In the run-up to a show like Gamescom, my inbox becomes a blazing trash fire. Inundated with requests to see everything from phenomenal-looking indies to Barbie Horse Ride and Rescue, literally thousands of game demo invitations charge into my mailbox, waving their banners and blowing their trumpets in a bid to attract attention.

One game that successfully breached my email defences was Chronicles: Medieval. Revealed during Summer Games Fest, this impressive real-time strategy (RTS) not only promised massive, strategic top-down battles blended with third-person combat – but narration from beloved homegrown actor Tom Hardy. And who was I to turn down battering virtual knights with Tom Hardy?

Tom? Hardly

"So, unfortunately, as much as we love him, Tom is not a part of the game.” Sam Millen, lead producer at Raw Power Games, tells me, sheepishly, as I ask him about the Bane actor’s involvement, “We do have the incredible Lars Mikkelsen in the game, acting in a chronicler role - taking a voiceover role during character creation and during parts of the game. Sadly, Tom Hardy was only part of the trailer.”

After ten emails and two press releases mentioning the BAFTA-winning actor, it turns out it was all a Trojan Horse-esque ruse to get me in the room. While hardly an auspicious start to my behind-closed-doors preview, thankfully for its Danish developers, Chronicles: Medieval's absurdly ambitious gameplay soon wins me over in spite of its misleading marketing, breaching my defences and ensuring a total coverage victory.

Chronicles: Medieval is an impressively nerdy history epic. What I didn’t expect, however, was just how deep the game’s roleplaying elements go. Slated for an Early Access launch in 2027, Raw Power Game’s debut mixes gargantuan troop battles that would make an eight-year-old blush with a vast, explorable world map. In this rags-to-riches tale, players start out as a lowly peasant and build their standing through questing, diplomacy, and, of course, combat.

From campsites to castles

“You start off without owning any land. You're just out in the wild.” Explains Millen, “Maybe you have a couple of men you grew up with in the town who walked together and decided to join you as you embark into the world.”

Drawing equal inspiration from Civilization, Heroes of Might and Magic, and Mount & Blade, the game’s 1313 AD map spans from the Pyrenees to Poland, and the British Isles down to Northern Italy. Yet while this could have been a broad Age of Empires-esque offering, Chronicles: Medieval revels in the specificities. Players can zoom all the way down to inspect individual sub-regions and see the individual lords that rule them.

“Each culture - England, France, HRE [Holy Roman Empire] - has different army trees: levies, household troops,” says Millen. “You're going to be able to upgrade your troops as they get experience…then you'll potentially take troops from being levied infantry to being cavalry, progressing up those paths."

There’s a heavy dose of Crusader Kings here, too. You can buy tavern intelligence to reveal the map, recruit mercenaries, or bypass combat entirely if you’d rather play that way. Despite its trailer's combat focus, the pair sell me Chronicles as a true sandbox. Whether you want to become a wealthy peaceful merchant, a feared bandit, or even claim the English crown through political backstabbing, the choice is yours.

"It's a sandbox, so you can do whatever you want to do in the world.” says Millen, “There's no linear narrative that you're set off on. If you want to become the best merchant in the world and become incredibly rich from just selling goods, you can absolutely do that.”

Boots on the ground

As a sea of hundreds of troops forms up along a hillside of gently swaying grass, Millen zooms in on each soldier individually, explaining the historical significance of their crests, weapons, and surroundings. In other words, this is a game built by massive history nerds, people who are deeply knowledgeable about the time period they’re depicting and passionate about getting it right.

Supporting up to 2,000 total troops on the battlefield, forces are drawn from three distinct unit categories: household troops, peasantry, and mercenaries. The pre-battle prep begins in classic top-down fashion, as Millen manoeuvres his various units, placing archers at the rear, pikemen at the front, and sending a troupe of covert cavalry up around the tree-shaded hillside to flank the French forces on the other side of the plain.

Once the battle commences, you drop straight into a third-person action-role-playing game (RPG) perspective, sending your commander charging into battle. As if he’s in a fully fledged action RPG. As Millen’s militia gallops towards the advancing enemy horde, he sends four of them flying with his poleaxe. Dismounting and unsheathing his blade, he parries sword blow after sword blow with impressive fluidity. Here, enemies have posture meters, parries require precise timing, and every weapon swing carries weight.

"When you start out, you have low proficiency, so your attacks are slower, and you don't have combos,” says Senior Game Designer Gareth Bourn, “But over time, you unlock a lot of these skills — including leadership skills.”

While cleaving heads on the battlefield will rile up and inspire your troops, if the tide of combat starts to turn in the enemy’s favour — or you take a particularly nasty hit from an enemy - your leadership skills will determine whether they leave you high and dry.

Mace swings and morale

"The troops have their own minds.” says Millen, “You take morale hits if you die — but they're going to keep fighting for their lives until the point where their morale dips, either because they're losing or because they've already taken a hit from seeing their commander die. If you're a great leader. They want to stay with you and fight to the death longer… They can pick up your injured body, take you back to a settlement, and you can recover, rest, get your health back, replenish your troops, and fight another day.”

As commander, you can still tell units to advance, hold, etc. In the thick of combat, keeping a hands-on tactical element amidst the manic melee. Yet as if to prove that this game’s show-stealing battles aren’t its only parlour trick, players can choose to bypass them entirely. "You can even auto-resolve battles if you want,” says Bourn, “But every time somebody clicks auto-resolve, I shed a small tear.”

While Chronicles: Medieval’s ambitions are impressive, these aren’t just lofty 1.0 goals. I'm surprised to learn that when it hits Early Access, Chronicles: Medieval will also include a four-player co-op campaign map alongside custom battle modes. The duo tells me that full sieges — complete with trebuchets, crumbling walls, and battering rams - are slated to follow shortly after, with full console support and robust modding tools planned for the official 1.0 release.

Part intricately detailed Total War-esque historical battle simulator, part player-choice-driven RPG, much like a ragtag group of bannermen holding off a king’s army, Chronicles Medieval punches far above its weight. Developed by a hundred people in Copenhagen, this ambitious RTS could well give Microsoft and 2K’s Age of Empires and Civilization a run for their money.

While it might have been a false Hollywood bluster that got me in the door, it's the admirable ambition, detail, and potential that have kept me thinking about Chronicles: Medieval since. If Raw Power Games can deliver even half of the features they’ve promised, this could be a very special RTS indeed.

As the entire gaming industry rearranges its calendar to avoid Rockstar’s impending juggernaut, another long-gestating title has quietly emerged from the shadows – 1666: Amsterdam.

Hitting Steam Early Access in August this year after 13 years of development hell, this tale of kittens and covens is the brainchild of Assassin’s Creed 1 & 2 and Prince of Persia: The Sands of Time director Patrice Désilets.

Revealed during 2026’s Summer Games Fest, this tale of sorcery and witchcraft immediately turned heads. As a hooded wiccan soared across thatched rooftops and ducked and weaved her way between soldier-ridden canals, crimson curses and talking cats teased an exciting, thematically unique-looking adventure. Yet despite its refreshing spin on open-world traversal, this time-bending tale took its first steps into existence all the way back in 2011.

"It’s been a long journey,” reflects Désilets with a warm but weary smile, sitting opposite me in a brightly-lit cardboard office, erected in the business area at Gamescom. Picking up the controller to proudly show off his gothic adventure, he takes one of the game’s dual protagonists for a covert stroll down a packed Amsterdam canal street, before shaking off her civilian disguise and raining hex-induced havoc on the screaming populace.

Once championed as THQ’s Assassin’s Creed beater, the inspiration for 1666 first took hold on a 2011 holiday to the Dutch capital. “Walking around, seeing the houses in Amsterdam. It’s really a level design city,” recalls Désilets, ”I could picture an Assassin’s Creed there…there's a facade, and all the mystery. Then I passed a building inscribed with ‘Anno 1666.’ and immediately I thought, 'What if I do a game about the Devil set in Amsterdam?'"

The Devil’s in the details

Originally envisioned as a cinematic Satan simulator, 1666: Amsterdam was meant to be the more gothic-feeling spiritual successor to Désilets’ 2010 smash hit game, Assassin's Creed: Brotherhood.

Thanks to his previous pedigree, publisher THQ was in, and so he spent the next two years crafting what we’ll affectionately call ‘The Devil’s Creed’. Yet like many deals made involving the devil, it soon appeared that Désilets Lucifer-staring epic was afflicted with a thematically fitting curse. “I started working on it for two years, then THQ went bankrupt.” Désilets recalls.

With THQ then suddenly snapped up by his old employers, Ubisoft, Désilets found himself briefly back at the Assassin’s Creed publisher, who almost immediately let him go – leaving poor old Lucifer trapped in development hell. Fiercely fighting his former employers for the rights back, Désilets eventually regained custody of his demonic baby in 2016. “…we got everything back…the 3D, all the concept art, and some animation,” Désilets grins. Yet in the years that had passed, something had changed for Désilets.

Despite having grafted to reacquire his IP, as time passed, Désilets no longer found himself excited about writing a game starring the devil’s apprentice. The reason? He was put off by a certain hit 2016 TV show. “Lucifer, the TV series, came out, and meanwhile, I'm like, 'Is it really the subject matter I want to attack?” shrugged Désilets. While Désilets had fallen out of love with the devil, he still found himself drawn to the idea of an evil-infested 17th-century Amsterdam. “What if there was a witch hunt going on in Amsterdam in 1666?” Désilets posited, as a new story unfolded before his eyes, “The only thing that I really kept was this idea of also doing 1999 [a dual time period alongside 1666], playing off two sides of the same coin: reflection and duality. That's when I pitched it to my team in January 2020, and [I] rewrote almost everything.”

Processing phobias through polygons

Thanks to being forced into remote work during the pandemic, retooling 1666 took the team at Panache six long years. While the initial Satan Sim was a Désilets-led project through and through, with his newest studio having previously made the 2011 survival game Ancestors, he’d grown to trust his team at Panache, letting their ideas help remould his 2011 original prototype in unexpected ways. “It’s funny, because it’s really a team game, yet in some ways, if we go back to my original PowerPoint, it's crazy how similar it is.”

Yet for all its violence, hexes and history, there is a familiarly agile, gravity-defying protagonist at the centre of Panache’s latest. "... I feel like I'm always doing the same game ... about traversal…” laughs Désilets, “the cat is my way to play with verticality [here]... Even in Ancestors, where you were an ape in a clan, it was still this sheer pleasure of going around a 3D world. Once I analysed myself: 'Why is this always in my fucking games?' "Because I've got vertigo! In real life, I can't climb – so maybe that’s why [you climb in my games]. Don't ask me to climb anything. It's a joke in my family.”

While soaring across canals as an agile Wiccan - or darting across rooftops as a black cat - provides the vertical thrills that Désilets physically can't enjoy in real life, 1666 departs radically from the traditional Assassin's Creed formula in one surprising way. “Each time you fail, there's a reshuffle of your abilities. You must continue your investigation with a different set - that’s where the roguelite element appears,” explains Désilets. “You might fail as a fighter and come back leaning towards witchcraft, so you have to think differently. You go through the challenges the way you want - it's your game, man. If you want to spend six hours in a row being the cat, I don't care.”

Swapping polish for collaboration

“You're not supposed to do [Early Access] with a narrative game,” shrugs Désilets, “But we're giving you roughly between 10 to 15 hours' worth of narrative-driven, open-world-ish content... [and] our goal is to have 1.0 by the end of next year, with console releases at the same time.”

For Désilets, the old way of shipping a game was done out of necessity, not choice, with the Assassin’s Creed veteran always wishing that he had more player feedback and additional time to refine his art.

“There was a time in game development when the game was finished... There was nothing you could do — no patches. That time drove me crazy.” Says Désilets . “Now... give me feedback…and then I can tweak. This is what game making is all about: having a dialogue. Yes, I can make a really badass character like Altaïr, but if you take the pad and make him turn around in circles, the badass character is gone because you're the one finishing the game. It's a co-creation.”

Narrative-driven games usually rely heavily on cinematic polish to pull players in, making 1666’s rough-around-the-edges Early Access release a peculiar sight. Yet in an industry plagued by ballooning budgets and sweeping studio closures, launching a single-player adventure this way might not just be an anomaly- but a canny survival strategy.

“I do my games with my team a bit in a bubble,” says Désilets when I ask him about the precarious current state of the industry. “I'm not always in those conferences. I'm not totally sealed off from what's going on, but you walk around, there’s an energy here…that reminds me of the old times... and lots of quality new games. Maybe we can attract new gamers into what people liked in 2007 when they first played Assassin's Creed. It's got that vibe, right?”

One of the best-known facets of the Harry Potter world is Hogwarts, and its houses. Gryffindor, Slytherin, Hufflepuff, Ravenclaw; you know the animals, the colors, the characters, the traits. You've probably even been sorted yourself, or have self-prescribed a house.

But which really are you? No, I don't mean which do you want to be, or even which one another tool like Pottermore has sorted you into. There's an even more reliable system, which exceeds even the Sorting Hat in its prescience and accuracy.

That's right: a personality quiz. Not any old personality quiz, but a tech one — this is TechRadar, after all — which will sort you into your accurate Hogwarts house. It's like you're living through the events of Harry Potter, but all based on whether you prefer Windows, MacOS or Linux, or what kind of robot vacuum you'd buy.

We've sorted questions into three batches. First up, we'll start with some simple buying decisions. Then, we'll investigate how you make such buying decisions. Finally, we'll put you into some real-world scenarios, to see how you'd act in certain tech situations.

All of these will help you know once and for all: which Harry Potter Hogwarts house do you belong to? Or, at least, which house do your tech habits point you towards?

Big fan of the franchise? We've also got a 30-question Harry Potter quiz, to prove it. And if you need to use your tech decisions as a horoscope to further refine your personality, we also have a Beatles personality quiz to see which member of the famous band you're in.

Or, you could stick to Harry Potter, and find a way to enjoy the franchise more:

Asus has beaten every other manufacturer to market with the ROG Rapture GT-BN98, the world's first commercially available Wi-Fi 8 gaming router.

It's a bold flagship release for a wireless standard that isn't even finalized yet — and with zero Wi-Fi 8 client devices currently on sale, reviewers had to get creative just to test what it can actually do. I dug into three in-depth reviews to see whether it's worth buying now.

What reviews say

Build quality drew consistent praise. It's "extremely well constructed with very good attention to detail," said Blacktubi, while Guru3D leaned into the router's dramatic looks, joking that unboxing it feels like "a small alien spacecraft has crash-landed on the desk."

YouTuber Landpet was blunt about the performance side, calling it flat-out "an absolute beast of a router."

On paper, the GT-BN98's theoretical link rates are identical to its Wi-Fi 7 predecessor, the GT-BE98 — Blacktubi pointed out the headline numbers "are the exact same." But real-world range and throughput told a different story.

Design

The chassis is large by any standard. It measures roughly 35cm by 35cm and weighs over 2kg. It needs "a proper, dedicated spot" rather than a small shelf, and ruling out wall mounting altogether given the size.

The antennas dominate the look, adjustable around the perimeter and giving the router an almost insect-like silhouette when fully extended. The top panel mixes a metallic finish with geometric detailing and translucent panels, plus the expected Aura RGB lighting.

Build quality feels solid throughout — no flex, clean panel fit, antenna arms that hold their position. Front-edge LEDs give a quick status read without opening the app.

Thermally, this Asus router added a thicker aluminum top plate and an upgraded heatsink layer for passive cooling. Stress testing turned up no signs of throttling.

Design & ports

Inside is a 2.6GHz quad-core processor and 2GB of RAM, running a quad-band radio setup — 2.4GHz, dual 5GHz bands, and a single 6GHz band — for a combined theoretical throughput of 25,000Mbps.

A North America-only "Pro" variant swaps the second 5GHz band for an extra 6GHz band on the UNII-8 spectrum, which isn't approved for use in most other regions.

For wired gear, there are two 10GbE ports (one general-purpose, one dedicated gaming port), four 2.5GbE ports, and a single gigabit port, with up to six LAN ports active depending on configuration. USB 3.2 Gen1 and USB 2.0 round things out for network storage or 4G/5G backup internet.

Performance

Real-world speed testing — done entirely with Wi-Fi 7 client devices, since Wi-Fi 8 hardware doesn't exist yet — found wired and wireless backhaul holding up strongly even at 100 feet away

Over a wired 10G connection, speeds matched his 5 Gbps ISP plan exactly in both directions. On Wi-Fi, internet speeds hit 4.1 Gbps down and 2 Gbps up.

Local network testing — where speed isn't capped by the ISP — pushed even higher: 4.6 Gbps down and 4.2 Gbps up over wireless backhaul between two mesh units, with wired backhaul close behind at 4.5 Gbps down and 4.2 Gbps up.

Range held up well too. At 20ft, speeds stayed around 4 Gbps down; at 50ft outdoors, 2.7 Gbps down; and even at 100ft, it still managed 1.92 Gbps down, though upload dropped sharply to 718 Mbps.

A more controlled comparison, pairing two GT-BN98 units against two GT-BE98 units, found "significantly higher" throughput at long range with the newer hardware.

The more surprising finding: those gains aren't limited to future Wi-Fi 8 gear. Testing with an existing Wi-Fi 7 MacBook still measured "more than 10% higher average throughput" on the GT-BN98 versus the older Wi-Fi router — evidence that upgraded radio components benefit today's devices too, even without any Wi-Fi 8 features actually switched on.

Firmware

The ASUSWRT 6.0 was singled out as a real step up, with the interface "noticeably more modern and much more responsive" than Asus's decade-old firmware.

The headline addition is the new Adaptive QoE engine, which "smartly allocates network resources to balance gaming, streaming, and daily tasks," replacing the older Adaptive QoS system.

Security also got an upgrade via the revamped AiProtection suite, which was called "genuinely impressive" for including network-level ad and tracker blocking at no extra cost.

The new Wi-Fi Insight tool also correctly flagged interference from a Zigbee smart home devices on the 2.4GHz band, a decent addition which impressively “works as intended”.

Price

Pricing is still rolling out region by region ahead of the router's wider October 2026 launch.

The flagship GT-BN98 Pro is currently the only model on sale, listed on ASUS's Canadian eStore at CA$1,299.99.

A US launch price of around $899 has been floated by early previews but isn't yet official.

The standard GT-BN98 (non-Pro) has also been priced at €960 in Europe, though North American pricing for that specific model hasn't been confirmed yet.

Given the router is shipping ahead of Wi-Fi 8's official IEEE finalization — not expected until 2028 — buyers are paying an early-adopter premium on top of already-flagship router pricing.

Who should buy it

This is a router for enthusiasts buying ahead of the curve, not something you need for Wi-Fi 8 today — compatible client devices aren't expected until 2027 at the earliest.

Still, the benefits aren't purely theoretical: better range, backhaul speed, and firmware improvements apply right now, even to existing Wi-Fi 7 devices.

Verdict

The good

- Meaningfully better range and throughput, even on older Wi-Fi 7 devices

- Premium build quality with no thermal throttling under load

- Adaptive QoE and free-included ad/tracker blocking are genuine upgrades

The bad

- No Wi-Fi 8 client devices exist yet to fully prove the standard

- Large footprint requires dedicated desk space

- High price tag

The verdict

Worth it if you want the best router hardware available today and don't mind future-proofing ahead of the Wi-Fi 8 ecosystem. Hold off if you're not running multi-gig internet or don't need flagship-level range right now.

Mass-Effect-a-likes are a lot like interstellar buses: you wait ten years for one to arrive, and then two warp into view at once.

Unveiled at this year’s Summer Game Fest, both The Expanse: Osiris Reborn and Exodus look set to capitalise on the gaping hole left by BioWare’s beloved MIA franchise. Yet where Exodus’ Matthew McConaughey-starring project left Gamescom to rave reviews, in my infinite wisdom, I ventured into the far reaches of the outer rim to investigate the other Mass Effect-inspired epic: a sci-fi action-adventure based on the hit TV show, The Expanse.

With nine books to its name and six seasons of acclaimed television, Osiris Reborn was, I wagered, the candidate most likely to deliver a deep, narrative-led role-playing game (RPG) experience. Yet after an hour and a half spent sneaking across space stations, blasting soldiers, and fleeing stalking mutants, it turns out The Expanse isn’t really a narratively rich Mass Effect revival at all. It’s a so-so sci-fi Gears of War.

“We start the game from the Phoebe incident... It's set between Season 1 and Season 2 of the TV series... within the first two books,” explains Natalia Turina, lead level designer, as I pick up the controller and take my first steps.

Put into the space boots of the player protagonist and her twin, J, I join the duo as they infiltrate a space station owned by a shady private military corporation, Protogen. After a recent shipjacking by their crew, the duo has snuck deep into enemy territory with one vital task — delete the shipjacking’s virtual evidence from Protogen servers, and get the hell out.

Less normandy, more office bureaucracy

Thanks to its rich source material, I went into my demo expecting a deftly written, choice-driven world. Yet while the fiction of The Expanse promises a glimpse into a complex future, playing Osiris Reborn feels like traveling back into console gaming’s beige-looking past.

After slowly inspecting the busy cargo ship undercover, I duck and weave between armed enemy forces without arousing any suspicion, eavesdropping on ship personnel before heading to the ship’s computers.

As I pretend to be shipping a piece of intergalactic cargo and helmet-clad guards stare me down with assault rifles cocked, this should be a masterclass in tension and subterfuge. Instead, as I’m tasked with walking back and forth between computer terminals, I’m not immersed in the universe’s knotty politics — I’m bogged down by the minutiae of intergalactic shipping logistics.

“For choices in an RPG to feel meaningful, you need to believe in this world,” says Turina. “Every space needs to make sense — what is this place, how does it operate? This is why we spend so much time on the environment.”

It’s an answer that makes sense on paper, yet while rich in detail, there’s nothing exciting about the lore nuggets that I uncover. Well, I guess I certainly know how intergalactic shipping containers operate. After some more perfunctory dialogue, my cover is blown. Luckily, just as the alarm sounds and the station is alerted to our presence, I’m saved by an ambush. As a squad of a rival militia faction breaches the airlock, the formerly mundane depot erupts into a hail of gunfire.

A world stuck in the Xbox 360 era

Thankfully, combat fares a bit better than the world-building. Part Gears of War, part Binary Domain, there’s a retro 2010 feel to its firefights — for better and for worse.

Much like Commander Shepherd before me, I sprint between cover, deploy a drone, and blind-fire into a series of heavily armoured mercenaries, swapping between an assault rifle and shotgun. As I toss grenades at advancing militia and shatter glass over crouching enemies, I’m told players are free to change their specialisms on the fly, tweaking their tactical approach as needed via extensive skill trees.

It’s all functional lizard brain stuff, but while enjoyable, I worry that the combat [so far] doesn't feel dynamic enough to truly elevate it from the Xbox 360 era of cover shooters — let alone compete with modern peers.

Character interactions also struggle to hit the mark. When I encounter a lone marine — his hardened status signalled primarily by relentless swearing — our protagonist and her companion accept him as an ally almost immediately, despite being deep in enemy territory and under fire from multiple forces. It’s a jarring, immersion-breaking moment that once again highlights the quality of BioWare’s dialogue.

For all the granular detail painstakingly poured into Osiris Reborn’s lavishly rendered environments, its characterisation and dialogue currently feel paper-thin.

Running from horror, missing the tension

A late-demo pivot sees me being stalked by a Protogen-developed experimental mutant who broke containment amidst all the gunfire. Whoops. Impervious to bullets, the creature forces me into a frantic sprint down half-lit corridors. Where in Resident Evil or Dead Space this would be an exercise in sphincter-clenching terror, blending eerie exploration with a heart-pounding chase sequence, here the pursuit lacks any real tension. From the creature’s unconvincing AI to a complete lack of cinematic flair, I merely felt like I was running from A to B, meekly jogging toward a doorway to trigger the next cutscene.

As shiny as Owlcat’s mix of cover shooter and RPG is, I came away from my hour-and-a-half Osiris Reborn demo feeling fairly underwhelmed. Mass Effect won hearts not just through its worldbuilding, but through the strength of its writing: sharp wit, genuine heart, and moral dilemmas that stayed with you long after the credits rolled. It’s why during its ten-year absence, fans of BioWare’s space opera have been pining for a franchise to fill that galaxy-sized hole. So far, it sadly doesn’t look like Osiris Reborn is up to the task.

While I’m told the full game will feature wider exploration and more side missions and roleplaying, what I saw left me concerned that Osiris Reborn is less a Mass Effect spiritual successor and more a pale imitation.

Owlcat Games may borrow the same core Mass Effect ingredients, it appears to cook them together with none of the same skill or soul; its blend of corridor walks, light puzzles, and chest-high cover shooting feels closer to Kane & Lynch than Knights of the Old Republic.

With the full game not due out until next year, there is, of course, still time to refine the formula. Here’s hoping that the final adventure lives up to the title — expanding its source material’s beloved universe and, with any luck, helping its RPG aspirations be reborn.

I've been wearing the Apple Watch Series 12 to bed every night for a week now, which means, after seven nights of calibration, I finally got the benefits of the revamped Vitals app and the new Readiness Score. The Watch has to learn your baseline metrics so it can see outliers, which is why it takes a week for the Vitals app to gauge whether your heart rate variability, respiratory rate, blood oxygen and more fall within your 'typical' range.

Readiness is the last metric, as it aggregates your sleep quality based on these parameters, compares it with your activity levels and gives you a score from 1 to 10 on how energetic and prepared-for-the-day you might be feeling.

This is the first time Apple devices have had a Readiness metric, compared to the long-serving nature of the features on the best Garmin watches, Samsung and Google wearables, and many others, so naturally, I wanted to compare it with another wearable device.

These days, I'm most often using my Garmin Cirqa for sleep tracking, so I kept wearing it on my right wrist, and the Apple Watch Series 12 on my left. Below, I've compared my sleep scores from across the week for accuracy, as well as my Readiness at the end of it all.

Sleep scores across the week

I fully expected Apple and Garmin to provide disparate scores at first, only to tighten up as the week progressed and Apple learned more about my body. Sure enough, that's exactly what happened, as Apple gave my first night a score of 93 compared to Garmin's dismal 62.

After that, you can see above that the graph follows the same general trends: a rising incline as the week progresses, with a dip in the middle, and a high final night.

The main outlier other than night 1 is night 6, for which I have no explanation. Garmin gave me a score as high as 87, while Apple gave me a mere 68. You can see the difference in the graphs above, but I've listed all my scores in the table below, too.

Perhaps there will always be these outliers here and there. Research from health data API creator Terra measured 77,000 nights' sleep across six major wearables companies, and found that while most trackers are accurate enough to converge on duration and general trends, sometimes outliers in their data exist due to the different sensors and algorithms used between manufacturers.

The study stated that "once you break sleep into stages [light, deep, REM phases of sleep], agreement [between wearables] falls apart: REM, deep, and light can diverge by hours in direct comparisons."

However, are these outliers statistically important, as long as the wearables are all charting similar trends coming to the same conclusions?

Readiness

At the end of the week, I got my first Readiness score from Apple. It was a 7/10, indicating that I was 'ready' without being 100% prepared for a mammoth workout.

Checking the Garmin Connect app, I was surprised to see a Readiness score of 70/100, indicating that even though there were some outliers along the way, the wearables came to the exact same conclusion after a week of constant wear. Success!

Conclusion: Pay attention to the trends

While some discrepancies were present during the week, Garmin and Apple both came to the same conclusion on Readiness and offered a similar-shaped sleep score graph, indicating that they're internally consistent and generally quite accurate.

I'd be happy to wear either one to track my sleep, Readiness and fitness in general terms. However, the differences in individual data points are a reminder not to get too caught up into the weeds of the data: as clever as they are, these devices are imperfect, and they estimate. You can see differences in my Apple Watch Series 12 vs Garmin Fenix 9 Pro test too.

If you find yourself stressing too much over your scores or your data, or you feel differently than your watch says you ought to feel, prioritize the messages from your body over the ones from your wearable — and don't be afraid to get a night of untracked sleep now and again to free yourself from sleep score anxiety.

Welcome to the September 2026 Blu-ray Bounty. If this is your first time reading, you can check out previous editions of the Blu-ray Bounty at the link. This is where we test the latest 4K releases each month, to recommend whether you should add them to your 4K collection.

4K Blu-ray is the best way to view movies at home and is my go-to source for testing the best TVs and best soundbars as TechRadar’s AV reviewer. Plus, I’m a collector myself, and always adding new titles to my already-packed shelves.

This month, I’ll be testing ten new titles from September 2026. And there’s some big news: we’ve upgraded our in-house sound system! After using the Samsung HW-Q990C soundbar system since the start, we’ve updated to a full separate Denon and Bowers & Wilkins home theater system, consisting of the Denon AVC-3900H AV receiver, a pair of B&W 603 S3 floorstanders for fronts, a pair of B&W 607 S3 bookshelves for surrounds, a B&W HTM6 S3 centre speaker and a B&W ASW610 subwoofer. This is a 5.1 system for now, but we’ll be adding in two side speakers soon to create a 7.1 system.

Elsewhere our setup remains the same. The TV is the LG G6, one of 2026’s best OLED TVs and the Panasonic DP-UB820 4K Blu-ray player, one of the best 4K Blu-ray players on the market. As usual, I’ll be testing each disc on its picture and audio quality and not the movies themselves.

Supergirl (Warner Bros.)

Supergirl follows the story of Kara Zor-El AKA Supergirl,(Milly Alcock) Superman’s cousin, as she is thrown into an unlikely alliance to hunt down Krem (Matthias Schoenaerts), a space trafficker and outlaw, after he harms her dog Krypto.

Supergirl looks superb in 4K. Colors are both punchy, shown in the neon signs at the various bars or the laser from Supergirl’s eyes and the various weapons of criminals, and true-to-life, shown in the iconic colors of the blue, red and yellow in the Supergirl costume. Scenes in space showcase the disc’s rich black tones and strong contrast, with the balance between the black of space and the bright stars and Sun. Textures are crisp, with people looking realistic and environments and objects packed with detail.

The disc has a Dolby Atmos soundtrack that is a real showcase for the object-based sound. As Supergirl flies around, the trajectory of sound accurately follows her movement on screen, with the sound mapped precisely from speaker to speaker. Spaceship engines and the boom of Supergirl taking off deliver meaty bass that’s well-controlled, and fight scenes generate plenty of hefty impact from front channels.

Star Wars: The Mandalorian & Grogu (Disney)

A spin-off movie from The Mandalorian TV show, Star Wars: The Mandalorian & Grogu follows the two titular characters as they are tasked with the rescue of Rotta the Hutt, the only surviving heir to Jabba the Hutt.

Presented in Dolby Vision, The Mandalorian & Grogu looks excellent. It’s a slick presentation with lifelike textures and detail: characters take on a 3D-like depth that makes them appear true-to-life. There’s also plenty of high contrast scenes with rich black tones that look great on the G6 OLED. The balance between light and dark tones was superb too, with the gleam of the Mandalorian’s armor contrasting well with the dark surroundings.

Colors are also delivered with some nice pop, with the reds of lasers and blaster fire looking vibrant. On a wooded planet, the greens of the trees are lush and natural.

The Dolby Atmos soundtrack here is brilliant. During the opening scene, as the Mandalorian infiltrates an Imperial warlord’s hideout, gunfire sounds hefty with plenty of impact. Explosions generated a weighty rumble that had our testing room rattling, but later fight scenes showed that the bass was tightly controlled.

Sound placement is superb too. As gunfire scatters from one end of a room to another, the trajectory is accurately mapped from the front to surround speakers with real precision.

Hamilton (Disney)

One of the most widely known musicals in recent theater, Hamilton follows the story of Alexander Hamilton, one of the founding fathers of the US. The musical is known for using different styles of music in its songs, primarily hip-hop. This home video version is a recording of the original 2015-2016 Broadway run.

The 4K disc supports a Dolby Atmos soundtrack that sounds excellent. Vocals are cleanly delivered and well balanced with other elements in the mix. Backing choirs and vocalists feed through the rear speakers, and individual instruments are precisely mapped to each channel. In ‘Alexander Hamilton’ (the opening number), there’s tight bass, punchy drums and excellent use of Dolby Atmos’ object-based profile, with every element combining to conjure up an immersive soundstage.

Other tracks such as You’ll Be Back demonstrate the Atmos soundtrack’s more delicate strings and pianos, while other tracks such as Guns and Ships again show the bass-driven hip-hop styling mixed with epic strings and rapped vocals. All of it sounds impressive, especially through our new home theater system.

The picture here is presented in Dolby Vision and looks great. Black tones created by the use of stage lighting are deep and rich, actors look true-to-life, there’s plenty of refined detail in costumes, and colors have a nice vibrancy to them throughout the entire production. The fast movement of the actors and cameras will also provide a test for your display. Shadow detail is brilliant throughout as well, with the spotlights and stage lights creating some impressive-looking shadows.

12 Angry Men (1957) (Criterion)

12 Angry Men (1957) follows a group of 12 jurors debating whether to convict or acquit a teen charged with murder. 11 jurors agree, but 1 juror’s caution forces the men to reconsider the evidence.

As with most Criterion releases of classic movies, this restoration breathes new life into the picture. Textures have been upscaled nicely, giving people in particular a fresh, 3D-like look and textures in the environment have been refined brilliantly in the deliberation room. The black and white picture also shows strong contrast, with a terrific range of gray tones. There’s a fair amount of grain retained here, but it suits the movie.

The supported soundtrack is only monaural, so there’s nothing here to really show off your home theater. But the soundtrack has been cleaned up, meaning dialogue is clear throughout and other effects such as the shuffling of chairs is clear too.

Obsession (Universal Pictures)

Obsession follows the story of Ian (Michael Johnston), who uses a ‘One Wish Willow’ to wish that his crush, Nikki (Inde Navarrette), will fall in love with him. His desire is granted, but he soon discovers that this triggers a dark and dangerous series of events.

Presented in Dolby Vision, Obsession looks great in 4K. There’s a lot of deep dark tones throughout, with some strong contrast on display, created by the use of lighting. Textures are crisp, with characters appearing lifelike throughout the movie. There’s some stellar shadow detail here as well, with a lot of detail legible even in particularly dark scenes. Colors, while not especially varied, do look bold and vibrant when used, particularly reds, which look both punchy and natural.

There’s a Dolby Atmos soundtrack supported here, which is most effective with the movie's tense and foreboding score. Bass synths generate a lot of rumble from the subwoofer, and there’s a real sense of immersion as the other synths swirl across the speakers. Other effects also sound great, delivered with real precision and clarity. In one scene where Nikki smashes a bottle, the sound of splintering glass is realistic, and the following visceral moments sound gruesomely lifelike as well. Moments of violence are rendered with spine-chilling detail too.

Disclosure Day (Universal Pictures)

Disclosure Day follows the story of Margaret Fairchild (Emily Blunt), a meteorologist and Dr Daniel Kellner (Josh O’Connor), a cybersecurity specialist, who end up on the run from a government agent (Colin Firth) after discovering extraterrestrial life is real, and they attempt to reveal the information to the world.

This is a slick-looking 4K presentation, with lifelike textures and detail. Colors are punchy and realistic throughout the movie, whether it’s clothing, cars or environments, such as the green fields of the forest where Kellner hides out. High contrast scenes look excellent as well, with a great balance between light and dark tones. HDR highlights really pop on screen too, shown with rear headlights in cars and spotlights used later in the movie.

The Dolby Atmos soundtrack featured here is fantastic. Gunshots have plenty of impact, explosions and car crashes generate a hefty bass response, and object-based sounds are expansive. As cars drive through a dusty, gravel path, the sound of the gravel under the tires is accurately mapped to rear channels, and seamlessly synchronizes with the on-screen action.

Batman: Knightfall Part 1 (Warner Bros)

Batman: Knightfall Part 1 is the first part in a planned trilogy of animated movies that adapts the ‘Knightfall’ storyline in Batman comics, that ran from 1993-94. The first part follows Batman as he confronts the escaped inmates of Arkham Asylum, who have been freed by Bane.

This is a clean, crisp animation presented in 4K, Dolby Vision. Colors are satisfyingly vivid and give the picture a real comic book look. Reds and oranges are especially eye-catching, shown again with blood and the fire of Azrael’s claws. The animation itself also appears intricate, with sharp textures that give the picture depth. Black tones are rich and look great on our LG G6 OLED.

The disc supports a DTS-HD 5.1 MA soundtrack that sounds superb. In the opening scene, as Azrael ambushes members of the Kobra Cult, gunfire delivers good punchiness, Azrael’s movements are tracked accurately and as he stalks around the gang, the sound follows the action precisely. Punches have good heft and weight to them and the soundtrack makes every use of each channel, giving every sound room to breathe.

Little Odessa (Criterion)

Little Odessa tells the story of Joshua Shapira (Tim Roth), a hitman who accepts a contract in the Brooklyn neighborhood where he grew up, where upon his return, he reconnects with his estranged Soviet immigrant family.

This is another great 4K restoration from Criterion. Textures have been upscaled well and cleaned up, giving the movie a more modern look. Night scenes are perfect for showing off the disc’s strong contrast, with rich shadow detail, deep dark tones and a good balance with brighter tones, such as streetlights and snow. Colors look natural throughout, and even scenes with snow benefit from the new Dolby Vision HDR grade that adds more brightness to the white tones.

The disc supports a DTS-HD 2.0 surround MA soundtrack. While this won't be a tought test for your home theater system, the general ambience of the Brooklyn streets adds some good atmosphere to the viewing experience, with car horns, nearby pedestrians and the subway all adding some life. Speech is also clear throughout.

Barbarian (20th Century Studios)

Barbarian follows the story of Tess (Georgina Campbell), a woman who books an AirBnB and discovers that it’s been double-booked by a man named Keith (Bill Skarsgård). After Tess decides to stay anyway, she soon discovers dark secrets within the house.

This is another great-looking 4K Blu-ray. Where your display will be really tested is with the movie’s many dark scenes. On our LG G6, black tones were inky and rich, and contrast was strong as the light from torches as well as dimly lit lamps juxtaposed dark surroundings. Shadow detail is also excellent, and textures have an authentic look, particularly with people. While the spectrum of colors on offer is relatively slim, they appear vibrant and natural in equal measure when used.

A DTS-HD 5.1 MA soundtrack is featured here. From the opening scenes, the use of surround adds to the immersion, with the pounding rain still giving a sense of height, despite the absence of actual height channels. The movie’s tense, mysterious score is punchy, with the bass synths generating some good low-end tones that whip up a chilling atmosphere. Other more subtle effects such as the scratching of rocks and shallow breathing are detailed enough to really set your nerves on edge.

The Abominable Snowman (1957) (Hammer)

The Abominable Snowman follows Dr John Rollason (Peter Cushing), a British scientist who joins an expedition led by American Tom Friend (Forrest Tucker) to find the legendary Yeti in the Himalayas.

As I’ve found with the last few Hammer releases I’ve tested, The Abominable Snowman looks very good in 4K. The restoration looks excellent, with textures upscaled to give the movie a much fresher look. This is shown most of all in people’s skin, which is given a 3D-like level of detail. The black-and-white picture looks great too, with deep blacks, punchy whites and a good range of gray tones. Shadow detail is fantastic too, with a great use of shadows to create strong-contrast scenes. Grain has been cleaned up a lot and though there is some left, this is a clean-looking presentation.

There are two soundtrack options here: mono and 5.1 surround. While the mono is no doubt more accurate to the original movie (due to the timeframe), the 5.1 does add an extra layer. Sound effects such as shuffling feet through snow and the howling winds are given more space, while speech is mapped to center channels for a clearer sound. Scenes where avalanches crash produce surprisingly strong bass as well, with some solid rumble emanating from our subwoofer.

The pressure to get my son a phone arrived sooner than I was expecting. He was barely halfway through Year 6 (that's 5th Grade for any American readers), and his friends started taking their newly acquired smartphones on trips to the park, when previously their hands would be full of footballs and frisbees. I felt like I had missed a parent’s meeting where it was decided that this was the life we wanted for our kids.

The realization that I was well behind the curve in this area kickstarted a lot of late-night Googling, where I spent hours researching the best phone for my son. I initially thought I’d just get him a dumb phone. At least then I could pretty much guarantee protecting him from harmful content. But this wouldn’t have allowed me to track his location, and for safety reasons, that was a non-negotiable for me.

Then I discovered child-friendly phones like the British-designed Sayph or the more well-known HMD Fuse. These types of phones enable parents to restrict app use, create a whitelist of contacts, and even create a school mode. They also often have additional filters for blocking inappropriate content. I liked the idea, but the handset options were limited.

It was only then that I stumbled across the Google Family Link app, which enables parents and their children to create healthy, positive digital habits. Through simple, user-friendly tools, you can do almost everything that a dedicated child-friendly phone can with Family Link, and if you use it to the fullest, you can create a safe online environment for your child.

Google Family Link works across Android, ChromeOS, and iOS devices — the child's phone, however, must be an Android device, since Google doesn't offer the 'kids' version of the Family Link app for iOS; only the 'parent' version.

I’m not a helicopter parent, honest!

I’m sure I’m not the only one who likes to know where their child is. The advent of GPS has not only made it possible to track your child but also made it feel slightly creepy to do so. The fact that you know where they are without them knowing where you are feels slightly odd. But then again, they are your children, and you’ve been tasked with keeping them safe.

I like to know whether my son has arrived at school, if he’s still at the park, or which friend's house he’s decided to go to. As much as I want to know where he is, I don’t want him feeling that he’s got to be constantly on his phone texting me updates. That would go against my desire to keep him as screen-free as possible for as long as possible.

This is where Google Family Link becomes your ally. With it, you can not only keep an eye on their location but also create specific places named Family Places. Whenever your child enters or leaves these locations, the parent receives a notification. With that, you can keep a loose eye on them without following their little avatar on Google Maps as they cycle to school. Full disclaimer: yes, I have done that as well.

"No, I’m not giving you access to that app”

Given this was going to be my son’s first ever experience of a smartphone, I was keen to only give him the very basics. I didn’t want to provide access to anything that was remotely addictive for fear that he wouldn’t be able to put the phone down. I was surprised how easy it was to achieve this with Google Family Link.

All I had to do was navigate to the ‘Block and limit apps’ tab to see a full list of available apps on his phone. From there, I was able to select an individual app and either block it completely or set a daily time limit. The interface is incredibly well laid out, such that even the least tech-confident parent could manage to navigate it.

I also wanted some level of control over his contacts. Okay, now I really am starting to sound like a control freak. With Family Link, you can not only set ground rules for incoming and outgoing calls and texts but also require any contact to be approved before they can be contacted.

The text interactions my child might have then got me thinking about issues he might run into with group chats. I’m sure we’ve all read the stories of bullying and inappropriate sharing of private photos, and I didn’t want my son to be exposed to this.

I was therefore grateful to see that Google offers sensitive content warnings, which add a layer of protection for Google Messages that blurs incoming nude photos and discourages the sending of nude photos. I can’t comment on its effectiveness, although it’s nice to know it’s there. It’s worth noting that many child-specific phones offer greater levels of nudity protection, if this is a concern of yours.

Locking down websites because the world is a scary place

My son needs ongoing access to some select websites to be able to do his homework, but beyond this, I was able to control which sites he was able to visit when signed in to Chrome. The app enables you to activate a filter that blocks explicit sites, although even Google admits in the app that “No filter is perfect.”

I therefore decided to only allow approved sites. I have to say, this has been the only element of my Google Family Link experience that has been a pain. I was hoping to just copy and paste URLs into the app and let Google take care of the rest. Unfortunately, that didn’t work. Because a lot of sites have subdomains, especially Microsoft websites, I found that unless I entered all the specific URLs, Family Link wasn’t clever enough to be able to implement what I needed it to.

I eventually got it working and was glad that my son was able to request access to specific websites simply by typing them into Chrome and selecting ‘Request Access’. That took care of any that I missed. The request pops up on my phone, and as soon as I approve it, he can access the site straight away.

A safety platform that evolves

One of the best things about Google Family Link is how easy it is to add restrictions as well as remove them. As my child gets older, I’ll want to give him more freedoms, and with the app, it's simple and straightforward for me to do so.

With child-friendly phones, they always have the risk of being 'childish' and will therefore quickly become outdated for your child. With Google Family Link, it’s possible to make any phone child-safe, whether they have the latest Samsung Galaxy or a generations-old Google Pixel.

I'm glad, then, that rather than buying some expensive new child-specific phone, I settled on an app that allows my son to use a 'normal-looking' Android smartphone that I can control remotely. Google Family Link isn't perfect, but it's the easiest and most feature-packed solution to child safety I've come across so far.

Monster Hunter Wilds: Ascendance promises a fresh expedition into the skies, with players summoned there by the resurgence of ancient dragons and the promise of upgraded weaponry. At Tokyo Game Show 2026, I went hands-on with this upcoming expansion, getting up close and personal with the downloadable content's (DLC) interconnected aerial archipelagoes and battling against two brutal monsters with the use of the brand-new Boost Bracer.

The demo opens in a camp, with the reliable Alma serving up two important quests: Hunt the Alpha Araketa, a level 2 monster, and Slay the Teostra, a level 5 monster. Before jumping straight into the hunt, though, it only felt right to check things out at home first. The camp itself is a visual marvel in keeping with the high standards set by the base game — something here about how it looked; check demo vids online.

The demo is keen to focus on gameplay, though, and understandably offers little in the way of non-player character (NPC) chatter. Still, I was pleased to find a small jump across a ravine hides a barrel-rolling side game that, while unnecessary, was a fun side activity nonetheless. But I digress; it was finally time to break some monster parts.

Falling with style

The first expedition dropped me on a jagged island floating high above any visible Earth, where I was set to track the Araketa. The design of the space lands somewhere between James Cameron’s Pandora and the ruins of a medieval coastal castle. Rocky structures hang in the air above my eyeline, with aggressive tree roots tunnelling through them and disrupting their typical form.

As I came upon an unnervingly empty open area, I spied what looked like petrified dragon wings and tails buried in the white rubble, the blistering blue sky illuminating the stone and casting a dusty screen — so much so that I could almost feel my throat dry up as I stepped through the chalky stone.

Eventually, I spotted the Alpha, but it wasn’t alone. Typical of head honchos, the beast had two underlings, identical in species but much smaller in size.

New to Monster Hunter, the Araketa is a wyvern with a smooth, stone-like exterior painted with what looks like a turquoise-and-yellow platter pattern. Unlike a typical dragon-faced beast, the Araketa features a bulbous, boulder-like head, two wings with sharp claws at the back, and a long, winding tail. While beautiful, the most important note is that all three of these extremities are used in combat.

As I entered the battlefield and alerted the trio to my presence, the Alpha let out an ear-splitting screech before taking battle formation. As one of the underlings jumped up into the air, the Alpha and the second monster flew towards me at a startling pace. In no time, their attacks began, the younger creature attempting to headbutt me with its hammerhead, while the Alpha flanked me from behind.

Diving side to side, the Alpha then took its turn, digging its claw into the earth and performing a spin attack that knocked me off balance. As I regained my footing, the third skybound monster, which had been suspiciously absent, used its momentum to swoop down and take a swipe, too. It was overwhelming, but exhilarating.

My weapon of choice for this battle was the Insect Glaive, which proved useful for platforming around their bombardment. Thankfully, as I started to feel overwhelmed, I remembered I had a new tool to flex: Boost Driver attacks.

Another new addition to Monster Hunter Wilds is that players now wear a Boost Bracer, which enables weapon-specific special attack types. At the bottom of your screen, you’ll see a gauge with three charge slots. As you attack the monster, these slots will illuminate light blue, indicating you have charge. When in position, you can press R1, circle, and triangle in tandem to unleash a visually striking brutal attack.

For the glaive, this saw my character throwing his body into a series of explosive shifting sweeps. Beyond the spectacle and boosted damage, the upgraded moves did well to reset momentum, and I was particularly pleased to see I didn’t need any specific resources to harness its power.

As the battle drew to a close, I had just enough time to look around and take in the ambient wildlife of the new biome, spotting all kinds of creatures from rocky-backed alligators to fluffy, chicken-like birds.

Among the structures, I also got to experience the new wind mechanics mentioned in the marketing material and made use of my Seikret to dive across them, which was fun if not a little rigid — perhaps I’ll have more control in the full release.

The belly of the beast

The second quest sent me to a wholly oppositional location. Instead of blue skies and white rock, I was met with sludge and volcanic ash plumes rising from various organic chimneys dotted across the horizon. This area was composed of tones of grey, occasionally disrupted by orange or red illuminations.

The castle ruins had also been swapped out for what looked like giant pots, akin to Elden Ring's Iron Fist Alexander if he fell into some wet tarmac. Unlike the ambient wildlife of the first area, there was little more than Sisyphian dung beetles here, continuously rolling a solitary ball up a hill.

Descending to the bowels of the cave system, I spotted my prey, slumbering amongst fire and brimstone. Unlike the Araketa, the Teostra isn’t new to the franchise, with the elder dragon appearing in Monster Hunter World and long before that in Monster Hunter 2. However, it is the beast's first appearance in a long time, and he’s packing some serious heat this time around.

Before I could begin my barrage, a pop-up on screen alerted me to an unskippable step — if I didn’t drink a cool drink, I would lose health during the fight, presumably from boiling alive in my own skin. I’m happy to report that ignoring this message was a silly mistake, and it wasn’t long before I fainted and was forced to make the descent again.

Salve applied; I dove headfirst into what already appeared to be a complicated fight. Double my height, the Teostra’s feline form was blazing, its attacks involving a mixture of fizzing fire bombs and cat-like swipes. When used in tandem, it became steadily overwhelming, and I was pleased I had my attention-stealing Palico to draw away some of the heat, both literally and metaphorically.

To keep from burning alive, I leapt in and out of its proximity, buckling its back legs and taking huge swaths of damage — safe to say, in the twenty minutes I had to win, I was not successful. Thankfully for the sake of my pride, as I returned to the camp a loser, I noticed the player next to me was struggling against the Teostra as well; even with a different weapon and impressive dexterity, they failed to kill the beast in the time allotted.

It’s clear Monster Hunter: Wilds Ascendance intends to escalate the hunting in the base game, enabling players to wield powerful tools in environments designed to genuinely test their preparation and survival skills. As someone who dropped off the base game after hitting credits, this expansion provides more than enough reason to kit back up and head into its refreshed, uncompromising world.

Monster Hunter Wilds: Ascendance is releasing in 2027.

The race to crown the best camera phone of 2026 is heating up.

Back in January, Samsung released the globally available Samsung Galaxy S26 Ultra, before Oppo and Xiaomi dropped the Oppo Find X9 Ultra and Xiaomi 17 Ultra, respectively, both of which were made available in the UK but not the US (Australian buyers, meanwhile, got the Xiaomi, but not the Oppo).

Google entered the race with the Google Pixel 11 Pro in August, and now Apple has released the iPhone 18 Pro, which is earning plaudits for its variable aperture feature and pro-level video capture capabilities.

So, wherever you are in the world, you’ve likely got at least three compelling camera phones to consider at the time of writing, and I took two of them — the iPhone 18 Pro and Galaxy S26 Ultra — on a recent trip to Hawaii to see how they stack up against one another.

Below, I’ve added examples of both phones’ wide, zoom, macro, ultra-wide, and selfie-taking capabilities to help you decide which model is right for you. Then, at the end, I've summarized my thoughts on how they compare.

Specs

Before we dig into the comparison, here’s a quick roundup of each phone’s key camera specs:

In addition to the optical zoom distances listed above, both phones offer an additional ‘optical-quality’ zoom (8x for the iPhone, 2x and 10x for the Samsung), but they do so via a sensor crop rather than fixed focal lengths, so I’ve prioritized their true optical zoom distances in the ‘zoom’ section below.

It’s also worth noting that, to save storage, both the iPhone 18 Pro and Galaxy S26 Ultra capture images in smaller file sizes than their large sensors suggest by default. Specifically, the iPhone defaults to 24MP (using Apple’s ‘super-high-resolution’ imaging pipeline), while the Galaxy S26 Ultra defaults to 12MP (via pixel-binning).

I stuck with these out-of-the-box shooting settings for each of the comparisons below.

Wide

iPhone

Samsung

In this picture of a statue by the beach — “The Acrobats” by John Robinson — we can immediately see the difference between the two phones’ approaches to color science.

The iPhone’s image is the more natural of the two, with more pronounced shadows and muted colors, versus the Samsung’s brighter, more contrast-heavy image. This difference is particularly apparent on the statue’s torso and patch of grass to the right of the scene.

iPhone

Samsung

Here, again, the Samsung’s image has punchier, arguably more artificial-looking colors, but the difference isn’t quite so stark as in the first image. Both photos are equally detailed.

iPhone

Samsung

Again, only a subtle difference between the two images, but the leaves are certainly greener in the Galaxy’s shot (see the top right-hand side of the image)

iPhone

Samsung

The iPhone and Samsung take radically different approaches to capturing this shadowy pathway. The former’s image is more realistic, with strong detail maintained where the leaves meet the sky. The Galaxy’s image is impressively detailed, too, but its colors are much more stylistic than natural, which is emerging as a theme in this comparison.

iPhone

Samsung

This, for me, is the clearest example yet of the color science difference I’ve referenced below every image so far. Both of these shots are equally detailed, but the Samsung has definitely dialed up the saturation versus the iPhone’s more neutral shot.

iPhone

Samsung

Moving on to a sunset image, and the iPhone does a good job at maintaining the clarity of both the foreground and background, while the Galaxy goes full “Oh, so you want a sunset shot?” mode. I definitely don’t hate the Monet-style sky the Samsung has captured here — it's certainly the more aesthetic of the two — but in doing so, the phone obscures almost everyone in the foreground. The iPhone’s shot is, again, more balanced, but much less stylistic.

iPhone

Samsung

This image, though, proves that the Galaxy S26 Ultra can effectively balance foreground and background in low light. I actually think it does a slightly better job than the iPhone here, but there’s not much in it.

iPhone

Samsung

Again, pretty similar results here, but the iPhone definitely chooses a warmer hue that fades from orange to red, while the Samsung goes from red to an almost pink color. This is, in fairness, a particularly tricky shot to capture.

Zoom

iPhone

Samsung

Back to that statue again, this time at 4x optical zoom for the iPhone and 5x optical zoom for the Samsung.

The level of detail captured is impressive in both cases, but the Galaxy’s image is, in my opinion, the sharper of the two, which makes sense given the phone's optical advantage over the iPhone — the Galaxy essentially has more information for its processing algorithms to work with, resulting in a slightly more detailed image.

I also like the way the Galaxy dials down the color of the blurred clouds in the background to really bring the statue into focus, which isn't something you'd expect given the phone's gung-ho approach to colors we've observed so far.

iPhone

Samsung

Interestingly, in these photos — taken at the same optical distances as above — the Samsung has again dialed down the colors to produce a much grayer image than the iPhone, though I'm not sure that approach works for this particular scene.

The iPhone's image is warmer and, in parts (see the cracks in the pavement), a touch more detailed.

iPhone

Samsung

This is an example of each phone’s maximum digital zoom capabilities — 40x for the iPhone, 100x for the Samsung. Yes, 100 is a bigger number than 40, but you can see why Apple doesn’t bother to offer anything beyond that figure. In both cases, the phone is guessing at details via processing trickery, rather than actually being able to capture them, which results in two pretty unusable images.

Still, the Samsung will produce iPhone-like results up until around the 50x mark, so it's a more versatile zoom camera overall.

Macro and ultra-wide

iPhone

Samsung

Macro mode — where your phone automatically switches to its ultra-wide camera lens when you move close to a small subject — was my favorite test to carry out on both the iPhone 18 Pro and Galaxy S26 Ultra, because macro photography is something Apple and Samsung phones have been good at for years.

In this example, as we’ve seen previously, the iPhone 18 Pro opts for more realistic, neutral colors, while the Galaxy’s shot has way more contrast. The greens in these two images are quite literally different shades.

iPhone

Samsung

Again, the level of detail here is comparable across both shots, but this is the first instance where I actually prefer the Galaxy’s deeper, more artistic-looking colors. The iPhone's brighter image is closer to reality, though.

iPhone

Samsung

In this example, it’s the other way around; I prefer the iPhone’s colors to the Samsung’s. However, you’ll notice — if you zoom in on the vascular tissue of the leaves — that the Galaxy phone actually captures more detail.

iPhone

Samsung

As in my comparison between the iPhone 17 Pro and OnePlus 15’s cameras, I’ve only included one regular ultra-wide example here because I was so preoccupied with the phones’ wide and telephoto lenses while shooting in Hawaii that I forgot to take more with their respective ultra-wide lenses (perhaps my colleague was right in saying that phones don’t need ultra-wide lenses at all — but that’s a debate for another day).

In any case, everything I said about the first few photos in this comparison applies again here. Both phones capture excellent detail — I'd probably give the edge to the iPhone, since it captures the graininess of the sand slightly better than the Samsung — but their approaches to color science differ dramatically, which is evident in the sand, sky, and grass pictured above. The iPhone captures the scene as it is; the Samsung captures the scene as it thinks you want it to be.

Selfies

iPhone

Samsung

Both of these selfies are detailed and well-balanced in terms of color, but the iPhone's image is the slightly sharper of the two, no doubt because the 18 Pro has a larger 18MP front-facing sensor than the Galaxy S26 Ultra’s 12MP sensor. It does a better job of maintaining the clarity of my flyaway hairs, for instance.

It’s worth mentioning, too, that the iPhone 18 Pro’s ‘Center Stage’ selfie camera uses subject tracking to automatically keep you in the frame, and also lets you switch between portrait and landscape modes, depending on how many subjects are in frame. I've used that particular feature a lot, so it's definitely not a gimmick.

Verdict

There’s no doubt that both of these phones are among the best camera phones money can buy in 2026, but they do differ on several fronts.

For everyday shots in good lighting, both the iPhone 18 Pro and Galaxy S26 Ultra — in their respective default shooting modes — capture plenty of detail, but you’ll get more natural-looking shots from the former, and more vibrant, stylized shots from the latter.

In low light, the iPhone takes slightly brighter pictures and, as a result, sometimes captures more of the scene than the Galaxy, but Samsung’s style-focused color science can similarly lead to better results than the iPhone. It all depends on what you’re shooting and which aesthetic you prefer.

When it comes to zoom, both phones perform well, though the Galaxy will get you closer to your subject than the iPhone without sacrificing detail (5x optical zoom vs 4x optical zoom). You also get an additional 3x optical zoom option on the Samsung, which you don’t get on the iPhone. If I wanted the most versatile zoom shooter, I’d choose the Samsung.

For selfies, though, the iPhone is the clear winner, with its larger 18MP sensor and genuinely useful framing tricks.

As mentioned, the iPhone 18 Pro also boasts a variable aperture feature. Since I wanted to compare both of these phones using their out-of-the-box shooting settings, I didn’t manually control the aperture on the iPhone in any of the above shots, but Apple does give you the option to do so, whereas Samsung does not.

Changing the aperture is useful for tweaking how much of a given background you want to keep in focus (as my colleague Lance demonstrates in a separate piece), and you can also achieve some nice lens flare effects in low light while doing so. But this is very much a pro-level feature that most users won’t play around with too much.

Overall, then, I’d say the Galaxy S26 Ultra is my preferred camera phone for zoom photography and stylized shots, but the iPhone 18 Pro is the more consistent wide shooter in both well-lit and low-light scenarios. It’s got a better selfie camera, too, and since Apple’s new pro-level controls unlock more possibilities for the iPhone, it’s the phone I’ll be sticking with until the next wave of camera-focused Chinese flagships rolls around.

Incidentally, I'll be publishing separate Portrait mode and video comparisons shortly, so stick with TechRadar for more on these two top-end camera phones.

It’s 4 pm on an unusually warm first day of October as a small crowd begins forming outside the Bristol Royal Infirmary, the south-west UK city’s central hospital. Outpatients arriving for appointments begin to take notice as the demonstration against US company Palantir’s involvement with the UK’s health service starts to take shape.

Placards are raised: “No Palantir in our NHS”, “Ditch Palantir”, “Good at harvesting your NHS data”. The passing traffic starts beeping its support.

Bristol is one of the 45 locations across over 30 cities around the UK where people are taking to the streets on October 1, 2026. The biggest action rolls out in front of Palantir’s London offices, with hundreds there to protest. Here, the number is closer to a vocal 50 or so.

Former and current health workers, campaigners from Amnesty International, the Good Law Project and more, and concerned citizens have all joined the International Day against Palantir to ask the British government to stop paying a US firm to manage UK citizen health data.

Headquartered in Miami, Palantir Technologies is an analytics firm specialising in AI-powered tools designed to manage large datasets. Its flagship software platforms, Gotham and Foundry, are widely used by government organizations and businesses globally to uncover data patterns and support decision-making operations.

The promise of its products certainly makes sense. Its list of clients, however, is what today’s demonstrators find less than palatable.

Among its most controversial partners is the US Immigration and Customs Enforcement, better known as ICE, which has been using Palantir tech for over a decade. More recently, the Israeli military has also begun using its powerful software to assist with on-the-ground operations.

It was in 2023 that Palantir managed to secure a £330 million contract within the UK public health sector. The then-Prime Minister, Rishi Sunak, tasked the American company with building a new platform to manage all UK citizen health data across the country in what was called the NHS Federated Data Platform (FDP).

Three years and two Prime Ministers later, the project is under increasing scrutiny, with some suggestions from insiders that many of the hospitals that have been signed up to Palantir are still yet to even log into the system.

With more than one reason for protester complaint on the table, TechRadar went along to the Bristol arm of the International Day of Action to find out what it was that the demonstrators cared about the most — their data, their finances or their politics?

Palantir’s NHS Federated Data Platform (FDP): the privacy concerns

A national programme led by NHS England, the Federated Data Platform (FDP) is designed to help NHS organisations bring together information from hundreds of different databases in a single platform. The goal is undoubtedly attractive: improve efficiency by supporting health worker delivery of patient care, improving operational performance, and assisting staff to make better-informed decisions.

According to the NHS security and privacy page, “the protection of people’s data is at the heart” of FDP’s design, with measures like strong network security, purpose-based access to data, and encryption all in place. It certainly sounds well conceived and considered.

Furthermore, under the terms of the contract with the NHS, Palantir Technologies is named strictly as a "data processor". That means it cannot decide how the information that its software processes is used. In other words, Palantir could face direct liability should it fail to secure this data or violate NHS instructions.

A Palantir spokesperson explained to TechRadar that FDP is a software that helps the NHS better manage the data it already holds. “How that software is used is controlled by the NHS Trusts who use it, with data — legally and contractually — only able to be processed strictly in accordance with their instructions, and all data remaining in the UK,” they added.

Despite these rules of engagement, all of the protesters expressed their lack of trust in Palantir when it came to handling their most sensitive data. They were not convinced that Palantir was not able to access their information at will.

“We simply cannot believe this won’t happen,” one protester told us.

These concerns do not exist in a vacuum. In May, the Financial Times reported on an internal NHS briefing note which explicitly earmarked "unlimited access" to identifiable patient data for certain contractor admin roles.

Most protesters also expressed their fear that Palantir’s NHS system could also be repurposed one day for wider surveillance across different government bodies, such as immigration.

Matt, a health worker and campaigner in Bristol, told TechRadar that there’s both an emotional and technical component to his worries about Palantir.

“All Palantir software is interoperable with each other. So, the data that is collected under the FDP can be dragged and dropped to their policing software if a government chose to do that,” he said.

Matt refers to what came out of a January investigation by The Nerve that found 34 active contracts across at least 10 UK government departments, as well as local councils and police authorities. A scenario that, according to opponents, could make it technically easier for a cross-pollination of healthcare data and law enforcement work to occur.

Again, these fears aren’t entirely unfounded. While Palantir’s UK CEO, Louis Mosley, insists that Palantir has “no interest” in UK patient data, the company’s track record outside the country seems to tell a different story.

404Media revealed that, in the US, citizen health data is fed to a Palantir-developed app called ELITE, which is used by ICE agents to map and target suspected illegal immigrants.

Small wonder, then, why Matt and others outside the Bristol Royal Infirmary are expressing their concerns.

Is Palantir’s technology worth the risk?

Make no mistake, though. This was mostly a protest of politics rather than privacy. Most people at the rally were here to object to Palantir's deal with the NHS based on the company’s involvement in ongoing global conflicts and immigration raids.

Either way, though, as the Head of Tech and Data at Good Law Project (one of the campaigns behind the UK actions), Duncan McCann, told TechRadar, procurement decisions based on ethics and morals are rare, whatever field of concern they lie in. The hard facts of cost and utility most usually win the day.

And it's because of this, McCann explains, that the tide is turning against the US firm.

“When we started working on this campaign a year ago, all the momentum was with Palantir. They were getting new contracts all the time. The assumption was that this contract was just going to get renewed. It wasn't even really a question. Now, when I look around, I see them losing contracts,” said McCann, mentioning the Homes for Ukraine as the most telling example.

And, for McCann, the reason for Palantir’s decline is simpler than the politics. Its technology isn’t just more expensive — Ukraine said the government saved millions by replacing Palantir’s software — but it seems to be actually failing in its goal of making the NHS more efficient.

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.

While the UK government claims to have already seen significant improvements in its data management flow, there have been multiple reports alleging that the FDP software is not delivering on that promise.

A report by Democracy for Sale, for example, says that staff have found that Palantir's newly built platform is 10 times slower than the existing system, citing an internal NHS document.

Palantir has strongly rejected these allegations, however. In a statement to Democracy for Sale, the US firm argued that "comparing the two systems is not like-for-like", adding that its platform "introduces guardrails to improve data correctness, enforce purpose-based access and allow more controlled data sharing".

Approached by TechRadar, a Palantir spokesperson said:

“Palantir software is helping to deliver better care, with Trusts using it recording 117,000 additional operations, a 14.3 per cent reduction in discharge delays for long stay patients and a 5.6 per cent improvement in the number of people finding out whether or not they have cancer within 28 days.”

Impressive-sounding stats, but back in June 2026, the Financial Times reported that England’s health service acknowledged that its main claims about the benefits of the FDP aren’t backed up by robust evidence, with MPs now calling for a new audit.

The jury, it seems, is still out on the Palantir systems’ efficacy, leaving the company’s reputation a peg or two lower than where it stood when the 2023 UK government originally signed the deal.

How likely is it for Burnham to scrap the deal?

As the deadline to renew the deal approaches, campaigners feel that now is their best shot at making the government listen to their concerns. Indeed, this may be more likely than some people realise, given the new Prime Minister’s position on digital policy.

During Andy Burnham’s almost decade as Mayor of Greater Manchester, neither municipal nor public health departments awarded any contracts to Palantir. Instead, Manchester’s NHS leaders spent six years building their own analytics infrastructure.

For campaigners against Palantir’s involvement in the NHS, then, scrapping the deal could be a good fit for Burnham’s political agenda.

"It proves that we can do it ourselves and save money by doing that, while creating new jobs,” McCann from the Good Law Project told TechRadar.

What’s certain is that the new Prime Minister is under increasing political pressure to act.

During this week's Labour Party conference, 20,000 party members backed a motion to cancel the contract with the US firm. Although the motion was not chosen for debate, the number of signatories still signals the discontent among the political rows.

A great majority of Labour voters (76%) also believe Palantir’s contract should not be renewed when it expires in February 2027. That echoes the views of the Science, Innovation and Technology Committee that, in June, urged the government to exercise the 2027 break clause and “either develop an in-house replacement or seek an alternative UK provider”.

That said, as Wired reported, there are plenty of politicians and advisers close to Burnham who are also quietly advocating for the company’s work in the UK. The recent announcement that former Labour Deputy Leader, Tom Watson, is joining Palantir as senior vice president perfectly paints the rifts within the governing Party.

For the people filling the streets, however, it isn’t negotiable. Ultimately, as health worker and campaigner Matt points out, the change could, and should, start locally.

"Here, in Bristol," he says, "we're really trying to build up the pressure and give the message that NHS Trusts actually do have the right to refuse to use the software. We really want them to take a strong stance on that."

What are the Palantir protesters asking for

By Wednesday, September 30, the day before the demonstrations, 44,000 people had filed legal objections to their health data being shared with, stored in, or used by the FDP system operated using the US company’s AI technology. Thousands of people have also signed petitions to ask to scrap the £330 million deal in the past months.

Currently, patients cannot opt out of having their sensitive healthcare data stored on the FDP, despite a 2023 YouGov poll finding that nearly half of all UK adults would likely opt out of data sharing if a private company like Palantir began processing their personal information.

“I think when people are aware of it, they're very concerned,” health worker Matt told TechRadar, pointing out the number of people turning up to the rally. “It really shows the strength of the public feeling about this.”

The erosion of trust in the NHS was a recurrent topic when we spoke to those opposing Palantir in Bristol. Many told TechRadar that they would opt out if they could.

And while people admitted that a better system to manage health data may be needed, they didn’t feel that Palantir should be the answer.

“We do need platforms that can speak to each other, but we need ones that we can trust,” another protester told us.

Most people said that they would feel more comfortable if a UK company managed their data instead. They said a domestic firm would be easier to oversee and hold to account.

Economics was also a big concern.

"Why can't these jobs be kept in Britain for British companies to run these systems? Why do we need to have a foreign company coming in and have dangerous motives to actually do this work? It's not necessary," another protester told us.

As we left the rally, the placards still waiving in the distance and the hospital entrance regaining its usual look, it was that word 'trust' that echoed in our mind.

It's almost impossible to separate the issues of data privacy from global politics, when it comes to Palantir. It's tangled up in every line of code.

Demonstrators in Bristol weren't rallying around a single, isolated grievance. They were there to resist a model where healthcare, data, and defense contracts bleed into one another.

At its very core, the entire issue around people’s health data privacy hinges on public trust.

Whether driven by fear of corporate overreach, frustration at the use of government funds, or deeper ethical objections, the battle over Palantir goes beyond digital security – the public is demanding assurances about who holds their most sensitive information.

And, for those taking to the streets today, the US tech giant had failed to earn their trust.

New details about Grand Theft Auto 6 have emerged ahead of the game's launch, courtesy of an exclusive cover story by Game Informer, confirming even more gameplay features and immersive systems that will bring Leonida and Vice City to life.

There's a lot to unpack, from an expansive open world to an array of side activities to engage with, and the ability to post on social media and doomscroll, but I've put together the top five things that stood out.

1. A map bigger and better than ever before

GTA 6's world, which consists of six primary regions including Vice City, the Leonida Keys, Grassrivers, Port Gellhorn, Ambrosia, and Mount Kalaha National Park, all culminates in a massive open world that Rockstar Games has ever created "by some degree," according to Rockstar North co-studio head Aaron Garbut.

Essentially, it's twice as big as GTA 5's world, and is more dense, populated, and interactive.

"This new world is different," Garbut said. "We obsessed about every part – the rooftops, the alleys, the internals and everything in between to the point we hit a critical mass with the numbers of our interiors, and you start to feel the buildings are more real inside and out. This world is a cohesive whole with the density of content and a hand-crafted quality to every bit."

2. An improved weather system

GTA 6 will bring realism to the next level and offer a brand new "re-engineered" dynamic and localized weather system, and players will experience everything from hot, sunny weather to rain showers, lightning storms, and more. Puddles will also remain on the ground after the rain passes, realistic rainbows will appear in the sky, and lighting will change depending on the situation.

"We have re-engineered our weather system to make everything much more localized and persistent, and really wanted the world to feel dynamically solid and achieve a sense of scale we haven't felt before," said Rockstar's VP of art for lighting and rendering, Owen Shepherd.

"The world features new fully dynamic cloudscapes that roll across the map; these give us a broad spectrum of weather from the clear blue skies of a Miami winter to the vertiginous cumulonimbus clouds with sporadic lighting that you see most afternoons in the Floridian summer and autumn."

3. Sooo many animals

Similar to Red Dead Redemption 2, GTA 6 will feature a wide variety of animals, including dogs you can pet, with Rockstar confirming 170 animal species in total.

Many of these are Leonida-native species that have never appeared in a Rockstar Games title, including over 60 bird types, more than 30 land mammals, various fish, reptiles, and more.

Rockstar vice president of dynamic art Michael Kane revealed that the game will feature creatures like flamingos, pelicans, and alligators; these are just a few of what players will encounter.

"Florida has a diverse animal ecosystem that we wanted to faithfully include across the different biomes in Leonida – urban and rural environs, aquatic life both inland and ocean, and the wetland habitats like the swamps and Everglades," Kane said. "The animal art and population teams are incredibly passionate about delivering animals that are accurately researched and look visually correct in every respect, alongside the supporting movement and animation to bring them to life."

4. More authentic NPCs and a passion for fashion

As we've seen from the trailers and gameplay presentation, GTA 6 will have a ridiculous number of handcrafted non-player characters (NPCs)— with over 600,000 animations combined—and will bring a new sense of realism to Vice City.

Rockstar has previously confirmed that the studio established a new Los Angeles-based team of writers, directors, and producers to focus solely on realizing non-player characters; according to Kane in Game Informer's interview, it was a "huge artist-driven effort" to make them "authentic and representative in service to both the story requirements and so that the entire Leonida map feels visually diverse and believable."

NPCs will now have different heights, skin and body types, individual features, with Kane adding that Rockstar "paid a crazy amount of attention" to making culturally correct tattoo styles that fit Leonida's diverse communities and regions, with hair and fashion also being heavily considered.

"We’ve been able to really go hard on that with creating the biggest range of garments and accessories we’ve ever made – supported with rich, realistic materials," Kane said.

"That surfacing includes color, patterns and branding that are both area-appropriate, but also tied to context (states of dirt/weathering and fitment) for the characters themselves and their social situation and how they inhabit Leonida. Shoes, sneakers, and colorways that we know carry cultural weight as items – likewise watches, sunglasses, or hats."

5. Robberies and the wanted system

As we've seen in the gameplay footage, players can take control of either Lucia or Jason. This applies across the game, including for heists and robberies.

However, alongside a six-star Wanted system, Rockstar has introduced a new "Heat" system that means cops can gather information on the co-protagonist's appearance, including their clothes, hairstyle, and the vehicle they're driving.

Changing Lucia and Jason's appearance and ditching their vehicle can help players lose cops, but separating the pair can give them a better advantage by splitting the cops' attention.

However, getting arrested will result in a "Busted" screen even if the other protagonist gets away, so players need to think carefully about their escape plans.

The data center boom powering a lot of AI infrastructure has provoked pushback from a wide array of groups critical of how they are built and run. The backlash has been strong enough that even in Texas, which had become a hub of new data centers, Governor Greg Abbott felt the need to order an audit of those seeking connections to the state’s electricity grid and to tell the Texas Commission on Environmental Quality to stop issuing permits sought for them until the audit is complete.

But a rather large complication is making that audit seem inadequate. Thousands of diesel generators and increasingly large private power plants sit next to the data centers. Some developers are building or planning enough power generation on-site that the distinction between a data center and an industrial power facility starts getting fuzzy.

It's a problem raised in an open letter sent to Abbott by the NAACP Texas State Conference and the NAACP Center for Environmental and Climate Justice. The groups argue that Texas’s new measures remain too focused on data centers seeking connections to the Texas grid, potentially leaving a giant loophole for those that generate or source their own electricity. The letter also argues that the audit does not sufficiently address cumulative effects, including pollution, water consumption, and noise.

As the letter points out, a Floodlight and The Texas Tribune investigation found that at least 38 Texas data centers have obtained minor-source permits for on-site power since 2024, covering more than 2,100 diesel generators. The permitting category is also associated with considerably smaller sources of pollution, including dry cleaners, auto body shops, concrete batch plants, and rock crushers. Texas’s new regulatory response may not actually address the issues brought by the data centers and their support infrastructure.

Minor permit, major story

The "dry cleaner permit" involves categories of minor-source air permits made to streamline approval for small businesses and avoid more intensive environmental review. The idea is that regulators don't need an elaborate approval process every time someone installs a set of dry cleaning machines or pumps at a gas station.

AI data centers test the logic through sheer scale. Floodlight found that OpenAI’s Stargate campus in Abilene had already built 10 gas-powered turbines and 62 backup diesel generators and was seeking approval for another 41 turbines and 18 generators. The existing turbines and generators were initially authorized through minor permits that did not require the environmental studies, public notice, and public comment associated with major permits.

The investigation also found several projects with estimated emissions just shy of triggering more scrutiny. A CyrusOne data center in Whitney estimated 249.1 tons of nitrogen oxide emissions annually, where 250 tons would have triggered more stringent permitting, while a Vantage facility outside San Antonio received permission for 99.8 tons annually in an area with a 100-ton threshold.

More than half of the data centers Floodlight identified submitted annual nitrogen oxide estimates just below thresholds requiring public input and more detailed environmental review. It raises questions of whether the regulatory thresholds make sense if they can be evaded so easily. The NAACP letter argues that minor-source permits can effectively remove communities from the process when they shouldn't be. It has already issued a notice of intent to sue Vantage Data Centers and VoltaGrid. While the allegations are still pending litigation, they show why the definition of a "minor" source has become so fraught.

Crackdown with loopholes

Abbott's recent responses acknowledge that Texas has a problem worth addressing. His audit directive states that projects that fail can be denied connection to the Texas grid. And the pause on more data centers until the audit is done will certainly slow things down for a time.

But while encouraging data centers to avoid burdening the grid might seem like a way to rein them in, letting them have their own power plants requires a different regulatory response. There are still emissions, water demands, and neighborhood effects regardless of whether it's linked to the grid. Abbot did tell regulators to make permitting decisions with the state's broader data-center directives. The loophole may be smaller than it first seems.

Still, this is becoming a recurring theme of the AI infrastructure boom. Technology is moving at a scale and speed that regulations written for a very different industrial landscape struggle to accommodate. A rule that seems perfectly sensible for a shop with a boiler or a building with a couple of emergency generators looks rather different when applied dozens of times across facilities consuming enormous amounts of power.

Texas has recognized that the AI data-center rush needs guardrails. The next test is whether those guardrails follow the actual environmental impact of the projects rather than simply the route their power cable takes.

As businesses everywhere continue to get to grips with the reality of AI-powered workflows, Snowflake has revealed its view of what it will take for a company to become a truly agentic enterprise.

Speaking at the company’s Snowflake World Tour London event, senior leaders laid out how a number of factors will all contribute to agentic success.

The 'Agentic Enterprise' is here

“There's no enterprise AI strategy without a data strategy,” declared James Hall, Snowflake Country Manager, UK&I as he kicked off the keynote.

“You need the right data foundation, trusted, governed, secure and accessible…in just 12 months, we’ve moved from talking about the potential of AI to putting AI to work…it’s now all about making AI real for business as we enter the era of the agentic enterprise.”

Hall highlighted the success of UK&I customers such as mobile operator GiffGaff, with Snowflake CEO Sridhar Ramaswamy also later joined on stage by the CEO of the London Stock Exchange Group (LSEG) as proof of this.

But it was Christian Kleinerman, EVP Product Management, who took to the stage to run through the factors the company sees as vital for an agentic enterprise.

“We are all part of the era of the agentic enterprise,” he declared, defining this term as an organization that has been able to embed AI powered agents into its core operating processes, co-ordinating with humans, setting directions, setting guardrails and automating as much as possible - all with the ultimate goal of help workers be more productive, and creating better business outcomes.

“The promise of AI is unparalleled, in my mind, to other technologies that we've seen in our lifetime,” Kleinerman added, addressing a growing myth around the technology, namely that some organizations believe AI is so good and smart, the data will sort itself out.

“But if anything, the opposite is true,” Kleinerman said the most amazing results you can envisage with AI come from having a great data strategy.”

Kleinerman noted three main elements for an enterprise to be truly agentic - an AI model, data context, and software and apps - all feeding into what Snowflake calls an “agentic control plane”.

When it comes to picking a model, Kleinerman noted the company is “committed to model choice for all of you” - as there is clearly no way to know who will win out in AI models even within the next few months.

Data context is becoming increasingly crucial, he added, as understanding your business matters a lot to how AI can be beneficial for you, and having the right software and apps will really allow AI to integrate with the rest of your enterprise infrastructure.

“The truly amazing results come when you really understand your data”, he added, “we want to un-silo your data, we want to help you govern your data...so humans can be more data-driven - AI has supercharged that ability.”

All of this flows into the agentic control plane - “the glue that holds all this together”, Kleinerman concluded, “the agentic enterprise truly comes to life when the AI models can interoperate with the rest of your enterprise infrastructure…(and) hopefully it feels like an avalanche of innovation...our goal is to continue accelerating.”

“We’re operating in a fast-moving environment, arguably the biggest risk any business could take right now is by standing still - but moving fast with data and AI shouldn’t mean losing control, it means having the foundation, the flexibility and the confidence to move quickly,” Hill added.

“The era of the agentic enterprise isn't coming, it's here - it’s happening right now in production.”

If you work in an office you’ll know we’re deep into performance review season. But how would you feel if you found out that your boss had used AI to write your yearly appraisal? I know I’d be pretty annoyed to learn that AI was deciding whether I was worthy of a raise this year.

AI use by employees is becoming commonplace, but it's also being adopted by managers. A recent survey by Headway of 1,000 managers found that 33% had used AI to write feedback or prepare a performance review, while 31% had turned to AI to resolve team conflicts. A further 27% had used AI to help write uncomfortable conversations they needed to have.

That may sound like a lot, but the figure is actually down a bit from 41% admitting they have used AI to draft or revise a performance review in a similar survey from a year ago, but still high.

The ease AI

For managers, the appeal of using AI is easy to understand. Faced with having to write several performance reviews each year it can be a real timesaver to dump the work onto AI, however the risk is that what should be very personal performance feedback is being outsourced to a bot.

Managers' AI use is not stopping at performance reviews, either. The survey also found that 57% of managers have already scripted or rehearsed a difficult conversation before having it using AI.

Thalia-Maria Tourikis, a qualified health coach and burnout prevention and recovery expert at Headway, commented:

“When encountered with a difficult conversation, one might think that using AI can help organize their thoughts. However, you should not give up that interaction to automation. When human emotions such as trust and empathy are involved, it is best conveyed when written by a person. The perfection of AI reads as superficial and doesn’t have the same impact.”

A sliding scale

Bosses using AI to help rate employee performance isn’t the only threat from AI that workers face, there’s also the threat of being monitored by AI. Owl Labs' State of Hybrid Work survey spoke to 8,000 workers worldwide, which found that 58% of employees believe their employer has deployed AI-based monitoring tools.

The Headway survey doesn’t make it clear whether managers are using AI to simply improving their grammar, or change the occasional word for something better, or wether their actually generating the text of the performance review entirely using AI. There’s a big difference between these two extremes and one involves much more cause for concern than the other.

It’s especially concerning with performance reviews because these are often link to financial compensation. If your performance review makes a difference to your annual salary, and wether you deserve that raise or not, then finding out that it’s really AI deciding your future rather than your human manager would be especially galling.

If you ever need a sign that AI is coming for Hollywood then it must be that Jeffrey Katzenberg, the veteran Hollywood executive who co-founded Dreamworks, has just announced that he’s launching an AI-focused studio.

Crucially, he told The Wrap that AI output was “slop” until very recently, but that in the last couple of months something has changed. Now he wants to make feature films and TV with AI, arguing that falling production costs will mean more films and new forms of storytelling.

At the same time The Gifted, an AI-generated trailer, has just won a competition that puts $2.5m towards turning it into a proper feature film. Future Vision XPRIZE is one of the world's largest film competitions for hopeful and optimistic science fiction filmmaking that answers the question: what does a future worth building actually look like?

You can watch the trailer for The Gifted here.

Unfortunately, the industry focus on AI seems to be happening at exactly the same time that the audience is being turned off by it.

Filmustage, a movie production planning platform, analyzed Google Trends data and found searches rising on both sides of the AI debate. It found that the phrase “movies with no AI” was up 345% in the past month, while “movies made with AI” was up 112% in the same period.

Google Trends percentage increases can often be massive, so we shouldn't read too much into the numbers themselves, but the overall trend is upwards. The motivations behind those searches may be very different, but both suggest audiences increasingly want to know one thing before they watch: was AI involved?

I don't think it's hard to understand why. You only have to look at an AI film to understand why people have reservations.

You can judge the trailer for The Gifted yourself, but while it may be a stunning technical achievement that AI has made this three-minute video, if I’m judging it as a movie trailer… then I don’t want to watch the movie.

The flaws are obvious. AI is clearly very good at spectacle: lightning, explosions, collapsing buildings and impossible camera angles. Not to mention lead characters who look almost like famous actors you know, but not quite. (Is that Amy Adams in The Gifted's poster? No, it's not quite Amy Adams.)

The harder test is whether it can sustain believable characters, performances and emotional continuity for two hours rather than three minutes. If it does, then it's going to need much better dialogue than what's presented here.

I remember watching the pure-AI film Hell Grind in a theater recently, and that just about kept my attention for its 90-minute duration, but I was ready for it to be over by the end. It also played to AI’s strengths, by not trying to make its world look ‘real’. It's when AI pretends to be real life that I think it falls down the most.

Pure AI films... still look like AI

Actors, perhaps seeing a future in which their skills are not as widely appreciated, are famously against AI, but even they seem to have accepted the overall trend towards it now.

As Deadline reports, Tom Cruise earned spontaneous applause for a six minute speech during a recent BAFTA “Life in Pictures” conversation in London. Cruise talked about how AI “is coming”, but “people want to see real things”.

“There is nothing like human” said Cruise, citing a summer of blockbuster films as proof that “people want human experience.”

From what I've seen of AI movies, I'd say Cruise is exactly right. AI still can't produce an experience that looks human and the connection to the audience that goes along with that. But with even Tom Cruise acknowledging that AI is coming, and executives like Katzenberg jumping on board, AI movies seem almost inevitable now, especially with the saving in production costs that they bring with them.

But while AI might be economical, it definitely has its problems, and if people are already searching for “movies with no AI” then Hollywood has a branding problem before the AI movie era has even properly begun.

AI as another tool for filmmakers

But perhaps before we assume that AI in films means the end to human actors and an influx of AI slop, let’s talk to one of the AI studios making waves at the moment, Utopai Studios, who creates original films and television series using their own proprietary PAI software.

I talked to Kevin Chong, Chief Operating Officer at Utopai Studios.

“We don’t think of AI films as a separate genre”, explains Chong. “AI is another tool filmmakers can use to tell stories, just as visual effects, digital cinematography and virtual production became part of the filmmaking process. It can help filmmakers create worlds, achieve a level of scale that might otherwise be difficult or prohibitively expensive, and give creators more choices.”

And what about the end of human actors? “At Utopai Studios, we don’t see AI as a replacement for human actors, and that isn’t the future we’re building toward”, Chong explains. “There may be stories where a completely synthetic character makes sense creatively, just as animation has always created characters that don’t physically exist. But that is very different from saying audiences are going to stop wanting human performances.

“Performance is one of the most human parts of filmmaking”, Chong continues. “Actors bring interpretation, emotion, instinct and life experience to a character in ways that are fundamental to storytelling. AI can become another part of the filmmaking toolkit — including helping extend performances, create environments or enable things that would be difficult to achieve traditionally — but the creative decisions still need to come from people.”

Tom Cruise is right: there is nothing quite like human, but perhaps there will be a way for humans and AI to coexist in the future of movies after all.

October 1 is International Coffee Day, and if you've been thinking of upgrading your coffee maker, it's the perfect opportunity. As TechRadar's resident coffee expert, I've put together a list of my three favorite bean-to-cup espresso machines that make preparing perfectly extracted espresso a breeze, even if you've never used fresh coffee before.

There's something for everyone here. If you're on a tight budget, or are making the leap from coffee capsules to whole beans, the compact and affordable Philips Baristina is the perfect choice for you. Interested in tinkering with brew options and understanding how different factors affect the taste of your drink? Take a look at the Sage Barista Impress With Cold Extraction. Want a fully automated machine that produces even complicated drinks with the tap of a button? The Philips Café Aromis is the coffee maker of your dreams.

Many retailers are offering special discounts for International Coffee Day too, and you'll find the best deals on each coffee machine below as well.

Philips Baristina

The Philips Baristina is the most affordable bean-to-cup coffee maker I've ever tested, and easily one of the best. In fact, I recommend it so frequently, my colleagues are probably quite tired of hearing about it. It makes brewing espresso with fresh coffee beans just as simple as using Nespresso capsules or K-Cups, and opens up a whole world of speciality coffee without requiring any specialist skills.

Load your coffee beans into the top of the Baristina, fill the water tank, and you're ready to get started. Push the filter handle into the machine, hit the start button, and it will automatically grind just the right amount of coffee, tamp it down, and then slide the handle underneath the brewing group, which will extract your shot using the ideal water temperature and pressure. When it's finished, just tap out the used coffee grounds, wipe out the filter, and you're ready to go again.

Unlike most bean-to-cup coffee machines, the Baristina doesn't have a case that covers everything up, which helps keep everything clean and hygienic. There's no enclosed space for stray coffee grounds and moisture to build up, so maintenance is a breeze.

The biggest downside of the original Baristina is the lack of a steam wand, which means there's no way to prepare lattes and cappuccinos. However, the machine is available bundled with a good quality standalone milk frother for a small additional cost, and earlier this year Philips launched the Baristina Latte, which has an integrated steam wand.

Read our full Philips Baristina review

Philips Café Aromis 8000 Series

At the other end of the scale, we have the Philips Café Aromis — the company's flagship bean-to-cup coffee machine with a huge array of hot and cold drinks options, excellent milk frothing, and app compatibility that makes customizing your brews a breeze. It's gorgeous as well, with its sleek silver chassis, crisp touchscreen, and a natural wood cap on top of the dispenser to complete the look.

Drinks are almost infinitely customizable, and the cold menu is a particular highlight. Unlike some bean-to-cup coffee makers I've tested, which just dump a shot of hot espresso over ice, the Aromis can extract coffee using water straight from its tank for a mellow flavor akin to traditionally made cold-brew. It also has separate milk carafes for frothing hot and cold milk, and each one can be disassembled into three dishwasher-safe parts for easy cleaning. So, there's no need for scrubbing milk residue from pipes and hoses.

It's not the only bean-to-cup coffee machine with its own app, but it's the only one I've used where the app offers something useful apart from a quick-start guide and warranty details. The Philips HomeID app makes it easy to create personalized user profiles, and even had a clever chatbot-style feature that asks you questions about your flavor preferences and then tweaks the coffee maker's settings accordingly.

The Aromis is definitely one of the more expensive espresso machines I've reviewed this year, but its price tag isn't excessive considering just how much you're getting, and you could easily pay half as much again for an alternative from a different brand.

Read our full Philips Café Aromis 8000 Series review

Sage Barista Touch Impress With Cold Extraction

This bean-to-cup machine (which is sold under the brand name Sage in the UK, and Breville elsewhere for complicated reasons) is an upgraded version of the original Barista Touch Impress, and like the Philips Café Aromis, it offers genuine cold extraction for smooth-tasting iced coffees (as well as delicious hot-brewed espresso).

Unlike the other two bean-to-cup machines in this list, this one is semi-automatic, which means that although you don't have to get your hands dirty, it gives you some control over the coffee-making process if you want, introducing you to barista skills in a friendly, easy-to-follow way. You'll discover the effect of each variable on your finished drink (including dosage, grind size, tamp pressure, and brewing temperature), and the machine will offer feedback and suggestions if it doesn't taste quite right.

The Puck Impress System, which tamps your freshly ground coffee with a pull of a handle, is particularly satisfying and greatly reduces mess compared to tamping by hand. Unlike the Breville Oracle Jet below, I didn't have any issues with static electricity making coffee cling to the grinder.

The Barista Touch Impress offers both manual and automatic milk steaming options too, so you can have your foam dispensed into your finished drink, or prepare it yourself and have fun practicing your latte art.

Read our full Sage Barista Touch Impress With Cold Extraction review

Coffee capsules are incredibly quick and convenient, but they have one big drawback: the used pods left over at the end.

Some capsules are relatively easy to recycle — Nespresso pods, for example, are made entirely from aluminum, and Nestlé has well-established recycling program to handle them, but other brands use capsules made from a combination of metal and plastic, which is much more difficult to process. Whatever the material, though, there is still a considerable amount of waste, which takes resources to recycle, and some of which will inevitably end up in landfill.

Using ground coffee or whole beans is a much less wasteful way to prepare an espresso, but it also takes longer, and although it's a fun hobby for some of us, not everyone is interested in getting into the technicalities of brewing a balanced shot.

That's where the Lavazza Tabli system comes in, using tablets made entirely of freshly ground coffee. Each one gives you exactly the right amount of coffee for a single serving, so there's no weighing required, and there are no loose grounds to make a mess of your counter. The whole process is just as clean and simple as using coffee pods, but without the empty capsules to worry about at the end. It seems like such a simple idea, it seems surprising it's not hit the mainstream before — but how does it compare with other brewing methods?

The Tabli coffee maker launched in the US in July, and is now available in the UK as well. It's one of the most affordable coffee makers I've tested this year, with a list price of $139.99 in the US and £199 in the UK (though it's currently discounted to £133 at John Lewis.

It's an unusual-looking machine, as you can see in the photo above, and relatively large for a single-serve espresso machine. It measures 12.9 x 6.5 x 15.4 inches / 327 x 165 x 390mm (H x W x D) with a 1.3 quart / 1.2 liter water tank on the back, and comes in a choice of three colors: black, white, and walnut brown.

The interface is as simple as that of any capsule coffee maker. There are three soft-touch buttons on top for selecting your drink size (single, double, or lungo) and the upper section slides open smoothly to reveal the brewing basket for inserting your coffee tablet.

The Tabli machine has a removeable drip tray, so you can use it with espresso cups, or larger mugs if you'd prefer to use your freshly brewed shot as the foundation for a longer drink like a cappuccino. However, if that's how you like your coffee, you should be aware that the Tabli doesn't include a steam wand, so you'll need to purchase a standalone milk frother like the Smeg Mini Milk Frother or Dreo BaristaMaker.

Used coffee tabs drop into a container behind the drip tray, just like used grounds would in a conventional bean-to-cup coffee maker. The grounds are dry to the touch, and are suspended in a basket so you can easily tip them into a food waste caddy, then pour away any water left behind after the Tabli rinses itself (there's no need to perform a rinse cycle yourself)

Tabli coffee tabs come in card boxes, and are held in molded paper trays, much like egg cartons. There is a thin plastic film over the top of the tray, but it's the bare minimum necessary to keep out moisture. Everything else is easily recyclable in regular household waste.

Some capsule coffee makers are cheap to buy, but end up proving surprisingly expensive in the long run due to the price of the pods. Bosch Tassimo machines, for example, are very affordable, but the pods are some of the priciest around.

Thankfully that's not the case with the Tabli system. A box of 12 tabs costs £5.89 in the UK, which is about 49p per drink, and very reasonable. You can also pick up a multipack of eight boxes for £40, which brings the cost per drink down to 42p.

Tabs are given a strength score on a scale of 1-13, and although there are currently only six types available, if the system proves successful I'm hopeful that Lavazza will increase its range.

There aren't any of the elaborate flavors you get with Nespresso or Tassimo pods; all Tabli tabs are made using nothing but compressed coffee, with no additives, and in my opinion that's no bad thing.

I've tried two Tabli varieties so far (Qualità Rossa and Qualità Oro), and both produced a smooth-tasting, easy-drinking espresso without the astringent taste I associate with most coffee tabs. I'll be interested to see how they compare to the Espresso Ristretto, which has a higher intensity score of 12/13 and should have a punchier flavor.

This isn't Lavazza's first experiment with shaking up home coffee-making. Last year the company released the Lavazza Assoluta — a good-looking espresso machine that lets you experiment with different beans without getting into the tricky business of optimizing grind size and temperature. Install Lavazza's mobile app on your phone, scan the barcode on your bag of coffee, and the Assoluta will adjust its own brew settings accordingly.

It was an interesting idea, but so far the Tabli strikes me as a better concept, and one that doesn't require any extra work on the user's part. I'll continue to put the machine through its paces over the coming weeks, and will bring you a full review very soon.

Your email platform can now write a newsletter and pick the moment it lands in someone’s inbox. If you're a marketer running campaigns with a thin team and a thinner budget, that can feel like a lifeline. You're in good company, too, with 40% of marketers naming expanded AI use among their top three email priorities for 2026.

But the flip side of that is your subscribers are using AI too, mostly to decide whether your email deserves their attention. Surveys suggest that 40% of US consumers may trust a retailer's emails less if they knew AI had written them. That’s why we wanted to talk about where AI earns its place in an email workflow, where it needs strong oversight, and where it just immediately backfires.

How does AI assist with writing emails?

AI in email falls into one of two camps. Generative models write and rewrite copy from a prompt, while machine learning models study your historical data to predict who is likely to engage or drift away.

Adoption is already mainstream. In the State of Email 2026 report from Litmus, 28% of marketers said AI was deeply built into their email workflows and another 34% used it regularly. Only 5% weren't using it at all, while generative tools for copy and images topped the list of most impactful uses.

The payoff is mostly speed. The same survey found the majority of marketers now produce and send campaigns within three days, with advanced adopters likely to ship an email in under a day. Some common jobs AI handles include:

• Drafting subject lines and body copy from a short brief.
• Rewriting a draft for a different tone or audience segment.
• Predicting the best send time for each subscriber based on past engagement.
• Scoring contacts by churn risk or likely lifetime value.
• Flagging invalid or risky addresses before they damage deliverability.
• Summarizing campaign performance in plain language.
• Building automated sequences triggered by subscriber behavior.

Still what AI can't do is know your customers the way your account managers do. It predicts plausible text and likely outcomes from patterns, which is a different skill from understanding why a loyal client suddenly went quiet or making judgment calls on what gets your customers’ attention.

Types of AI email assistants

AI now turns up at almost every stage of an email program, from the draft window to the reporting dashboard. Some of it you'll have switched on deliberately, while some arrives quietly in a product update.

Knowing which type of tool you're dealing with tells you how much damage a bad output can do. A clumsy subject line suggestion is easy to ignore, whereas an automation agent sending the wrong offer to your entire list is a much bigger problem.

AI writing assistants

Generative writers now come standard in most email marketing platforms, sitting inside the editor to draft or rewrite copy on request. In fact, GenAI tools for copy and images are often the most impactful AI use case for marketers overall.

The same models also live inside everyday inbox and office apps, where they suggest replies that mimic your tone. Those assistants suit one-to-one sales follow-ups better than bulk campaigns, which need a consistent brand voice across thousands of sends.

List management and hygiene tools

This is less glamorous work, but it decides who receives your email at all. Email verification services increasingly use AI scoring to judge how likely an address is to engage, which helps with catch-all domains that accept every message and can't be confirmed through standard checks.

That matters a lot more now than it used to. Google tells bulk senders to keep user-reported spam rates below 0.1% and never let them reach 0.3%, so a stale list can sink your deliverability long before your copy gets a chance.

Scheduling and automation platforms

Send-time optimization is one of the oldest AI features in email marketing. Platforms that offer it study each subscriber's past opens and clicks to predict when they're most likely to engage, then stagger delivery to match.

Automation is where things are moving faster, though, with newer platforms bundling bots and agents that build multi-step sequences from a single goal and suggest changes based on performance. Marketers expect that shift to continue, as 44% predict more than half of their email operations will be AI-driven by the end of 2026.

Analytics and predictive insight tools

Predictive analytics turns your send history into forward-looking scores, such as each customer's likely lifetime value or churn risk. Those scores can feed segments that update automatically as behavior changes.

Reporting assistants are the newer addition, explaining in plain language which campaigns are driving results. Analyzing campaign performance to identify trends ranked as the most impactful AI use for 15% of marketers in the same Litmus survey.

Why your emails shouldn't rely on AI alone

Readers can often tell when nobody was home while an email was being written. For example, Gartner reports that 49% of US consumers believe generative AI has made the quality of content worse.

The penalty grows with the relationship at stake. In a study of more than 1,100 US professionals, researchers from the University of Florida and USC found that over 80% saw supervisors as sincere when they used AI for light edits. That figure fell to between 40% and 52% for heavily AI-assisted messages, which should give any founder or account manager pause before automating a personal note to a client.

Accuracy is a separate headache, because an AI model will happily write "Final hours!" for a sale that actually runs until Sunday. Washington state's email law saw nearly 200 class actions over subject lines like that after a 2025 court ruling. A 2026 amendment now requires proof that the sender knew a subject line was misleading, but saying "the AI wrote it" isn't a defense that’s likely to work.

Inbox AI also changes who reads your email first. The majority of consumers now make decisions based on AI-generated summaries alone, so vague copy can easily get smushed into nothing before a person ever sees it. Ultimately, the safest move is to put your main point where no algorithm can miss it.

5 tips for writing better emails with AI

You don’t have to ban AI from your email workflow, but you need to decide which tasks are worth handing off and which tasks need human review, then build those checkpoints into the process.

1. Start every draft with a detailed brief

Generic prompts produce generic emails. Give the model your audience segment, the offer's exact terms, the call to action, and a couple of past emails that performed well so it can match your voice.

Treat the output as a first draft for an editor. We recommend rewriting the opening line and the call to action yourself, since those are where readers decide whether a message sounds like a person.

2. Fact-check every claim and deadline

Build a pre-send checklist that compares prices, discount percentages, expiry dates, and product names against your source of truth. Pay particular attention to subject lines and preview text, because those are what inbox summaries and legal complaints tend to latch onto.

Anything with urgency language deserves a second human reviewer. If there's any chance a promotion gets extended, rewrite the subject line so it stays true either way.

3. Put approval gates on automated sequences

Let the platform draft a welcome series or a win-back flow, but review every email and trigger before it goes live. Treat any AI-built sequence like a new hire's first campaign, where nothing ships until someone senior has read it. After launch, spot-check live sends each week rather than assuming the automation behaves.

Set a cap on how often any single contact can be emailed across all your automations. Overlapping AI-built flows are an easy way to creep toward Google's spam-rate ceiling without noticing.

4. Question what the analytics tell you

Before acting on an AI recommendation, ask what data it's based on. A send-time suggestion built on a few weeks of activity from a small list is a guess wearing a lab coat.

Use AI to spot patterns, then confirm them with a controlled A/B test before rebuilding your program around them. The model can find the correlation, but a human still has to decide whether it means anything.

5. Protect customer data and keep personal emails personal

Check how a vendor handles your data before pasting customer details into any AI tool. Look for a written commitment that third-party AI providers can't train their models on your prompts or subscriber data, plus a clear answer on where that data is processed.

For relationship emails, such as a thank-you to a key account, AI shouldn't do anything more than light editing. You can use AI to write the email for you, but making the reader believe that you meant it is still your job.

I was a free-range kid before the term existed. Growing up before cell phones, my parents' chief communication device was a very loud voice, and one only used after the sun set and "dinner!!" was served. Freckle Phone creator Sam Terris tells me he, too, was free-range and, as he recalls it, had a lot of sovereignty.

The loss of that independence, the ability to wander over to your friend's house, play outside on a swing set by yourself, or go exploring in the creek that runs down the hill behind your house, has, researchers claim, played a significant role in children's growing anxiety and depression. It's something 'Free-Range Kids' advocate Let It Grow President Lenore Skenazy has built an entire advocacy group around.

In her TED Talk on the subject, she said that where there were once three worlds: "the kid world filled with bikes, adventures and playing, and the adult world, so boring... then there was the family world where everyone was together, but now they've all been sort of mashed up together, and especially [so] since phones, because even when kids aren't physically with their kids, they can be texting and talking, and tracking them."

This makes some intuitive sense, and naturally, what better way to solve it than with another phone?

"It's the first phone for kids that encourages real-world adventure," says Terris when I talk to him about the cute, square, and wearable device.

There was a lot of internal argument at Freckle about whether or not to call it a phone. But in the end, Terris says, "If we don't call it a phone, all the questions that we're asking are those like, 'Can you make video calls?', 'Is there a camera?', 'Does it have table stakes, phone functionality [and so on]?' So if we can call it what it is, which is a phone, it allows us a little bit more space to talk about some of the more novel stuff that we built."

In truth, the roughly 3x3-inch x 0.61-inch thick device doesn't look or work entirely like a traditional phone.

Yes, it's a 5G device running on Freckle Mobile's MVNO. Yes, there's a phone number. And texting. And calling with parent-approved numbers. There are two cameras (12MP and 5MP), microphones, and speakers. But the resemblance kind of ends there. A skinned-over Android system offers no downloadable apps or access to the internet; instead, there are tools and AI [features] custom-designed for real-world adventure and exploration.

Freckle's platform includes Quests, Chat (AI or tough questions sent to parents), Maps, walkie-talkies, and custom characters. In some ways, it's like a kid's backpack full of fun stuff to help, as Terris told me, "facilitate those stories and adventures."

Terris thinks that current products designed to help parents keep tabs on their children are "misguided". They are, he told me, "thinly veiled ankle monitors or like parental spyware, like leashes."

Freckle Phone is different and designed to return that freedom to kids (ages 7 to 13) while maintaining a level of connection that might alleviate modern parents' concerns.

The interface is simple, fun, and has on-board AI (Freckle uses a mix of models, including Gemini and a small local model) that automatically engages and can help children understand the world they're exploring. Kids can ask a question or use the camera to take a picture of something in front of them — a plant, animal, or object — and get instant details about it.

The phone also encourages play, letting Freckle Phone owners use NFC to tap the phones and initiate the walkie-talkie mode. The phone also includes tap-to-pay in case parents loaded up the onboard wallet with money for the grocery store, fast food, or a bus ride home (if parents let their kids go that far).

At the end of their day, Freckle Phone offers a way for the child to share their adventures. Each Freckle Phone has a customizable on-screen character who's wearing a backpack.

"So in this little character's backpack are these things that we call artifacts. They're little mementos; these are generated by the tool usage. [There's] our curiosity camera, [or] if you're playing music, there's a little Music Player on here as well, even locations that you've gone to, like points of interest on your map, collaborative stories that you've told, the descriptions of those badges that you've earned."

It all sounds pretty good, and the Freckle Phone does look like fun. The square screen is a decent size, but with a thick black bezel around it that Terris insists has grown on him. The device should ship with a shoulder strap and in three colorways (green, blue, and burnt orange). In the video I saw, all the kids were wearing their Freckle Phones. It probably won't survive a drop in that creek, but the IP54 rating should handle a good rainstorm.

The phone may be about returning kids to a free-roaming past, but it's also connected to habits of the present. Freckle Phone will ship with little charms called "dots" that you can magnetically attach to the back of the phone, and there are digital cards you can swap with the aforementioned NFC capabilities.

It's also potentially priced right, starting at just $199 for a device with 4GB of RAM and 64GB of onboard, non-expandable storage. The wireless service will run you $25 a month.

Terris admits that they didn't, in this first version, invest too much in the camera specs, but "we have found that kids like the act of taking photos more so than rummaging through them later." Maybe so, but parents looking at the photos later might feel otherwise.

The more I looked at Freckle Phone, I realized that "phone" might be the wrong description of it. Freckle Phone is potentially another wearable AI device. Recent history has not been kind to the Humane and Rabbit R1s of the world. How, I asked Terris, could his startup avoid a similar fate?

"I think the issue with the 'flops from before' [a term I used] is like, they weren't solving a really clear need, and there wasn't a really obvious use case," said Terris, who bought a Rabbit R1 early on, and while he really appreciated the effort, the shortcomings were also all too clear.

"[There] wasn't an obvious place for it to plug into my life, and I think some of the software presented was not really realized, and so it was kind of like a half-baked product. I think, with Freckle, what's amazing about it is that there are the enriching experiences, through prompt activities and quests and helping kids find friends and navigate [the world], but there's also the underlying utility of communication."

The choice of including a colorful and useful screen was also important. "I think the screen is actually a great visual affordance that has really high utility." And that utility serves kids' needs, he added.

Long-term, Terris hopes to build on Freckle's foundation with a regular cadence of updates (he says they have a backlog of about 1,000 new features). Eventually, it will "kind of feel like Christmas morning for our users every couple of weeks, and they get to unlock a new feature."

Freckle Phone is set to ship in the US in December. There won't be any commercials or even a large ad spend. Instead, there'll be homegrown videos, some influencer collaborations, and maybe some updates based on feedback from one very special beta tester: Olympia Ohanian. Her father, Alexis Ohanian, is one of the company's investors (he's married to Serena Williams). "She's been a vocal beta tester, so she's really helpful," said Terris.

Is a wearable AI phone enough to set kids exploring and raise the spirits of an entire family? Only time will tell. The concept of Free Range Kids may have legs, but it may not yet be ready to run — with or without a Freckle Phone.

America.gov has a pitch that I suspect almost anyone who has dealt with the federal government will understand immediately. Instead of figuring out which department, or obscure corner of an agency, controls the thing you need, you tell one website what you are trying to accomplish. Then AI sorts out the bureaucracy for you.

The Trump administration launched the AI-powered America.gov this week as a single digital front door for federal services. Its ambitions eventually extend to passports, Medicare, Social Security, and government jobs, but the version available today starts with something much more familiar. There is a big conversational box where you can ask the government a question in ordinary language.

There is familiar technology behind it too. Google Gemini and Grok are being used. The system draws on information scattered across thousands of government websites and attempts to turn it into a useful answer with links back to official sources.

I spent some time trying it, and I can see the appeal. I am considerably less convinced that I am ready for what the government wants to build on top of it.

No learning curve

America.gov doesn't require much instruction. You type what you need in the same conversational style you would use with ChatGPT, Gemini, or Grok and let the site work out which pieces of government information are relevant. For anyone accustomed to AI chatbots, there is essentially no learning curve.

I started with the suggestion by the chatbot to ask how to replace a Social Security card. America.gov explained the basic process and directed me toward the appropriate government resources. I knew beforehand that my journey should begin with the Social Security Administration, but it was still helpful information.

I tried the same approach with passports and Medicare. Again, being able to describe the outcome I wanted was easier than figuring out the vocabulary used by the relevant agency. Tell America.gov what you are trying to accomplish rather than throwing a few keywords into the box and you'll probably be okay.

That said it's important to click the sources and not just take it on faith. America.gov can synthesize information, but generative AI adds another interpretive layer between you and the underlying government material. If I am checking a deadline or eligibility requirement, I would verify the important details on the agency page before acting on them. A hallucinated fact is annoying when an AI is helping plan dinner and considerably more consequential when it involves healthcare.

The site's first day also provided an unusually vivid example of why I don't want to treat its answers as immutable government truth. In my tests, the chatbot didn't provide any political information when I asked it questions, which is what the website originally promised.

According to an AP report, America.gov initially answered questions about subjects including who won the 2020 election, presidential term limits, and climate change using information from government sources. Then some of the answers changed. That isn't evidence of a security problem, but it raises questions of who decides what a government AI service chatbot is allowed to say. If America.gov is going to become a primary interface between Americans and their government, changes to its behavior should not disappear invisibly behind a chat box.

Gov AI privacy

The current chatbot is only the first stage. The White House says America.gov will eventually allow people to complete government transactions rather than merely tell them where to go. The executive order also calls for Login.gov integration so people can authenticate themselves and access participating services through the same portal.

Passport services and Medicare enrollment are planned for the next couple of years, as are legal name changes and federal job applications. The broader idea is that government systems should stop repeatedly asking people for information another part of the government already possesses. If implemented well, that could remove some genuinely miserable bureaucracy. But those are also the features that make me nervous.

Right now, I can use America.gov without giving it much of anything. Asking how passport renewal works is basically an information request. The promised version could sit between my authenticated identity and some of the most sensitive interactions I have with the federal government. That requires an enormous amount of trust.

The executive order creating America.gov says agencies should retain custody and control of their own records, that the system should avoid creating a centralized federal record on Americans, and that personal information should be protected.

Those are good promises. Recent history gives me reasons to want considerably more than promises. The Government Accountability Office examined access given to a Department of Government Efficiency team working with Treasury's Bureau of the Fiscal Service in 2025 and found that one DOGE employee had access to three federal payment systems and could view, copy and print data. Investigators also found that Treasury had issued the employee a laptop without first ensuring that the employee had agreed to follow security rules intended to prevent data from being transmitted outside the bureau.

And, in March, GAO reported that federal agencies do not always have the technology and resources needed to ensure privacy protections when using AI. Its panel of experts warned that AI can expose personal information contained in underlying datasets, while GAO concluded that existing government-wide guidance did not fully address the major privacy challenges it identified. That report landed only six months before the government launched an AI service it intends to expand across some of its largest public-facing programs.

None of this means Gemini or Grok is quietly reading my Social Security record when I ask America.gov a question. It does mean I want answers before America.gov evolves from helpful government concierge into authenticated government middleman.

The administration is asking Americans to eventually trust America.gov. Given the documented problems around government data access and the GAO's warning that federal AI privacy guidance remains incomplete, that trust should be earned technically rather than requested rhetorically.

Peloton is name that remains synonymous with stationary exercise equipment, despite the company's ups and downs. It experienced huge highs as Covid forced people to exercise indoors, and became a market leader in the best treadmills and exercise bikes, even if it was never quite able to recapture the zeitgeist position it once held as restrictions lifted and people returned to their gyms.

Six years on, the premium exercise equipment company shows no sign of slowing down: its slickly-produced live and recorded classes feature instructors who are treated like celebrities in their own right, and its algorithmic, addictive content meshes with its high-end machinery to make for a 'magic formula' which keeps Peloton power users addicted to indoor cardio.

Now Peloton has announced a trio of new treadmills that will launch starting October 1: the Peloton Tread Flex, the Tread Vision, and the Tread Vision Plus.

Let's start with the Peloton Tread Flex. Costing $2,195 / £1,999 / AU$3,499, it's Peloton's cheapest treadmill, and its cheapest product so far, although it won't be available in regions outside the US until spring 2027. It's designed for smaller spaces and smaller budgets, and it's also Peloton's first folding treadmill, with four caster wheels to help you tow it around in its upright configuration without needing tools.

"We've put a lot of work into that," Nick Caldwell, Peloton's Chief Product Officer, tells me when I meet him and VP of Product Management Meg Higgins at Peloton HQ in London. "It's 46% of the footprint of our Tread Vision base platform, and also super quiet... You're getting great value with the capability to do folding and flex, and you're going to get a ton of new software features with this as well."

Some of those software features include the great-looking ReRun feature, which extracts and replicates the elevation profile of real-world routes, like parts of the New York City Marathon, and replicates them on the Tread. The Tread will respond by adjusting the gradient automatically to replicate the route, and show you your progress on a virtual map. Not only is this cool in its own right, it also allows runners hoping to tackle popular races around the world an opportunity to dry-run those iconic courses.

"Out of the box, we'll have some built-in ones," says Higgins. "We'll take the New York City Marathon and break it up into five digestible chunks... so you can get dialed-in training for the course." Higgins adds that runners will be able to upload GPX files from their outdoor runs and replicate them on the treadmill, too.

Next up is the Tread Vision, a more powerful non-folding, AI-powered model starting at $3,495 / £2,999 / AU$4,999. It features a new cushioning system underneath the tread for more comfort and energy return, and a redesigned button system with jump-up and jump-down options alongside the old dial system. But the real advancement is the Movement Tracking Camera.

"We've created a new form feedback system for runners", Caldwell tells me. "This is using vertical displacement to a) count your steps, give you a cadence figure, and b) it'll tell you whether or not your form is correct, and whether your running efficiency is being maximized".

Covering stats like vertical oscillation (how much you bounce while you run) as well as cadence, the system will give you a Running Efficiency Score, as well as personalized AI-powered feedback. We've seen similar features in products like the Peloton Guide, and incorporated into earlier Cross Training labelled devices, but here Peloton has applied it to a treadmill.

"We've already seen this improved folks who are trying to do marathon prep, and things like that," adds Caldwell, indicating that it's more accurate than similar running efficiency-based metrics from wearables such as the best Garmin watches.

Finally, we have the Tread+ Vision, priced at $6,695, with pricing for other regions TBC. This treadmill also features Sled Mode, a Hyrox-style workout that has you grip two vertical bars and replicate a sled-pushing exercise against variable levels of resistance.

"There is one other competitor that has this on the market, that treadmill costs $15,000," says Caldwell. "I know this thing isn't cheap, but, like, you know, for around seven grand US, you're getting an immense value in the form of the highest-quality energy return service, and now Sled Mode."

With all these treadmills, you get Peloton IQ features such as the new ReRun functionality and an IQ Pace Target, which uses AI to calculate times for you to aim to achieve for various distance. These features are also coming to historic treadmills too.

"The only thing that is gated is the Run Analysis because it uses that Computer Vision camera," says Higgins. "Our philosophy is if the hardware allows it, to give it to everyone."

You also naturally get access to Peloton's smorgasbord of live and historic classes, depending on your subscription tier. Heart rate zone-specific workouts are a new addition to the roster, allowing people to connect heart rate monitors and wearables from Garmin, Apple, Google and now Whoop, alongside new running, walking and hiking workouts.

Seeing the Vision

Seeing as we were at Peloton HQ, I had the opportunity to have a quick go on the Tread Vision to try some of the new running tools for myself. The cameras were quick to set up, putting in some information such as my height and capturing my body in vertical orientation as a phone camera does.

During a brief 10-minute test run, I can attest to feeling as though there has been some improvement of the energy return of the belt compared to my experience with the previous model. The belt was nice and spacious, and using the jump-up and jump-down buttons as opposed to, or in conjunction with, the dial was helpful when following the class's cadence as dictated by the instructor. No more spinning the dial too far and trying to course-correct during a class.

After that short test, I ended up with a fairly pedestrian Running Efficiency Score of 66, viewable in the Peloton app, with individual scores for specific metrics, and some apparently genuinely helpful advice on working on my vertical oscillation. However, given the brief nature of the trial, I imagine the feature would become more accurate with longer workouts — at least, that's my excuse.

I'm looking forward to trying all three new treadmills over a longer test period — stay tuned.

If you're British, you're automatically a Maxine Peake superfan — regardless of your age or generation. From Silk to the hugely overlooked The Bisexual on Channel 4, she's starred in banger after banger.

Now, she stars as delusional wife Jenna in gripping Australian Apple TV thriller Last Seen, which follows emergency dispatcher Ian (Patrick Brammall) trying to find his missing daughter after a mysterious phone call suggests she's alive years after going missing.

It's a gripping watch from start to finish, but I can't help wishing that Peake had more screen time. Did that stop me blurting out "I love you" during a Zoom call about the series? Absolutely not.

More than anything, I wanted to know what Peake recommended that fans should watch next, whether that was continuing the same vibe or finding something completely different to decompress.

She went with the latter not only by choosing my favorite British comedy series of 2026, but by picking something that isn't going to cost you a dime to watch on BBC iPlayer.

Maxine Peake is a huge fan of Ann Droid on BBC iPlayer — and I don't blame her

"I would go to the BBC and I would watch Ann Droid with Diane Morgan," Peake tells me. "It's another side of some Boltonian joy."

I'd completely forgotten that this is where both Peake and Morgan come from, but even without the geographical link, it's an excellent shout. if you've not yet seen the hit comedy series, Morgan plays Linda, a secondhand AnnDroid Z58/100 Basic Eldercare Robot gifted to widow Sue (Sue Johnston) by her son Michael (Kevin Ready).

Basically, it's like Channel 4's Humans meets a Royale Family and Motherland reunion that we weren't expecting, and it's nothing short of glorious. Clearly, everybody else who's watched thinks the same thing, as season 2 was renewed shortly after all six episodes of season 1 were released.

Morgan is truly the most talented, authentic and unique voice in Western comedy at the moment, but even those, this is huge praise from Peake. Why? her first job out of drama school was snarky Twinkle in Dinnerladies, the classic Victoria Wood sitcom.

Despite only running for two seasons, it's a cornerstone of British sitcom culture, a baton that Morgan has picked up and run with. It's unexpected, but the more I think about it, I'm not sure a that a better TV pairing is possible.

Today Samsung debuted a pair of new Tablets: the Samsung Galaxy Tab S12+ and the Samsung Galaxy Tab S12 Ultra. I’ve been taking the Ultra for a spin ahead of its October 7 launch and have been loving the massive display.

But the price hike, and minimal upgrades leave me questioning if it really needed it launch.

There’s a lot that’ll be familiar here for owners of the Samsung Galaxy Tab S11 Ultra when it comes to the S12 series. The AMOLED screen is still 14.6-inches with a 120Hz refresh rate and 1,600 nits peak brightness, the battery is still 11,600mAh with 45W charging (and boasts the same promised 23 hours of video playback as the S11 Ultra), and the S12 Ultra tablet is still IP68 rated meaning it could spend 30 minutes underwater at a depth of 1m and survive the experience.

The tablet is as thin (5.1mm) as it was before, with the same Silver and Gray color options to choose between as before. The only notable change is the chipset — the new 3nm processor boasts an up to 19% faster CPU, 17% better GPU performance, and a 51% improvement to its NPU processing for your AI needs — though there is also a new heat chamber for better heat dissipation, plus a new four-channel speaker setup for more immersive audio than last generation’s dual-speaker build.

I’ve been trying the Ultra model out and enjoying it so far — relying on it as a solid laptop alternative with the book keyboard cover attachment (sold separately), and also as a great mobile gaming and TV-watching device while I travel (a full review will be coming soon) — but considering how similar it is to the last generation, I am left wondering if Samsung should take a leaf out of Apple’s book and abandon its annual refresh approach to Galaxy Tab launches.

The Tab S12+ is an interesting beast. There was no S11+, so it’s a replacement for the S10+ from 2024, but there’s also no regular S12 launching alongside these models, so it's kind of the successor to the S11 too.

Samsung’s Galaxy Tab S12+ boasts a 12.6-inch AMOLED screen with the same 1,600 nits peak brightness as the Ultra. It also has the same chipset as the Ultra model, but there's a smaller 10,600mAh battery (unsurprising, since it's physically smaller).

Besides the lower battery capcaity and screen size it also lacks the Ultra’s 8MP ultra-wide rear camera — but otherwise has the same 13MP main on its back, and 12MP ultra-wide on its front.

Because the S12+ is billed as a step-up model, despite there not being a base S12 to step up from, it does mean Samsung’s cheapest 2026 model is a lot more pricey than its cheapest 2025 tablet (more on that in a second).

For someone on a budget thinking of upgrading, it’s a real shame the best option from Samsung might be a tablet that’s a year old, but unless we get a regular Tab S12 (I’m not holding my breath) that’s the position we find ourselves in.

Coming for the iPad Pro?

As we’ve come to expect in 2026 — though it still stings to see every single time — the tablets will cost more than what has come before. The Galaxy Tab S12 Ultra will set you back $1,399 / £1,399 / AU$2,299 — up from $1,299 / £1,269 / AU$2,099 – meanwhile the S12+ costs $1,199 / £1,199 / AU$1,999 for its cheapest model.

For both tablets the cheapest configuration comes with 256GB of storage and 12GB of RAM, with both also having a 512GB model, and the Ultra boasting a 1TB model that also ups the RAM to 16GB, though these configurations will cost more.

At these prices, Samsung is pitching its S12 lineup against Apple’s iPad Pro lines, but thankfully the prices are a little cheaper. Compared with the 2025 iPad Pro launch prices, Samsung’s tablets would be pricier — however Apple hiked its prices by up to $200 / £200 / AU$400, putting its 11-inch iPad Pro on-par with the S12+, and making its the 13-inch iPad Pro $100 / £100 pricier.

There are factors to consider beyond price, of course. The Samsung tablets have larger screens than their respective rivals and a longer quoted battery life, and come with the S Pen stylus in the box, while the Apple Pencil is sold separately.

But Apple’s tech is lighter — 444g and 579g compared with 553g and 692g — and while we need to do some more testing, Apple silicon tends to outperform its rivals.

All that aside, there’s plenty to still love about the S12-series Galaxy Tabs. I’ve been using the Ultra for a couple of weeks now and it’s genuinely a very useful machine. I’m currently writing this story using the tablet and its keyboard attachment (multitasking across documents, web pages, and my photos app with ease).

But again I’m left wondering if it really needed to exist considering the lackluster hardware upgrades and price increase. Right now I’m not sure, but let’s see how I feel when my review is ready

Summer is over, but the arrival of fall has brought with it a ton of excellent new gadgets. TechRadar's product testers have been busy reviewing tech from across a wide spectrum of markets, and we've uncovered some truly outstanding new products.

Below, I've rounded up all the products that scored 4.5 or 5 stars out of 5 in our in-depth tests. Perfect scorers in our monthly Reviews Recap are usually few and far between, but this past month we've had a bumper crop. There's the Sonos Beam Ultra, which we called "the ideal mid-sized Dolby Atmos soundbar", and Sennheiser's Momentum True Wireless 5 earbuds, which piqued our interest with their user-replaceable batteries and then won us over with their impeccable sound quality.

For those who like to take their tech out-and-about, the DJI Osmo 360 II fixes the first gen's flaws and gives GoPro and Insta360 "a serious run for their money", and the Garmin Fenix 9 Pro delivers "near-perfect performance" for a price tag we've politely described as "aspirational". Oh and there's a new phone out from a little brand called Apple — maybe you've heard of it? We were pretty impressed with the iPhone 18 Pro, too.

Scroll down to browse all the best tech we've tested over the past four weeks, and tap the View details button for a summary of what we thought, plus a link to read the full review for each.

Note: only the products that are available in your region will appear below.

5-star products

4.5-star products

Remember when the Apple Watch was just a fitness gadget, primarily devoted to notifications and workout tracking? Now it's like a medical technician on your wrist, measuring everything from heart rate (with readings every 5 seconds), temperature, blood oxygen, respiratory rate, and sleep. It's so comprehensive that it can now, with the enhanced Vital apps, give an overall sense of your current well-being.

Apple gave its now two-year-old Vitals section a makeover in preparation for the upcoming Readiness view in the soon-to-be-released redesigned Health App. In the meantime, you can see Readiness in Fitness. It uses readings from Apple Watch (Vitals, Sleep, Activity) to come up with a Readiness score. Basically, the number, on a scale from 1 to 10, tells you how ready you are to take on the challenges of the day. Today, I'm a 7!

I'm not a health nut, but I try to keep physically active and in shape, and I generally watch what I eat (and sometimes that means watching myself eat a donut). On my Apple Watch Series 12 — as with previous Apple Watches — I enjoy tracking my workouts and, ever since I started wearing the watch to sleep, my sleep score. Apparently I get, on average, roughly 4.75 hours or so of sleep a night (ugh).

For my needs, knowing that I completed a workout and that my sleep is good (or not) is enough, but perhaps I'm not typical. Certainly, Apple has seen an opportunity to transform the Apple Watch into a true and more proactive health wearable, one that can not only measure vitals in real time but also alert you to potentially serious issues.

Apple Watch knows a lot about you

We've all read stories of the Apple Watch detecting potential heart issues and people then getting checked out by a professional and discovering a real, albeit treatable issue. Apple Watch has long been able to help you in dire situations like a car crash or fall. When I took a spill not too long ago, I set off Apple Watch's fall detection (I wasn't hurt, so I stopped it from dialing emergency services), and I saw it do the same when a relative took a particularly dramatic tumble, where I think we let the call go through (she's fine). I do often recommend the Apple Watch as an eldercare device.

These are all positives, but I'd be lying if I didn't admit that the level of physical data my Apple Watch is now collecting and collating in the Health app (and Fitness App) is overwhelming.

With Apple leaning harder and harder into health tracking, Apple made a major sensor change for the Apple Watch Series 12. Its new sunburst-style heart rate sensor features large green LEDs (replacing the sparser red diodes) in a ring array capable of taking heart rate readings every five seconds (there's also 24% more watch surface-to-skin contact). Now I have Heart Rate Variability (HRV) readings that are fed to, among other things, the enhanced Vitals section in Health.

It's here that I happened to notice a single "Slightly Off" Vital reading and started spiralling, wondering, worrying, and also laughing because it would not be the first time I've been called "Slightly Off". Normal readings are called "Typical", and, I guess, if you're lucky or super healthy, "Favorable."

What's wrong with me?

Vitals combine respiratory rate, wrist temperature, Blood Oxygen, Recovery HRV, and sleep duration (each day shows the average for each measurement). The single "Slightly Off" reading was attributed to a high heart rate while sleeping. Apple Watch Series 12's ability to measure heart rate every five seconds (previous Apple Watches could only do it every five minutes) meant that this was probably a pretty accurate reading.

According to Vitals, my typical sleeping heart rate is 51bpm. On September 25th, the heart rate shown was 60BPM. That's not super high, but based on my average, it is, in fact, "Slightly off".

So what am I to do with this information? For now, nothing. After all, this is a single day and, since then, I've had nothing but "Typical" Vitals measurements.

We're still weeks or more away from a Health App redesign that might make all this data even more digestible, to say nothing of all the AI features (Live Rewind and Siri Recap) that are due in late 2026.

Even though Apple Watch Series 12 doesn't look much different from its predecessor, I've come to realize that the almost constant heart monitoring, and the sheer mountain of data the wearable is now collecting, is potentially a health game changer. I still find it all overwhelming, but I can guarantee I'll be watching like a hawk for the next "Slightly Off" reading.

Have you noticed the increase in local event posters made using AI? Suddenly I’m seeing them everywhere. They’re easy to identify because they all look similar, and generic. ChatGPT seems to love marker-felt-style fonts for headings, while the main image invariably has that unmistakable AI-slop look.

Perhaps the new ChatGPT Images 2.5, with its templates for event posters, is to blame, but to me it’s all a bit sickening, and it genuinely puts me off whatever the event is. I feel the same way when I see AI-generated posters in cafe windows showing food.

I’m not the only one either; AI is seen as an increasing threat by those who work in the creative industries. Scientist and professional cartoonist Dr. Maria Boyle, whose cartoons appear in New Scientist magazine and who goes by the name twisteddoodles, came to my attention for her cartoon about AI posters:

Getting the ick

I asked Dr. Boyle what she thinks of the AI-generated posters appearing everywhere.

“I always feel a sense of disappointment and creatively I get the ick,” she says. “It hurts most when it is places like the library or local schools who are meant to be bastions of creativity. The thing about AI-generated posters looking generic and the same means that essentially all events have the same branding. If AI is used by really good events but also really terrible events, you can’t trust AI art as a brand to equate quality.”

She finds it especially disappointing when businesses use AI to advertise their food with clearly artificial images.

“Recently I went to a coffee shop and all of the posters were AI generated, and the food had an uncanny valley nature to it; but one of the posters advertised a mango granola item I like to get. But because the person making the poster couldn't get the AI to make the food item look like what they served, they had to change what they served to look like the AI image. So now it is actually bigger and uses more ingredients.”

It’s a funny example, but it also gets at something more serious: what happens when the marketing stops reflecting the actual business?

Recent research from YouGov found that 78% of Britons would be less trusting of a brand that used AI-generated actors in its advertising, highlighting the potential risks for businesses that rely too heavily on artificial intelligence in consumer-facing campaigns.

Stephen Brookes, founder of UK-based creative agency NPK Media, warns that a purely AI-driven approach to marketing could damage customer trust, one of the most valuable assets a business can have.

“If every advert and social media post starts to look and sound the same because it has been generated using the same tool, businesses risk losing the personality that makes them stand out,” he says.

To Brookes, it’s not just about saving money in the short term, “A business might save money by generating an advert in a matter of minutes, but if that advert feels generic, artificial or disconnected from the reality of the business, the long-term cost could be much greater,” he adds.

“The best marketing doesn't just fill a content calendar. It communicates something real about the people, products and values behind a business.”

The mysterious talent fairy

I’m always struck by how a lot of people seem completely blind to the distinctive look of AI art, or don’t see any problem with using it at all. Dr. Boyle has noticed the same thing — and suggests it points to a deeper feeling of entitlement:

“Some people feel they should be allowed to use AI because of the perception that talent and artistic skill is something innate, and that if you aren't born with it you should have an entitlement to it and technology can give you that,” she says. “They believe that they aren't stealing from artists, they're simply giving themselves something that they were given by fate. But if you ask most artists, what they do has taken years of work and training to achieve and was not something gifted by the mysterious talent fairy.”

Of course, ugly posters are only the visible end of a much bigger issue. A few years ago, much of the debate around generative AI centered on whether companies should be allowed to train models on artists’ work without permission or payment. That debate hasn’t disappeared, but it feels much less prominent today.

“At this stage the models have been trained, they have access to the internet and are still training," Dr. Boyle adds. “Now I don't think the fight is with AI, the fight is with companies and individuals to see how much they value creative talent.

“That fight is people saying 'This is made with AI, I don't like it' because if AI creation becomes everyone's ick, not just mine, then that branding will be so damaged they will have to come back to us. Come back to me. I like money and I write really good copy.”

What if you could "spend the day making a shot that AI could make in a minute?" That's the somewhat facetious pitch for Polaroid's new all-analog Polaroid Mod Instant camera, and, in my experience, it's a challenge I accepted and thoroughly enjoyed.

No, it doesn't take all day to make some pretty wild instant Polaroid photos with the Mod, but it's also not as easy or unsatisfying as entering a prompt and waiting for a far-flung generative model to do the work.

Sitting somewhere between the Polaroid I-2 and the Polaroid Flip (in functionality if not price), the Polaroid Mod, which Polaroid unveiled today (September 29) and gave me some time with earlier this month, takes virtually every setting that might've been digital on a previous 21st-century Polaroid and made it analog, or more precisely, a series of dials.

With its metallic finish and familiar curves and angles — right down to the flip-open front film pack loader —it's instantly recognizable as a member of the Polaroid family. The gray body features a two-lens system: a 103mm main lens and a 95mm portrait lens. It uses an infrared autofocus system (it has just two focus zones, to keep things simple). There's a small dial on the front for exposure control with just three settings: auto, one step lower, or one step higher.

There's a large embedded flash and a red shutter button. On the back is a touch control for the flash, an LED light readout for the number of photos left, and a self-timer. There's a USB-C port to juice up the 750mAh battery, which should last through shooting over a dozen film packs.

The most important feature, though, and one that leads to an almost endless array of creative possibilities, is the mode dial around the main lens. In addition to the auto setting, it offers a Double-Exposure setting, Split Shot, Motion Freeze, and Light Painting. Each is represented by a tiny but clear cryptogram. The camera even ships with a special half-lens cover to assist in the Split Shot.

Like most of Polaroid's other full-size instant cameras, the Mod is designed to work with 600 film and Polaroid's I-Type instant film packs.

Let's shoot

Polaroid walked us through some of the basics of shooting with this relatively retro camera, and I was so struck by the simplicity of each operation that I felt empowered to try a few of them.

At 520g, the camera feels lightweight but substantial in the hand. It's also pretty rugged: I accidentally dropped the camera from a height of almost four feet. The film pack door popped open, but the Polaroid Mod was otherwise unharmed and worked for all my future shots.

I held the camera at eye level, looked through the large viewfinder, and snapped a photo of photographer and current Director of Polaroid's 20x24 Studio John Reuter. A second after I pressed the well-placed shutter button, I heard that iconic motor whir to life as the Mod spit out my first, still blue, undeveloped image. Polaroid film still takes 10 to 15 minutes to fully develop. When I finally had the finished image, I was impressed with the quality despite the suboptimal lighting.

I was anxious, though, to try out the wilder flights of photographic fancy, and with a colleague, I set about making my first Split Image.

First, I selected the mode by twisting the dial until I could see that icon. I asked my friend to stand with a drink and imagine she was toasting herself. On the camera, I placed the half-lens cover, which, as the name implies, leaves only half the film exposed.

In this mode, the camera will show a 1 on the tiny readout when you take the first photo.

I then repositioned my friend so she was shifted to the right and facing the opposite direction. Meanwhile, I took off the lens cover and switched it so the other half of the film would be exposed. On the readout, it shows a "2" to let you know the camera is ready for the second image.

I took the second image, and then the photo spit out of the Mod. The finished image was pretty cool, though I realize more care in the camera position and my subject would be needed for the maximum effect.

The Double Exposure setting is somewhat similar, but instead of using the lens cover, you shoot the first image with the lower exposure setting and then find your second subject and shoot the next image (again, the digital readout shows a "2") for the next part of your double image. The example we saw was an oversized live cat next to his tiny owner.

There's a motion freeze that opens a long exposure in which you can move the camera and then blast a flash at the end to capture a sharper image.

I was intrigued, though, by the Light Painting settings. After watching Polaroid try it with a variety of light sources, I asked Tom's Guide's Kate Kozuch to work with me on making my own Polaroid Mod light painting photos with an iPhone flashlight as a light source.

This Polaroid Mod mode gives you a full seven seconds of exposure to create some pretty wild images. Polaroid recommends a tripod, but I'm pretty good at holding still. For our shots, which we did in an almost fully darkened event space, I had Kate wave around the iPhone flashlight in a couple of different patterns. The results, which I saw 15 minutes later, were honestly far better than I expected.

I should be clear about something, though: most of these images, even in the best lighting, are arthouse quality. They can be grainy, moody, with sharp contrast between light and dark. I tend to love them, but if you're used to the clarity of an iPhone or Samsung Galaxy, you might be surprised, and not in a good way.

These images tend to tell a story with every bit of unpolished grain, so how you shoot it is as much a part of the tale as what you shot. With enough fiddling with the few available settings and true daylight, you might get some images approaching basic iPhone 13 quality but, again, that's not why you buy one of these cameras.

The camera, which goes on sale today, lists in the US for now (globally early next year) for $169.99 (a little cheaper than the Flip and roughly half the price of the more complex I-2), a comparatively affordable price, but keep in mind that Polaroid 600 and I-Type film can run you $20 for 8 pictures. At least you can get twice the imagery with the double exposure and split modes.

It’s preview season for League of Legends TCG Riftbound’s upcoming Radiance set, and we’ve got a brand new card to showcase for you today: Satchel Charge, as well as another look at the Bomb token it creates — a game piece I expect to make a bang in the set when it releases.

Satchel Charge costs three and recycles a Chaos rune to let you move a friendly unit as an Action, and then Deploy a Bomb to a battlefield.

Bombs deal 2 damage when they die (and set off all other Bombs you control at the same battlefield), which pairs very nicely with Deploy — a new keyword which lets you play a Gear card with Deploy directly to any battlefield (whether you control it or not), but the Gear dies when your opponent holds the location.

If you have multiple Bombs ready and waiting, you could seriously punish an opponent for claiming a battlefield by blasting their amassed units to smithereens.

Explosive plays

While an obvious use for Satchel Charge will be to save a unit by returning it to your base or moving it to a different battlefield, and leaving an unpleasant surprise behind for your opponent, it doesn’t have to be used this way.

It could instead be used to bring a unit and a Bomb to the same battlefield, or send your unit to one and a Bomb to the other. The versatility makes me think this could be a solid limited pickup to include in any Chaos-imbued Pre-Rift deck, but I wouldn’t be surprised if it saw some constructed play — especially if there’s a new Legend who plays nicely with Deploy and/or Bombs.

Though the big downside of Bombs is that they require an opponent to hold a location before they detonate, meaning you basically have to let them score another point before they face any consequences, which is certainly not ideal.

Plus, unless you have a few explosives ready, it’s possible your Bombs won’t be all that impressive when they do go off in terms of making a dent in your rival’s army.

Ziggs's Legend lets you make Bombs by exhausting it, while one of his revealed Champions makes Bombs (and other damage-dealing Gear cards) deal one more damage, so this card looks like an easy include in his deck.

Though we'll have to wait and see if these cards can do enough to make Bombs hit hard enough for them to be a new meta pick once Radiance launches on October 23.

• Vivo announced a new G2 Ultra 4.7x teleconverter lens alongside its flagship X500 series of phones
• The lens comes in the X500 Pro Max photographer's kit, or it can be purchased separately. It's compatible with all three X500 phones
• The phones and accessories are China-only for now, with a wider rollout TBD

My first time was… OK, but my second was a revelation. Behave, I'm talking about taking photos with a smartphone and teleconverter, of course, and Vivo has shown Oppo, Honor, and everyone else, in fact — including Apple — how it’s done.

Available separately (in China only) or in the Vivo X500 Pro Max photographer’s kit (much like the previous X300 Ultra-compatible model), there are now two teleconverter lenses that extend the phone’s reach: a 2.3x and a new ‘G2 Ultra’ 4.7x.

Like the Oppo Find X9 Ultra and its teleconverter lens, they are designed to be paired with the phone’s telephoto camera, but in this case, it’s a 3.5x lens with an 85mm effective focal length. Add the ‘G2 Ultra’ 4.7x teleconverter — which is a proper bit of gear with Zeiss optics — and you therefore have a 400mm lens, no digital zoom required.

What particularly impressed me on the Vivo X500 Pro Max is how its teleconverter is implemented within the phone's camera app. You check the teleconverter option in the camera menu, after which an icon always appears on the screen.

If you want to use the teleconverter, tap the teleconverter icon, follow the prompt to select which teleconverter is attached, and the phone corrects the image for use with the phone (the teleconverter uses mirrors, so the image would otherwise be upside down).

The Oppo phone, on the other hand, requires a dedicated teleconverter mode to optimize the image. This mode is severely simple and restricted, and therefore you cannot use it with the phone’s best camera features. The same can be said for new Honor Magic 9 Pro Max and the latest iPhone 18 devices paired with third-party options such as PGY Tech and its associated app (as demonstrated by @byteddyk).

Meanwhile, the Vivo X500 Pro Max doesn’t require a dedicated mode — it can be used with any of the other modes, with no real restriction in place, making the teleconverter inherently more useful.

A teleconverter with no limit

Working with any shooting mode and the 3.5x telephoto camera means that users can access the phone’s maximum 200MP resolution. You can select the Pro mode with RAW file format and full control over exposure, too.

Or how about SnapShot — a mode designed for action photography that automatically uses a high shutter speed, which you likely want to use at such a magnified 400mm focal length for wildlife and sports photos.

Oppo’s teleconverter mode can do none of those things, and it’s real case use is fairly limited, as I discovered when attempting to photograph a pro Champions League soccer match from my distant seat in the stadium.

The cherry on the X500 Pro Max and teleconverter cake is that the lens is supported by decent image stabilization, though you're not getting the 7EV in-body image stabilization like with the phone's main camera.

The lens comes with a collar and tripod mount, making it easy to attach to a tripod. It's best to use one, but I found that I could still shoot sharp images, smooth video, and track subjects handheld at 400mm.

It's also possible to extend the 400mm reach digitally, with 800mm, 1200mm, 1600mm, and 3200mm options, but I find that using the digital zoom with a Vivo phone makes for artificial-looking photos (when pixel peeping, at least. A better option is to up the resolution from the standard 12MP to 50MP or 200MP and then crop in. Check the gallery below to see how much you can crop in from a 200MP file.

I still have the Oppo Find X9 Ultra and its teleconverter, but I couldn’t ever get past the limitations of using it. The Vivo, on the other hand, will be the setup I reach for next time I want to grab super-telephoto shots with a phone.

For now, sadly, the Vivo X500 series and accessories are only available in China, with a wider rollout TBD, though these flagship Vivo devices have not sold directly in the US or UK before.

Still, the G2 Ultra 4.7x teleconverter is a real showcase of what smartphone cameras can be capable of — hopefully, one day, they will be more widely available.

If you're under the age of 30 and think of the name Celia Imrie, you likely think of the woman who farted in a log cabin during an episode of The Celebrity Traitors season 1.

Obviously you'd be factually correct, but this doesn't even begin to scratch the surface of just how iconic Imrie actually is. From Acorn Antiques and Dinnerladies to Calendar Girls and The Best Exotic Marigold Hotel, if there's been a worthwhile British drama, she's been there to improve it.

Now, the same can be said for War on HBO Max (international) and Sky TV (UK). Following two legal firms as they represent a spouse in a high-profile divorce trial, Imrie plays Budgie, a freelance barrister working for the newly established Taylor & Byrne.

From watching the trailer below, you might have assumed, like me, that Imrie was to play the High Court Judge overseeing the trial. According to Taylor and Byrne themselves, James McArdle and Phoebe Fox, Imrie had those onset so fooled by her power, that their opening scene in episode 1 had to be entirely reshot.

'She has such inherent power'

"She is such an incredibly charismatic woman and has such an inherent power," Fox tells me. "Originally, she just turns up and is in the office with us having a scene — but they actually had to reshoot her entering the office because everyone who watched it thought that she ran the company.

"We're the bosses, but that wasn't clear, apparently. She's a dream woman to work with."

And water is wet. This news doesn't surprise me in the slightest... but what I didn't realise is that Celebrity Traitors was airing at the same time as War was being filmed, which must have been like striking gold for any other cast members with a penchant for reality TV.

"While we were filming, Traitors was on," McArdle adds. "It was really hard for the director to get me to concentrate because I just kept trying to get spoilers out of Celia and Nick (Mohammed). Celia and I knew each other from way before, so I thought she would tell me... but she never did."

Oh, didn't I mention that there's also a Celebrity Traitors reunion within the cast too? More on that later this week. For now, set your clocks ready to watch War this Friday (October 2).

September was a surprisingly stacked month for horror. I was spoiled for choice when putting together this month's horror movie recommendations, and I've got a real mixed bag to cover here too.

Most notably, September saw new entries into the Creep and Silent Hill franchises. This means there's some familiarity here for fans keen to dive into these well-established horror worlds. But, as always, Shudder had a good helping of new originals to keep us entertained too.

I spent much of this month with the following three titles, and here's what I thought of each one.

TV — The Creep Tapes season 3

Where to watch: Shudder, AMC+
Lucy's verdict: Scream

Mark Duplass can do no wrong, in my opinion. After making an appearance in the hugely successful Backrooms this year, he's gone back to his original horror roots. Indeed, The Creep Tapes season 3 is the latest installment in his found footage nightmare, and it's incredibly good.

If you're keen to deep dive into Peachfuzz's character, season 3 does exactly that. Josef's (Duplass) serial-killer alter ego is front and center here, and this feels like it's going back to the Creep we know and love. We're uncomfortably up-close to Josef here, seeing many things from his point of view, making for an effective third season which might sit with you for a long time.

It's well worth checking out the first two Creep movies if you haven't already, which are available on Prime Video in the US and Netflix in the UK.

Movie — Parasomnia

Where to watch: Shudder
Lucy's verdict: Skip

Shudder's new movie Parasomnia has an interesting premise, but it ultimately didn't do enough to keep me engaged. The movie follows a woman who has recurring night terrors — and each time they happen, awful things happen around her in the real world.

When she struggles to keep these terrors at bay, a threat emerges and terrorizes her, which makes up much of the movie's 85-minute runtime. Unfortunately, Parasomnia suffers from an over-reliance on jump scares teamed with some predictable narrative beats, so I wasn't a huge fan of this movie.

It is on the shorter side, so you may want to try it anyway, or check out the line-up of new Shudder releases to find something new.

Game — Silent Hill: Townfall

Where to play: PS5, PC
Lucy's verdict: Scream

Silent Hill has a new spin-off game starring Welcome to Derry's Jovan Adepo. It's set in the coastal town of St. Amelia, and I spent my weekend getting lost in this creepy new world.

It's important to note that Silent Hill: Townfall is separate from the main storyline, so you won't find well-known characters like Pyramid Head here. Instead, you play as newcomer Simon in a slow-burn narrative that encourages you to explore in first-person. Despite a few initial glitches like the map not loading in, I did settle nicely into Silent Hill: Townfall, and I was particularly impressed by the central location, which is modeled off a quiet Scottish town.

Silent Hill: Townfall is a relatively short game which is easy to complete in a weekend, and I recommend giving it a go.

Recent reports suggest that although .COM remains the undisputed domain top dog, several other TLDs are starting to threaten that position – and .WEB may just be a strong contender to do the same.

Although the .WEB TLD has been tied up in legal and admin since it was auctioned off in 2016, it is now about to hit the market, offering a new generic domain which could rival rising stars like .AI and .IO, as well as established giants like .COM.

I caught up with Mike Chattan, Senior Director of Domain Operations at Network Solutions to get his thoughts on how .WEB may change the domain landscape and how businesses can take advantage of this new opportunity.

Can you tell us a little bit about the history of .WEB and why, despite its highprofile, it is only now becoming available for use?

During the last round of new generic top-level domain (gTLD) launches, .WEB was seen as an extension that challenges .COM. It’s short and closely tied to the history of the internet itself through the term 'World Wide Web.’

.WEB generated significant interest because it can give customers more options when their preferred .COM domain isn’t available. Seven applicants competed for .WEB, and the contention was resolved through an ICANN auction in 2016. The subsequent legal and regulatory process delayed the delegation of .WEB for years.

With ICANN now officially delegating .WEB to Verisign in July 2026, the registrar has said it plans to begin offering .WEB later this year. Now, we’re at the point where businesses and individuals can start considering what this extension could mean for them.

How do you see .WEB fitting into the domain landscape alongside powerhouses like .COM and .NET? 

.COM and .NET are what many people associate with a website or the internet, but a domain extension alone doesn’t determine a business’s success. You don’t need a .COM or .NET to build a successful business. It can help with familiarity, but at the end of the day, you drive your own success through your brand, your customer experience, and how effectively you market yourself.

For businesses using .WEB, the key is consistency and education. Use the domain across your website, marketing materials, social media, and other customer touchpoints so people become familiar with it. Don’t assume customers will automatically recognize a new extension or understand that it’s legitimate.

That education is important because people are more cautious about unfamiliar links and domains due to phishing and other forms of online abuse. So, build that trust around the domain over time.

Domains like .AI and .IO have exploded in growth in recent years. How could .WEB change the narrative for modern businesses? 

TLDs such as .IO,.SH, and .AI are technically country code top-level domains (ccTLDs). As we’ve seen, when ccTLDs have zero locality or other restrictions, they can act just like any other gTLD. .AI is a good example here because it’s popular among technology and artificial intelligence companies even though it’s the official ccTLD of Anguilla.

However, while we treat these extensions like gTLDs, some of them still apply ccTLD application rules, which makes it hard and more expensive for users to register.

Carrying and renewal costs are also problems for both corporations and single domain owners when it comes to these ccTLDs that have found other use cases. I will say that once the next round of TLDs comes out and customers have more options, these ccTLDs will become less relevant due to their pricing, unless the registries behind them make changes to accommodate non-country-specific customers. Customers might be able to find a better fit with new gTLDs for a quarter of the price.

How should SMBs evaluate if .WEB is the right TLD for them? 

.WEB is a brand-new TLD that gives SMBs the opportunity to register a name that may be unavailable in other TLDs.

But first, SMBs should consider if .WEB fits their brand, their audience, and the way they want customers to remember them. If they choose .WEB, they should use it consistently across their website, marketing, customer communications, and social channels. Like any new TLD, .WEB will require some education and repetition before it becomes familiar to customers.

Should brands be looking to secure .WEB domains as a defensive move? What are the risks associated with competitors buying up your brand associated .WEB domain? 

One concern that comes with new TLDs is brand protection. There is a lot to protect, and that comes with costs. That said, I don’t think .WEB requires a different brand-protection strategy from other TLDs. The same brand protection practices that apply to other TLDs should apply here as well.

For most brands, the important thing is protecting the second-level domain (SLD), regardless of the extension attached to it, and it needs to be done across the board.

How can brands measure the success of .WEB domains?

It will take time to understand how successful .WEB becomes. Registrations are an obvious starting point, but they don’t tell the whole story. I’d also look at renewal rates, the number of domains actively being used by businesses, adoption across different industries, and whether customers become familiar with .WEB to recognize and trust it.

The registry also has an important role to play. Historically, new TLDs have benefited when their registries invested in educating the market. Radix, for example, stood out during the last gTLD expansion by investing heavily in awareness and education around its TLDs. Verisign has a big opportunity with .WEB because they have an established .COM distribution network and experience operating domain registries.

I have this new friend, Charlie, who's been assisting me with work, vacation planning, and occasionally tapping me on the shoulder when he's concerned I forgot something. Charlie looks like a fat, furry Pillsbury doughboy, and he lives on my iPhone. That's right, I signed up for Meta's Muse AI, launched the app, named my avatar "Charlie," and dove headfirst into agentic AI in a way I honestly never have before.

If there's a defining feature of Muse, it's proactivity. Unlike ChatGPT, Gemini, or Claude, Muse is always running and constantly looking for ways to help and ensure you stay on task, on time, and generally don't mess up by missing crucial details.

Getting all this utility from an AI required a fairly significant leap of faith: my data, and I did this not without some trepidation. As I wrote on X the day this adventure began: "I let Muse AI look at my inbox, find an important message I missed, draft and then co-edit a response…and then send it. Have I lost my mind?"

It's not that I've never let an AI look at or access my data before. Google Gemini can see across Docs, Gmail, and Maps, but it's not natively agentic. So engagement with the data tends to happen at my request. It's not a busy beaver working behind the scenes to somehow make my life, if not better, then different.

Giving Muse access to my Gmail account did feel a bit like strapping on a parachute and jumping at 14,000 feet. In return, Meta promises some pretty stringent data privacy controls. Muse's constantly running personal AI computer lives in a virtualized, cloud-based environment and is shielded from other agents. Credentials are stored in a "Secure Credential Store," and the system's oversight ensures the agents don't get up to some risky business.

So, yes, I was scared when I gave Meta access, but I thought, "How can I understand this and tell my readers about it if I'm unwilling to try it and ensure it's not a dud, or worse, a disaster?" To be honest, I didn't expect much. Sure, the avatar, officially called Jolly, is adorable. He sits at the top of the app, working on his tiny laptop — it's like us, but cuter. I figured that promises of Muse actually helping me in any tangible way were remote. You can, by the way, name your avatar, so I chose Charlie (yes, I'm a big Charlie Chaplin fan).

Even what it was offering up top didn't sound inspiring. Muse, or Charlie, said it could:

• Manage my inbox
• Add or reschedule events in my calendar
• Find Facebook Marketplace deals

The last one really stuck in my craw because I basically hate Facebook Marketplace. It's too full of overeager buyers who want to show up at your house immediately ("Where is bed?")

Knowing, however, that I was committed to giving Muse a try, I agreed to give it access to my personal email (my work Gmail makes connecting to unsanctioned third-party systems next-to-impossible).

It took no time to make the connection, and I remember thinking, "Something like this should be harder, if only to deter mistakes." But with the deed done, I read over Muse/Charlie's offers to help. It could:

• Find an email or catchup on a long thread (that might be useful because Gmail's threading is notoriously abysmal)
• Draft a reply
• Unsubscribe from newsletters

Digging into my inbox

I was tempted to start with the last one because I have subscribed to way too many newsletters and alerts. My Gmail box is a mess, but since I was still unconvinced Muse could help me, I gave it what I considered a simple task: "Just check that there's nothing pressing I've missed."

Frankly, I expected little from this exercise because my personal email is not where my work gets done. Most critical emails are in my work address, so I was surprised when Muse/Charlie quickly told me that while there was "nothing on fire," it did find three that were "worth a glance."

The first was a Google security alert. With a decent level of self-awareness, Muse/Charlie said it was "just from connecting to me a minute ago, nothing to worry about."

The second was from my cell service reporting a recent change to my account. Charlie told me it was "Worth a quick look if you didn't make one."

The last one, though, surprised me. A PR rep named "Bill" had emailed my personal address about a new communication product. It was (and still is) embargoed information, so I can't give you details, but I can say that even from Muse/Charlie's very brief summary, I was intrigued.

"Open the Bill one," I told Muse/Charlie.

Instead of the entire email, Muse delivered an impactful summary. It was actually really useful, and Muse finished up by asking if I wanted it to draft a reply.

"Draft it," I wrote.

If it seems like I was being overly terse with Muse/Charlie, I simply didn't want to get drawn into this anthropomorphising game. I was here to find out if Agentic AI could work in my life and not develop yet another pointless AI chatbot friendship,

Muse's initial draft was short and serviceable, but it also overpromised, telling Bill that I wanted to set up a meeting or call. I preferred to learn more before I took that step.

I told Muse/Charlie to remove all references to a call or meeting; just agree to the embargo, get images and details, and then I'll consider how to follow up.

The next draft was perfect, and Muse/Charlie told me it was sitting in my Gmail drafts ready to send.

I told Muse/Charlie to send it but to add a CC to my work address.

Muse sent the mail but also offered to help with the story when it was time; I didn't respond to that because I have no interest in Agentic AI helping me to write anything beyond a simple email.

Muse knew less about me than I thought

Later that day, I had a speaking engagement and could not recall the details of where I was talking or where to park my car when I got there. I asked Muse to find in my email the original note from my contact. I misspelled the guy's name, but Muse still found it and gave me a concise and organized account of the location, parking, my contact's phone number, and even a useful map.

Muse doesn't automatically know everything about me, as I found when I asked Muse to map my trip to the speaking gig. When I told it I was heading from home, Muse admitted that it did not have my home address. Naturally, Muse asked for it, but I hesitated and only provided my hometown. For Muse, that was enough to give me turn-by-turn directions from one of the major roads near me.

Later, when I felt like I still needed some additional guidance on the talk, I asked to see the original invitation email, which Muse dutifully supplied.

Muse is always busy

I don't usually allow apps to send me notifications, but I rightly assumed that agentic AI might want to reach out and notify me. Sure enough, I noticed an alert from Muse/Charlie that Bill from the startup had already responded. Inside the Muse app, I found another concise summary of the email and an offer from Muse to pull the full product release from the email.

Muse is sort of relentless in its efforts to work for you and look out for what it perceives as your best interests. At one point, I got a security notification from Muse that it had spotted an email from Tripit, the app I was using to organize an upcoming vacation. I had forgotten and changed my password; Muse wanted to make sure I'd seen the email verifying that it was me making the change.

Emboldened to add to Muse's corpus of information about me, I decided to follow the tip about adding the context I'd built with another AI chatbot to Muse. Muse gave me the prompt to paste into the third-party AI and told me to copy and paste the response back into Muse. Gemini's summary of me was quite illuminating; most of it was spot-on and made me realize just how much time I've spent using AI over the last few years.

Muse/Charlie was also ready to address some discrepancies in the data. Months ago, I'd been asking Gemini about a potential two-week trip to Portugal. We eventually swapped that for Italy, but I never told Gemini. Muse noticed that while Gemini had Portugal, a recent email to a contact mentioned the upcoming Italy trip. So I told Muse, "One is outdated. We settled on Italy."

Agent = Assistant

Like any good assistant, Muse is far more organized than I am. It noted that my upcoming trip would leave little time to meet a deadline, and it wanted to make sure I filed my story before I left. I told Muse my work was almost done but did ask it to nudge me about another deadline.

Almost twenty years ago, I had an assistant. It didn't last long (budget cuts), but I remember how she was always more on top of my schedule and email than I was. I had a big-picture view of my work and strategy, but she covered the minutiae that made it all work. Muse/Charlie already feels a little like that. It's not doing my work, but it is already helping me stay on a more productive path by not letting me forget small but important details (and even some hidden connections).

When Meta CEO Mark Zuckerberg predicted over the summer we'd all have personal AI agents within five years, I thought it was an incredible self-serving statement. Sure, he believes that because he's investing billions in building them.

I don't feel that way anymore. Muse may be the first AI I'll use more consistently, and not for one-off prompt experiences, because it's always working on my behalf in the background. Its proactive nature means that it's constantly looking at my world, or rather the road ahead, to help me avoid potholes. Notifications about important emails, deadlines, and collisions I didn't see coming can be invaluable.

I think I'm ready to give Muse/Charlie more digital access to my life to see if/how it can improve it. I'm also trusting Mark Zuckerberg with my data, choosing for now to believe his promises of security and privacy. I pray he doesn't disappoint me.

Maybe that wearable Muse Charm isn't such a bad idea.

ChatGPT Voice has learned a useful new trick. When you're using the new Live voice model it can now use plugins and connected apps while you are talking to it, bringing services and brands into a spoken conversation.

I spent some time trying the combination with everyday tasks. Instead of asking ChatGPT about dinner and then opening a grocery app myself, for example, I can keep going by voice.

The change sounds relatively small until you use it because plugin support means some of those conversations can now lead somewhere beyond ChatGPT itself.

To use plugins with Voice you need to select the Live voice model. To make sure you're using Live go to Voice in Settings and make sure it is selected.