and more

Anthropic has unveiled a host of updates for Claude which could see the AI platform play a much more integral role in some of your most common work tasks.

Among the next expansions are tools which allow Claude to join video calls as a participant, meaning it learns first-hand from your meetings and can start work immediately.

Joining its existing suite of office software tools are new Whiteboards, Sheets and Data services looking to help users work and create more effectively - and new interactive connectors will now allow Claude to work inside the full versions of common workplace apps such as Adobe, Atlassian and HubSpot.

What's new with Claude for enterprise?

"A year ago, Claude changed how developers work," Anthropic wrote in a blog post announcing the new tools. "Engineers stopped asking AI for snippets and started handing it whole tasks, which freed them to spend their time deciding what to build. For most other teams, work has stayed much the same: the email gets written a little faster, the meeting comes with a summary."

"Now the frontier is moving to knowledge work, and people are starting to hand Claude whole tasks too."

The most striking new addition is undoubtedly Claude Meetings, which will allow Claude to join video calls as a visible participant.

Claude follows the conversation and starts on the work as people talk, taking notes, drafting plans, outlining timelines on whiteboards, answering chat questions, and utilizing connected work apps while conversations happen - and when the call ends, the notes, action items and everything Claude made are waiting in claude.ai.

Also new is Claude Whiteboards, a shared, real-time board which helps teams turn brainstorms and retrospectives into structured plans, timelines, and action items, and integrates directly into Claude Meetings, and Claude Sheets, a spreadsheet tool which allows teams to build and edit trackers and budgets collaboratively with Claude, complete with tabs, formulas, and dropdown menus.

Finally, Claude Data (formerly known as Claude Dashboards) connects company data sources (such as CRMs and warehouses) to generate dashboards, data stories, and narrated videos, featuring beta capabilities like certified queries and live sharing.

Alongside this, new Interactive Connectors will allow full versions of partner tools (like Adobe, Atlassian, HubSpot, Salesforce, NetSuite, and Workday) to open directly inside Claude, which can take actions directly within these apps with user approval.

Connectors respect existing user permissions, and admins can utilize features like Claude auto-connect, usage controls, and Smart Reports to monitor team productivity and costs.

Claude Meetings, Claude Whiteboard and Claude Sheets are all still in beta, with Claude Data and Interactive Connectors generally available now.

• Disney's new AI guide could make navigating Disneyland easier
• Works with Ray-Ban Metas, AirPods, and your iPhone
• Coming soon in beta for Magic Key holders through Imagineering Labs

When you're at a Disney park, or really any theme park, the idea is to be immersed in the experience, right? But there's also no sugarcoating the fact that a phone, especially one equipped with the Disneyland app, My Disney Experience for Walt Disney World, or any other park app, is pretty essential for planning and making the most of your day.

The app is essentially the brain of your visit, letting you check wait times for attractions, meet-and-greets, and other experiences, but also helping you find your way to food — I mean, we're all after a corn dog or a churro, right? — other services, and entertainment. But as we've heard from Disney Parks leaders in the past, constantly looking down at your phone can break the immersion.

Now, we're getting a glimpse at a potential alternative, and it's coming soon for a select group of Disneyland guests through a new Disney Imagineering app that's set to offer a rare peek behind the curtain.

Dubbed Imagineering Labs, the opt-in experience won't replace the main Disneyland app, but it will give guests the chance to try select experiences before they're fully developed or even guaranteed to launch. It's a pretty exciting approach, especially since feedback from guests could help shape what Imagineers ultimately bring to life.

The first experiment is an augmented reality Star Wars adventure called Secrets of the Spire, which is decidedly entertainment-focused and uses your iPhone's camera to overlay digital elements onto the environment and trigger physical reactions throughout Galaxy's Edge.

But the Disneyland Enchanted Guide is a bit more utility-focused, and it's seriously piquing my interest. It's an AI-powered assistant that lets you ask questions in natural language, like, 'Where's the nearest Dole Whip?' 'How long is the wait for Rise of the Resistance right now?' or whether you should come back later if the line is too long. It's also multimodal, meaning you can point your phone's camera at something in the park and ask about it.

Asa Kalama, Vice President and Executive Creative Director at Walt Disney Imagineering, described the thinking behind it to TechRadar: “Some of us are lucky enough to have that, you know, big Disney Parks fan as a friend, and there's no better park-going adventure than alongside them. They know all the tips, they know all the tricks, they can help you navigate your day. Well, what if we were able to provide that to every single guest that came to the park?”

And that's really the idea here. Rather than digging through menus or bouncing between different parts of an app, you can simply ask the Enchanted Guide and get an answer.

The interface is pretty neat, too. Rather than a traditional chatbot screen, the Enchanted Guide puts a familiar Disneyland map front and center, complete with attraction wait times and a glowing Mickey-shaped button at the bottom for asking questions. There are also buttons to type or use your camera, and the Guide can take into account your location, plans, and details you choose to share to make its recommendations more useful throughout the day.

It's using a combination of AI models to power the intelligence behind the Enchanted Guide, but there's been a whole host of customizations to make it feel uniquely Disney. Kalama explained that the team has worked to ensure the information comes from Disney's trusted sources, including real-time park information, and is delivered in a way that feels like the Disney voice, both literally and figuratively. From the pre-recorded preview I saw, it almost feels like chatting with a Cast Member you might stop on Main Street.

It gets even more interesting, though, when you can use another device to interact with it.

You might recall that at Meta Connect 2025, Disney Imagineering showed off an early tease of an experience at Disneyland using Ray-Ban Meta smart glasses, letting guests look at their surroundings, ask questions, and hear responses through the glasses. And while that was just an early glimpse, the Disneyland Enchanted Guide is bringing that idea to life, letting you interact with Disney's new AI assistant without constantly reaching for your phone.

Essentially, you can connect the Disneyland Enchanted Guide within the Imagineering Labs app to compatible Meta smart glasses using Meta's Wearables Device Access Toolkit. Once that's enabled and you're rocking the shades, you can chat with the Guide through the glasses without necessarily needing to pull out your connected phone.

The Enchanted Guide can make use of the glasses' microphones, speakers, and cameras. That means you could look at an attraction, ask about it, and get real-time information without even knowing its name.

Speaking with TechRadar, Chris Barbour, Director, Content & Ecosystem, Meta, explained that the partnership with Disney Imagineering started with a pretty simple idea: “The best of keep me immersed, keep me not having to look down at my phone, but also give me all the advantages of the technology being present for me.”

As Chris explained, one of his favorite parts is “being able to look at an attraction and get real-time information about that,” with the Guide using what the camera sees to understand what you're asking about.

On paper, and without having tested it yet, that could be incredibly handy within the park.

For instance, Asa shared a few examples, from checking the current wait time for an attraction to something a little more useful, like getting an alert when that wait drops below a certain threshold.

“Can you let me know when Dumbo the Flying Elephant has a wait time of under 10 minutes?”

The Enchanted Guide could then deliver a notification when that condition is met, leaving you free to enjoy other parts of the park in the meantime.

Imagine if it could also remind you when your next Lightning Lane reservation is coming up or if a slot opens for an attraction you want to experience. That's the kind of functionality I'd love to see. While Asa didn't confirm either capability, he did note that some features, including actually making Lightning Lane reservations, still live within the core Disneyland app. For now, the Enchanted Guide can explain how Lightning Lanes work, but will send you over to the main app when it's time to book one.

“There are certain capabilities that are on our roadmap that aren't yet currently implemented,” Kalama explained.

And while the Meta Ray-Ban integration is especially interesting, you don't necessarily need smart glasses to go hands-free with the Enchanted Guide. Bluetooth earbuds, including Apple's AirPods, will also work, letting you keep your iPhone tucked away while exploring Disneyland. AirPods won't support multi-modal on their own, as they don't have cameras ... yet.

Of course, wearing camera-equipped smart glasses in the parks could raise some privacy concerns. Meta Ray-Ban glasses have a built-in capture LED that indicates when the camera is in use, but that might not be immediately obvious to everyone around you. Guests are already accustomed to seeing phones and earbuds throughout the parks, so just be prepared for some looks.

There will even be a dedicated wake word for the Enchanted Guide, though it's still being finalized ahead of launch. That means you won't need to pull out your phone and tap the Mickey-shaped microphone button every time you want to ask a question. I'm especially curious to see what Disney lands on for the wake word.

And given that this is an AI assistant capable of understanding images and voice prompts, privacy is another important consideration. Both prompts and images are processed and then deleted, and the Enchanted Guide doesn't support video. So while it can use a camera to understand what you're looking at, it isn't designed to continuously analyze a live video feed.

Of course, this is still an experiment, and there's plenty we don't know about how well it will all work. I'm especially curious to see how quickly the Enchanted Guide responds, how accurately it understands what's happening around you, and whether it really can make navigating Disneyland easier.

And we'll have to wait a little longer to find out.

The Disneyland Enchanted Guide is still listed as coming soon within Imagineering Labs, and when it does launch, it'll initially be limited to Disneyland Magic Key holders. The experience will live within the Imagineering Labs app, where you'll sign in with your Disney account. With an eligible Magic Key linked to your profile, you'll be able to access the Guide once it's available for testing.

There will likely be some bugs, hiccups, or slowdowns along the way, but that's part of the whole Imagineering Labs approach. Rather than waiting until an experience is completely finished, Disney is opening up the development process and letting guests help shape what comes next.

A wider rollout is planned, though there's no firm date just yet, and for now, this is very much a Disneyland-focused experience.

But the goal behind it all is one that I think makes a lot of sense.

As Kalama put it, “The goal here really is to keep you invested and engaged with the physical world around you and the friends and family that you've brought to the park, to get you out of your device and back to your vacation as quickly as possible.”

And that's what has me excited about the Disneyland Enchanted Guide. Disney isn't trying to replace the magic of being in the parks with another screen or another digital experience. Instead, it's exploring how technology can help you make the most of your day without constantly having to reach for your smartphone.

There's still plenty to prove, especially when it comes to how well the AI works in a busy, sometimes unpredictable theme park environment. But if Disney can pull this off, the Enchanted Guide could become the kind of park companion you don't realize you need until you've spent a day with it.

And if it means more time enjoying Disneyland and less time staring at my iPhone, I'm all for it.

• Researchers showed Microsoft Entra Agent ID blueprints can be abused to socially engineer malicious AI agents into organizations
• A single approved blueprint can spawn hundreds of agent identities with access to corporate resources
• Defenders should restrict third-party agent consent, apply least-privilege controls, and monitor agent creation events

When Microsoft first introduced Entra Agent ID as a way organizations can identify and monitor AI agents, it most certainly did not think it would be used to smuggle malicious agents into corporate environments. Yet, that’s where we are today.

Entra Agent ID is an extension of Microsoft Entra. It’s a system that allows companies to create AI agents, give them digital identities, and then decide which services and what data they can access. Expecting organizations to deploy dozens (if not hundreds) of these agents, Microsoft also introduced “agent blueprints”, essentially templates companies can use to create multiple AI agent identities, each with their own permissions.

In a new report, security researchers Proofpoint laid out how a threat actor could create a malicious blueprint in their own Microsoft environment and then - using classic social engineering techniques - convince a victim to approve it in their own environment. To make matters worse, when a malicious agent enters the environment, they can “clone” themselves, creating multiple additional identities, all tasked with doing different things.

How the attack plays out

Proofpoint found the attack starts with the attacker setting up a blueprint in a Microsoft Entra environment they control. They can configure its credentials and the permissions they want it to receive and also make sure it can be authorized by other organizations.

Then, they create a link that leads to a genuine Microsoft authorization page. Unlike your run-of-the-mill phishing attack, this one is different because the login page doesn’t need to be fake and spoofed. The attacker does, however, need to have some basic information about the target organization, such as their tenant ID and details about an existing user they can specifically target.

Then comes the fun part: the victim is presented with the link (this can be done via email, social networks, instant messaging platforms) which asks them to approve the blueprint and its permissions. The link is legitimate, and so is the popup prompting the approval - nothing suggests this is actually a highly sophisticated phishing scam.

If they click the link and approve the blueprint, the malicious AI is added inside their organization.

The higher-authority the target - the bigger the potential damage, since for organization-wide application permissions, the person approving must have the necessary admin privileges. Ordinary users may be able to approve certain delegated permissions, depending on company policy.

One of the first things the AI agent receives, once installed, is a permission called AgentIdentity.CreateAsManager. This allows the threat actor to create additional identities linked to the blueprint that’s now inside their target’s environment. A single blueprint can support up to 250 agent identities, which means there can be a swarm of malicious bots running around at the same time.

Raising awareness while limiting access privilege

The attacker then obtains access tokens for those identities, making them valid members of the company with approved access to different resources which they can then exfiltrate as they please.

“Organizations should raise user awareness of the risks associated with agent consent, particularly when granting permissions to ‘third-party’ agents from untrusted sources,” Proofpoint warned. “Much like user consent settings limit users from granting consent to third-party apps, these settings also extend to agent blueprints and their agent identities.”

Proofpoint stressed agent blueprint ID is consistent across all tenants that give their consent, which is a common indicator to detect or remove known malicious agents. Applying the principle of least privilege will help limit the impact, the researchers concluded, utilizing controls like Conditional Access policies for agents and Entra ID protection for agents. “For monitoring, agent and blueprint events are surfaced in the audit log, in a similar way to app consent operations,” the report concludes.

• The smart Wyze Lever Lock is easy to install, and has built-in Wi-Fi
• The 2K Wyze Solar Cam comes bundled with a photovoltaic panel
• Both are affordably priced, and subscription plans aren't mandatory

Installing smart locks can be a hassle — you often have to drill new holes in your door, and usually spend time setting up a separate Wi-Fi hub to connect the lock to your home network too. It doesn't have to be that way, though.

Wyze — a tech company that's made a name for itself in affordable home security — has launched a new lever-operated smart lock that you can fit with nothing but a Philips screwdriver. Just unscrew your existing door handle, screw the Wyze Lever Lock into place, and you're done.

Most of the best smart locks we've tested here at TechRadar are deadbolts designed for your front door, but the Wyze Lever Lock is made for the other entryways that you might actually use more often, such as garage-entry doors and side entrances. You can use it on internal doors too, so you can protect your home office (and any expensive IT equipment and sensitive documents therein).

Your opening options include a fingerprint scanner (which Wyze claims will release the lock in 0.5 seconds), a touch-sensitive numeric keypad, and a pair of physical backup keys for emergencies.

The Smart Lever Lock has built-in Wi-Fi connectivity, so there's no need for an additional hub, and runs for 12 months using eight AA batteries. It's available now for $99.98 direct from Wyze, with free shipping.

2K recording, no wires

So you've secured your doors — now what about home surveillance? Wyze has you covered there as well. Last month the company launched a wireless outdoor security camera that you can fit anywhere, and is powered by just an hour of sunlight per day.

The Wyze Solar Cam comes bundled with a photovoltaic panel, which you can mount right beside the camera or up to 10 feet away so it catches as much sunlight as possible.

The camera has impressive recording chops too, streaming 2K video footage in wide dynamic range, revealing extra detail in bright areas and deep shadows.

Like the Lever Smart Lock, the camera is surprisingly affordable, with a list price of just $69.98 for a single camera direct from Wyze. If you don't need the solar panel, you can pick up the camera by itself for $59.98. Best of all, there's no monthly subscription necessary, though if you want cloud recording and AI features, plans start at $2.99 per month.

For more options without monthly payment plans, take a look at my roundup of the best subscription-free security cameras.

• A rumor claims MSI is planning some big price increases for Nvidia GPUs
• The RTX 5080 and RTX 5070 Ti could be hit with price rises of around $1,200 and $600 respectively
• Admittedly those are the worst-case scenarios, but with big price rises already in place, these GPUs are already pretty much unaffordable anyway – so even more modest bumps in cost are going to hurt

If you thought Nvidia's higher-end GPUs were pricey enough, you'd be spot on there – but they are going to get a lot more costly still based on the latest from the rumor mill.

Wccftech highlighted that leaker Harukaze5719 on X flagged up a post on ChannelGate (a common source of supply chain rumors in Asia) with details of what's supposedly a set of big prices increases from MSI.

We're told that the RTX 5080 could be visited with a price hike of between 1,300 to 8,000 Yuan this month in China, which equates to a jump of about $200 at minimum in the US, or $1,200 in the worst-case scenario (or the equivalent across other regions).

Yes, that isn't a typo, the claim is that MSI's RTX 5080 could jump by over a grand – the MSRP of this graphics card – towards the upper end of this predicted price increase.

The rumored hike for the RTX 5070 Ti is almost as frightening and could be from $120 all the way up to a potential $600 rise.

The source further claims a rise of between $75 and $150 for the RTX 5070, and smaller price bumps of around the $50 mark for the RTX 5060 and 5060 Ti. It's mostly the higher-tier GeForce GPUs which are being hit here, though, and possibly in a really nasty way.

Analysis: a pricing shambles

Previous buzz from ChannelGate (last week, as TechPowerUp noticed) regarding several other board makers suggested price rises in the order of $400 (in the US) for the RTX 5080 and $150 for the RTX 5070 Ti, which particularly in the latter case is at the lower-end of these fresh predictions. Keep a firm grip on that saltshaker, in other words, but everyone seems agreed that there are some major hikes coming to Nvidia GPUs in Asia, and those same pricing pressures will surely apply across the globe.

We'd better hope that the higher-end of these new forecasted hikes don't become a reality, anyway, as they'd leave the GPU market in a pricing shambles, frankly.

To take the US market for example, the RTX 5080 is already $2,000, so are we looking at it being nearer $3,000 come the end of November? It certainly seems like it's heading towards $2,500, which is a frankly astonishing amount for a non-flagship GPU. (Although given the RTX 5090's vanishing act, it is the de facto flagship at this point I suppose).

The RTX 5070 Ti is the more worrying prospect here, though, because the RTX 5080 has always been an expensive card – most folks won't fork out a grand (the MSRP) for a GPU anyway, as that's the budget for the whole PC (or most of it) for the majority. Whereas the RTX 5070 Ti is a realistic buy at its MSRP, but the GPU has already entered effectively unaffordable territory even now thanks to more recent price rises.

Currently in the US you'll pay at least $1,400 for an RTX 5070 Ti from big-name retailers like Newegg or Micro Center (and more for the beefier cards). If a $600 rise is inflicted on this graphics card, you're looking at $2,000, or closing on it, for the privilege of owning one. And this is supposed to be a mainstream mid-range (well, upper-mid-range) GPU, remember.

I was already prepared for this to happen – which is why I've been recommending buying a GPU sooner rather than later, if you need one, as this year has rolled on – although I wasn't expecting this kind of steep spike (and certainly not so suddenly).

This new rumor may be wrong, or the heftier price increases this leak is theorizing may not come to pass, but I'm pretty certain we are going to see all GPUs get bumped up in price as 2027 begins, and I fear there may yet be worse to come. (Certainly the broader RAM crisis is predicted to intensify next year).

Still, at least you might be able to get some money back on your purchase by letting Microsoft use your shiny new Nvidia GPU to train its AI (in one possible future timeline, that is).