• The European Transport Safety Council said Tesla offered to change the name of its tech
• Germany’s transport ministry claimed Full Self-Driving was misleading
• BMW aims to pile on the pressure with its own automated driving tech next year

Following intervention from Germany’s transport ministry, Tesla has agreed to formally change the name of its Full Self-Driving (FSD) partially automated driving technology to ‘Tesla Assisted Driving’ in Europe and the UK.

The move came after Germany’s Transport Minister, Steffen Bilger, aimed to legalize Tesla’s advanced driver assistance systems as quickly as possible, but went on to explain that the name was “somewhat misleading”.

“It is a driver assistance system that can support certain driving tasks such as accelerating, braking and steering. However, the driver must remain attentive at all times,” a statement read.

As a result, Tesla agreed to change the name of the technology to appease European lawmakers, with Tesla Assisted Driving suggested and subsequently agreed on by all involved.

The new name has now been added to Tesla websites in Europe, including the UK site.

It comes as many governments move to tighten the rules around the marketing of assisted driving systems, according to Fleet News.

In the UK, for example, the terms automated, automated driving, autonomous, autonomous driving, drive autonomously, drive itself, driverless, and self-driving will only be reserved for authorized or listed vehicles, such as fully self-driving taxis.

The UK’s Department for Transport released a statement that said that under the 2024 Automated Vehicles act, which is expected to come into force on January 7, 2027, a vehicle only meets the self-driving test if it “does not need to be monitored or controlled by an individual”.

BMW readies its Symbiotic Drive technology

The German automotive giant announced this week that it will expand its urban driver assistance functions in Germany early next year.

A suite of navigation-guided driver assistance functions, under the umbrella title Symbiotic Drive, will provide support from the motorway, through the countryside and into the city, BMW claims.

The SAE Level 2 technology will able to handle turning and right-of-way situations, urban lane changes and navigating roundabouts using the vehicles in-built satellite navigation system.

Unlike other advanced driver assistance systems, BMW says its technology is designed to work seamlessly with the driver, so they can intervene, accelerate, brake and steer without completely shutting off the system.

Lane changes are also handled in an innovative way, with interior, gaze-monitoring cameras detecting a driver’s intention to swap lanes and the steering wheel detecting inputs before the system autonomously executes the maneuver.

BMW says that the system only provides assistance in the form of a “warning and steering intervention” if the driver is actually unintentionally drifting out of their own lane or if the system detects an imminent collision.

However, the company reiterates that Symbiotic Drive is a Level 2 system, so requires a driver to pay attention and be ready to take over whenever the technology requires.

It will be available on iX3, iX4, all models in the 3 Series, 7 Series and X5 range in Germany to begin with, but further countries will soon follow.

The German marque says the hardware is already in place in the majority of circumstances and that BMW Symbiotic Drive will be delivered as an over-the-air software update.

Watch Barcelona vs Aston Villa live in the Champions League 2026/27 as Unai Emery takes his upstarts to the Camp Nou in search of a fourth win from four in the league phase. Hansi Flick's Blaugrana have lost two of their last three in all competitions and want a statement victory.

After successive defeats by PSG and Real Madrid in the Champions League and La Liga respectively, Barça got back to winning ways at the weekend with a 6-1 gubbing of Alaves. Braces for Raphinha and Lamine Yamal helped maintain their two-point lead at the Primera summit and Los Cules must focus on tonight even with a top-of-the-table clash with Atletico Madrid coming at the weekend. That duo have 29 goals already this season and will want more to guide the Catalans up from eighth in the league phase. Fermin Lopez, who has 12 goals contributions to his name already this season should also start

After seeing off Club Brugge, a woeful Fenerbahce and Viking, Villa now face the five-time European champions – for the first time in more than four decades – in the sort of elite fixture that Emery relishes. Nicolas Jackson scored again in the weekend draw with Fulham and the in-form Senegalese has the physique and finishing ability to put pressure on a defensive unit that has been without Eric Garcia, Jules Kounde, Xavi Espart and Andreas Christensen through injury. Christensen is back in training and may feature, but Frenkie de Jong was pressed into emergency center-back duty against Alaves, a challenge an in-form Jackson would relish.

Read on as we explain how to watch Barcelona vs Aston Villa in the Champions League 2026/27 for free.

Watch Barcelona vs Aston Villa for FREE

Barcelona vs Aston Villa is being shown on Virgin Media Play for free in the Republic of Ireland.

Coverage begins at 7.45pm on the Virgin Media Two channel.

Traveling abroad today? You can use a NordVPN (try risk-free) to unlock your Barcelona vs Aston Villa stream on Virgin Media Play — we used it to watch Como vs Man Utd recently and it worked perfectly.

Use a VPN to watch Barcelona vs Aston Villa live streams

A VPN is handy piece of software that can make your device appear as if it's back in your home country, so you can unlock your usual service. The best VPN right now? We recommend NordVPN – it does everything and comes with up to 75% off.

How to watch Barcelona vs Aston Villa live streams in the US

In the US, the Barcelona vs Aston Villa game is being shown on Paramount+.

The streaming service, which is showing all 189 games of the 2026/27 Champions League campaign, starts at $8.99/month or $89.99/year.

Alternatively, you can access Paramount+ virtually for free with a Walmart Plus 30-day trial for $1.

Visiting the US from Ireland? NordVPN can unblock your Virgin Media Play stream of the game.

How to watch Barcelona vs Aston Villa live streams in the UK

Barcelona vs Aston Villa is being shown on TNT Sports 1 in the UK.

You can add TNT Sports to your Sky, Virgin Media or EE TV package, or get an HBO Max plan that includes TNT Sports. Prices start at £25.99/month if you commit to a year.

Outside the UK? NordVPN can unblock your stream from anywhere in the world.

How to watch Barcelona vs Aston Villa live streams in Australia

Barcelona vs Aston Villa, along with all 189 games of the 2026/27 Champions League season, is available to watch on Stan Sport in Australia.

Stan Sport costs AU$20/month on top of a Stan subscription, which itself starts at AU$9.99/month.

Not in Australia right now? You can simply use a VPN like NordVPN to watch the action as if you were back home.

How to watch Barcelona vs Aston Villa live streams in Canada

The Barcelona vs Aston Villa game is available to watch on DAZN in Canada.

DAZN, which is showing all 189 Champions League 2026/27 fixtures, starts at CA$24.99/month, for a plan that also includes EFL, Carabao Cup, Bundesliga and NFL coverage.

If you're out of Canada but still want to tune in, explore the VPN route set out above, which will help you access your accounts from anywhere.

We test and review VPN services in the context of legal recreational uses. For example:1. Accessing a service from another country (subject to the terms and conditions of that service).2. Protecting your online security and strengthening your online privacy when abroad.We do not support or condone the illegal or malicious use of VPN services. Consuming pirated content that is paid-for is neither endorsed nor approved by Future Publishing.

• Watch Man Utd vs Roma on Prime Video (UK) / Paramount+ (US)
• Unlock your stream with NordVPN (save up to 75%)
• Man Utd vs Roma kick-off: Tuesday, November 3 at 3pm ET / 8pm GMT

Watch Man Utd vs Roma live streams in the Champions League 2026/27 as the Giallorossi attempt to heap more pressure on Michael Carrick by inflicting a fourth consecutive defeat on the Red Devils.

Almost two months without a victory, it has been a brutal period for all involved at Man Utd. Now 17th in the Premier League table after losing 2-0 to Chelsea, they last tasted victory 10 games ago against European debutants Sabah. Since then, players have attacked the coaching staff and criticised the club’s hierarchy, while Carrick has received the dreaded vote of confidence from the board. With confidence at an all-time low and the commitment of certain players being called into question, this feels like a monumental game for Carrick and co.

Roma will be confident of making it back-to-back wins in the Champions League and extending United’s winless run. Gian Piero Gasperini’s side head to Old Trafford in good form, having thrashed Udinese 5-2 at the weekend and scored 16 goals in their last four games. In-form Dutch forward Donyell Malan will relish the chance to test United’s struggling defence and will hope to add to his impressive tally of 11 goals in 12 games in all competitions.

Read on as we explain how to watch Man Utd vs Roma in the Champions League 2026/27.

Use a VPN to watch Man Utd vs Roma live streams

A VPN is handy piece of software that can make your device appear as if it's back in your home country, so you can unlock your usual service. The best VPN right now? We recommend NordVPN – it does everything and comes with up to 75% off.

How to watch Man Utd vs Roma live streams in the US

In the US, the Man Utd vs Roma game is being shown on Paramount+.

The streaming service starts at $8.99/month or $89.99/year. Alternatively, you can access Paramount+ virtually for free with a Walmart Plus 30-day trial for $1.

Outside of the US? NordVPN can unblock your stream from anywhere in the world.

How to watch Man Utd vs Roma live streams in the UK

In the UK, Man Utd vs Roma is being shown on Prime Video Sports.

Amazon Prime, which shows one Tuesday game each matchweek, costs £8.99/month or £95/year after a 30-day free trial. Students and 18- 24-year-olds can get it for £4.49/month or £47.49/year.

Outside the UK? NordVPN can unblock your stream from anywhere in the world.

How to watch Man Utd vs Roma live streams in Australia

Man Utd vs Roma is available to watch on Stan Sport in Australia.

Stan Sport costs AU$20/month on top of a Stan subscription, which itself starts at AU$9.99/month.

Not in Australia right now? You can simply use a VPN like NordVPN to watch the action as if you were back home.

How to watch Man Utd vs Roma live streams in Canada

The Man Utd vs Roma game is available to watch on DAZN in Canada.

DAZN, which is showing all 189 Champions League 2026/27 fixtures, starts at CA$24.99/month, for a plan that also includes EFL, Carabao Cup, Bundesliga and NFL coverage.

If you're out of Canada but still want to tune in, explore the VPN route set out above, which will help you access your accounts from anywhere.

We test and review VPN services in the context of legal recreational uses. For example:1. Accessing a service from another country (subject to the terms and conditions of that service).2. Protecting your online security and strengthening your online privacy when abroad.We do not support or condone the illegal or malicious use of VPN services. Consuming pirated content that is paid-for is neither endorsed nor approved by Future Publishing.

• Watch US Midterm Elections live streams on Fox News or MS NOW via YouTube TV free trial (US only)
• Use NordVPN to watch from anywhere

Voters across the US are preparing to head to the polls as President Trump awaits an unofficial barometer on how his constituents feel about his presidency so far.

Control of the US Congress is up for grabs, as the Democrats bid to take that control from the current Republican majority. And what a slim majority it is, with Trump's party currently holding only the minimum number of seats needed to enjoy control of the House of Representatives. 35 of the 100 Senate seats are up for grabs too, with the Democrats needing to control a total of 51 seats for that to change hands.

New battle lines have been drawn, literally, by both sides, with some electoral maps being redrawn more favorably for one side or the other, leading to the upcoming vote looking a little more unpredictable.

It's a big election, that could spell major changes to the remainder of President Trump's tenure, so read on as we show you how to watch the 2026 US midterm elections from anywhere in the world.

Can I watch 2026 US Midterm Elections live streams for free?

For viewers in the US, Fox News and MS NOW are available on YouTube TV or Hulu + Live TV, both of which offer a free trial.

If you're happy streaming coverage online, NBC News NOW, ABC News Live, and CBS News 24/7 all offer up-to-the-minute reporting on their free news streams.

International free streaming options include CBC Gem in Canada, BBC iPlayer in the UK and SBS on Demand in Australia.

Away from home? Watch your usual streaming services from anywhere with NordVPN (try for 30 days).

How to watch US Midterm Elections 2026 live streams from anywhere

For those away from home looking to watch US Midterm Elections coverage, you may be unable to watch the show like you normally would due to annoying regional restrictions. Luckily, there’s an easy solution.

Downloading a VPN allows you to stream online, no matter where you are. Use a VPN to watch US Midterm Elections live streams online from anywhere:

How to watch US Midterm Elections 2026 live streams in the US

In the US, all major networks will be broadcasting coverage of the 2026 Midterm Elections on Tuesday, November 3, although start times vary.

• CNN: 4pm ET / 1pm PT
• NBC News: 5pm ET / 2pm PT
• Fox News: 6pm ET / 3pm PT
• MS NOW: 6pm ET / 3pm PT
• ABC: 7pm ET / 4pm PT
• CBS: 7pm ET / 4pm PT

Don't have cable? You could opt for a cord-cutting service Sling TV, YouTube TV or Hulu + Live TV.

Away from the US? Use a VPN to watch online from abroad.

How to watch US Midterm Elections 2026 live streams online in Canada

Canadians can catch the latest from the US Midterm Elections on CBC News, with reporting running throughout the day and live coverage from 7pm ET.

Viewers in Canada can stream for free via CBC Gem.

Outside Canada right now? Unlock your usual streaming service with NordVPN.

How to watch US Midterm Elections 2026 live streams online in Australia

Various networks in Australia will be covering the US Midterm Elections, with free live streams available from around midday local time on Wednesday, November 4 on the likes of 9Now, SBS on Demand and ABC iView.

Away from OZ right now? Just use a VPN to alter your IP address so you can stream your favorite TV shows and films online just like you would at home. More details above.

How to watch US Midterm Elections 2026 live streams online in the UK

Both Sky News and BBC One will offer dedicated US Midterm Election coverage in the UK, kicking off late Tuesday evening.

BBC coverage is available for free to licence fee payers via BBC iPlayer.

Brit abroad? Just use a VPN to alter your IP address so you can stream your favorite TV shows and films online just like you would at home.

US Midterm Elections - Need to Know

We test and review VPN services in the context of legal recreational uses. For example:1. Accessing a service from another country (subject to the terms and conditions of that service).2. Protecting your online security and strengthening your online privacy when abroad.We do not support or condone the illegal or malicious use of VPN services. Consuming pirated content that is paid-for is neither endorsed nor approved by Future Publishing.

• The Remarried Empress premieres: Tue, November 3 on Hulu (US)
• Global stream: Disney+
• Access Hulu and Disney+ via DIRECTV
• Use NordVPN to watch any stream

If you're a fan of K-dramas like Crash Landing On You and Queen of Tears, you're going to be obssessed with this latest export from Korea. Read on for how to watch The Remarried Empress online from anywhere with a VPN.

The Remarried Empress actually started life as a Naver Webtoon by Alphatart, but it was so popular that it got the live-action treatment. The show tells the story of Empress Navier, who is divorced from Emperor Sovieshu so that he can marry his mistress, but finds a new husband and an unlikely alliance with Prince Heinrey of the Western Kingdom.

After a successful premiere at the Busan International Film Festival on October 9, the show was initially planned to release on November 4, but that release date got moved up by a day...

Below we have all the information you need on where to watch The Remarried Empress online and stream the show from wherever you are.

Can I watch The Remarried Empress for free?

Fans in the UK could catch (at least some of) The Remarried Empress for free, if they sign up to a Disney+ free-trial.

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

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

Use a VPN to watch The Remarried Empress from anywhere

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

Using a VPN is incredibly simple:

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

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

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

How to watch The Remarried Empress online in the US

The Remarried Empress is available for free to US viewers via the DIRECTV 5-day free trial.

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

Have one of these subscriptions but away when The Remarried Empress is on? You can still access your usual streaming services from anywhere by using a VPN.

How to watch The Remarried Empress online in Canada, UK, Australia and worldwide

Viewers outside the US can watch The Remarried Empress on Disney+, and the show will land there on November 3.

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

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

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

The Remarried Empress FAQ

The Remarried Empress trailer

The Remarried Empress - who's involved?

• Shin Min-a as Navier
• Ju Ji-hoon as Sovieshu
• Lee Jong-suk as Heinrey
• Lee Se-young as Rashta
• Lee Bong-ryun as Eliza
• Choi Dae-hoon as McKenna
• Park Ho-san as Viscount Lotteshu
• Jung Young-joo as Albani
• Nam Yoon-ho as Baron Lante
• Lee Joon-hyuk as King Wotton III
• Kang Han-na as Christa

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

• Security researcher combined audio analysis, flight data and a lot of number-crunching to identify audio locations
• Aircraft that are inaudible on the soundtrack can still reveal the recorder's location
• The technique is apparently accurate to within 100m

If, like me, you grew up watching corny FBI shows where computers could do the impossible through made-up technology with a single command of "enhance!", you're going to love the apparently impossible thing that YouTuber Colsto has just described in a new viral video.

Using a Doppler effect technique (more on that below), the location of a video or audio recording can be identified to within 100 meters through a combination of inaudible audio, flight data and math. This is no movie magic or scriptwriter invention: it's real and it's a little bit scary too.

I've embedded the video below and would really recommend watching it, because Colsto goes into great detail about how it's done, what it can't do and what precautions you might want to take if you don't want your background to betray you.

How your video background can betray you

As Colsto explains, the key is the 'Doppler effect'. That's the way sounds appear to change based on the distance between a moving source and the listener. You hear it all the time: it's what makes fire engine and ambulance sirens, car stereos and airplane propellers seem to change pitch as they go past.

What's clever here is that Colsto has focused on the sounds made by aircraft in the backgrounds of videos. In many parts of the world the skies are very busy, and a faraway airplane that's inaudible to you or other listeners might still leave a sonic footprint on your video or audio that can be isolated and analyzed. That's definitely the case in the US data Colsto used.

Once you've got the audio, the aircraft sound can then be matched with publicly available flight data and Doppler curve measurements. By triangulating the sound from multiple aircraft, you can then identify where the audio was recorded. This is enormously clever.

Colsto explains the detail in the video — what aircraft data was used, how it was analyzed, how the Doppler measurements were filtered and what corrections were made — but isn't posting any software or other tools to actually do it for you. Instead, you'll get an assessment of preventive measurements including why normal signal processing such as high-pass filters or noise reduction can be ineffective.

How scary is this?

Colsto was keen to point out that the video is designed to be educational about open-source intelligence (OSINT) techniques, rather than an attempt to fear-monger.

"The odds that someone can and does actually do this is pretty slim. In reality, there's probably a hundred easier ways to find you than this," he says.

However, these are some potential use cases. "The audio we used is just from some random security camera, but it could have just as well been footage from a investigation or a video online. Something with literally zero visual clues could be located using this," he added.

So despite the somewhat niche application, there's something deeply scary about this kind of thing, I think, and from similar revelations such as Wi-Fi routers that can be used to locate you indoors and Wi-Fi signals that can help track your movements, even if you're not carrying a phone that connects to them.

And I know that fear is probably a bit silly when everywhere I go I carry a phone, use a tablet or drive a car that's leaking identifiable information to God-knows-who. But I can't help asking the question: if the possibilities we know about are this scary, what on Earth is the tech we don't know about capable of?

Thinking of buying a new TV?

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

Bambu Lab has launched its Black Friday sale early this year, with running deals from now until December 1. There are plenty of 3D printers and accessories to choose from, but what's actually worth picking up in the sale?

Well, I asked our 3D printing expert Alastair which models he recommends from one of our top 3D printer makers. Alastair has tested every Bambu Lab 3D printer in his workshop for our best 3D printer guide and I wanted to know which machines he'd personally choose for readers. Here's what he suggests.

• For small business: The Bambu Lab X2D, which scored 4.5 stars in his review, " delivers exceptional print quality, material compatibility, and all at a superb price."

• For cosplayers and creators: The 4.5-star Bambu Lab H2S offers "enterprise-level print features and volume" with support for engineering materials.

• For beginners on a budget: The Bambu Lab A1 Mini earned a flawless 5 stars in his review, with exceptional speed, accuracy, features, and price.

Bambu Lab 3D printer deals

• Read his full Bambu Lab X2D review

• Read his full Bambu Lab H2S review

• Read his full Bambu Lab A1 Mini review

Bambu Lab filament sale

Alastair has also offered up some advice on which filaments he prefers. He says:

"Spend as much as you can. Filamentum, ColorFabb, and Polymaker are all great third-party filaments that I use. Otherwise branded filament from Bambu Lab and Prusa are good. esun and Sunlu as budget brands."

Bambu Lab is currently running a sale on filaments for Black Friday right here.

• The next game in the Elder Scrolls series is officially titled: The Elder Scrolls 6: Sentinel
• Bethesda Game Studios has also shared a teaser trailer hinting at the game's location, which appears to be Hammerfell
• The studio says, "It’ll be some time before we’re ready to share more" about Sentinel

Bethesda Game Studios has finally announced the name of the next game in its long-running Elder Scrolls series: The Elder Scrolls 6: Sentinel.

The news comes from a video celebrating the studio's 40th anniversary, which highlighted its many achievements and ended with Bethesda creative director Todd Howard offering a new look at the game.

"Everybody’s always asking about Elder Scrolls 6," said Howard. "And we thought, hey, given the moment, what a perfect time to tell you what The Elder Scrolls 6 is actually called."

The teaser trailer is short and doesn't showcase gameplay, but the camera pans over sand before a magic-looking scroll emerges. Then the game's title, as fans theorized, "The Elder Scrolls 6: Sentinel," was revealed.

The trailer description also provides an additional hint to the game, which reads, "The scroll offered only a shimmer, a veil thinly lifted to my eyes. A future lost. A victory undone. Sands of time scattered by fate," which is written in game lore by "Yazar, Moth Priest to High Queen Iradne of Hammerfell (4E 206)."

The Elder Scrolls 6: Sentinel has been in development for a very long time. Announced all the way back in 2018, today's teaser trailer and name drop are the first new pieces of information we've received about the game since.

In a follow-up post on X, Bethesda seemingly confirmed the game's location is set in Hammerfell, but it will be "some time" before more details

about the role-playing game (RPG) can be shared.

"Thank you for 40 years of shared memories," the studio said. "We can’t wait to make even more together with The Elder Scrolls 6: Sentinel. Development continues, and while it’ll be some time before we’re ready to share more, it feels great to finally reveal the name and location with you."

• Anthropic workers are likely making a good salary, plus stock-based compensation
• Revenue per employee is more than 2x Meta's and 2.6x Google's
• The company is targeting a valuation of around $1.5 trillion

Anthropic has confirmed in a recent S-1 filing that it spent $1.711 billion on personnel in its most recent three-month period, down slightly on the previous quarter's $1.857 billion but up around 3.5x compared with the same quarter in 2025 ($489 million).

With around 5,300 people on the books, it means the average worker set the company back over $322,000 for the quarter, or over $107,000 per month.

While these figures are likely wild overestimates for the average worker's pay, they show just how much the AI giant is having to spend on talent compared with last year.

Anthropic is spending billions on workers' salaries

Crucially, the average worker did not receive $107,000 in a single month, because these figures also include bonuses, employee benefits, payroll taxes and stock-based compensation. An estimated $1.1 billion of the quarterly total came from stock-based compensation, accounting for nearly two-thirds of the total sum.

Still, personnel costs accounted for around 15% of the company's revenue over the first nine months of 2026, which puts Anthropic in a positive position. The company spent $4.8 billion on personnel in the first three quarters.

Flip those figures around and we have a company that's making an estimated $6.6 million in revenue per employee, compared with $3.2 million at Meta and $2.5 million at Google, per Business Insider calculations.

Recruiting AI talent is expensive

Anthropic doesn't publish salary information, but given its huge expenditure relative to the number of employees, it's clear that recruiting AI talent is expensive. With stock-based compensation accounting for two-thirds of Anthropic's personnel expenditure, workers are likely being attracted by the potential of huge future wealth, rather than immediate compensation.

An upcoming IPO has Anthropic valued at somewhere around the $1.5 trillion mark, per recent Reuters reporting. That would put it ahead of OpenAI, which is now likely targeting a $1.25 trillion valuation, per a separate Reuters report.

• Rockstar Games has confirmed players can pre-load the digital version of Grand Theft Auto 6 on November 10
• Code-in-box editions should be available in stores starting November 12
• The pre-load file will be "up to 212GB depending on your device"

Rockstar Games has announced the file size for Grand Theft Auto 6 and the dates for when players can pre-load the game depending on their edition.

In a new blog post, the studio confirmed that digital pre-loading on PlayStation 5 will begin nine days before the game's November 19 launch, on Tuesday, November 10 at midnight local time "in your respective time zone."

For Xbox Series X and Xbox Series S, pre-loading begins globally on Tuesday, November 10 at 12AM GMT / 7PM EST / 6PM CST.

Meanwhile, code-in-box versions for both consoles should be available from retailers beginning November 12.

GTA 6 was speculated to be a massive game already due to its ambitious scale and impressive features, but Rockstar has officially confirmed that the pre-load file will be "up to 212 GB depending on your device."

"So make sure you've got ample space to begin pre-loading as soon as possible," the studio cautioned. Say goodbye to a fair few of your games if you want to install GTA 6.

In addition to the new details, Rockstar said it will be adding an official printable Grand Theft Auto 6 cover art poster and map of Leonida as a free digital download to the official site on November 19 for fans.

"A double-sided physical version of this poster will also be made available shortly after launch from the Rockstar Store — for $0.99 on its own or as a free gift with any purchase from the Rockstar Store," it added.

Rockstar recently released the cinematic GTA 6 launch trailer, which offers a deeper look into the game's story. Elsewhere, a new report claims the PC version could launch in early 2027.

You can definitely tell that Black Friday season is upon us, as big discounts on brand-new tech are popping up almost every day now. The latest early deal I've spotted is this record-low price for the Google Pixel Watch 5 at Amazon, which brings the latest version of the popular wearable down to $299.99.

Outside of a smaller $50 discount at the start of October, this is only the second time the Pixel Watch 5 has been on sale since its launch in August. And I don't expect it will get any cheaper than this over the next couple of months — even with Black Friday technically still over three weeks away.

The offer I've picked out here is for the standard 41mm model with Wi-Fi support only. However, the larger 45mm version and the watches with LTE support are also discounted by $100 right now at Amazon.

So, if you prefer a bigger screen or want your watch to be able to receive messages and alerts while away from your phone, you can choose between those options as well. For most, though, the standard 41mm version does everything you need in terms of health tracking, fitness tracking, and phone connectivity — and it's excellent value for money.

Google Pixel Watch 5 — record-low price

We called the Google Pixel Watch 5 "the best Android watch yet" in our four-star review.

While it does feel like a clear iterative upgrade over the Pixel Watch 3 and Pixel Watch 4, it's definitely a top choice for anyone picking up a wearable for the first time. It focuses on excelling in the key areas that matter, including reliability and accuracy in metric tracking, offering a comfortable fit and attractive design, and adding advanced features for those who need them, like Breathing Emergency Detection and Insulin Resistance Trends.

The updated Google Health app puts a lot of emphasis on AI, so your mileage may vary on how much you appreciate some of the more advanced health and fitness features that rely on your data, but this is a very capable smartwatch for Android users — now at an even better price.

• The Razer Raiju V3 Pro Signature Edition launches November 10 for $349.99 / €368.99
• The licensed PS5 controller features a magnesium alloy exoskeleton and an open polygonal top and back plate design with chamfered edges
• The controller is also optimized for PC and boasts "reliable precision, exceptional control, and blazing-fast speed"

Razer has announced the Raiju V3 Pro Signature Edition, a new version of its officially licensed PlayStation 5 controller built on a decade of experience for "the highest level of performance and craftsmanship."

Launching exclusively at Razer on November 10 for $349.99 / €368.99 in an initial limited release supply of 1,337 serialized units, the Raiju V3 Pro Signature Edition is Razer's "most ambitious" PS5 controller to date and combines materials chosen for "strength, durability, and finish."

The jet-black controller features a strong magnesium alloy exoskeleton, and the handles are wrapped in matte black textured TPU grip that resists oils and abrasion for a pristine finish for extended play sessions.

Layered over the touchpad, the Razer logo is laser-etched on a semi-transparent, tinted glass panel, which offers a subtle view of the open-air polygonal design underneath.

The Raiju V3 Pro Signature Edition also has an open polygonal top and back plate design with chamfered edges, and its nano coating leaves the internal elements exposed beneath the shell, keeping them safe from debris.

On top of that, the Razer controller boasts "reliable precision, exceptional control, and blazing-fast speed," which is achieved through its symmetrical thumbstick layout and contactless TMR thumbstick sensors for high-precision aim.

Built for PS5 consoles and optimized for PC with a 2,000Hz polling rate thanks to Razer HyperPolling Wireless for sharper movement and tracking, the controller also offers six mappable multi-function buttons, two claw-grip bumpers, and four mouse-click back buttons.

"With the Raiju V3 Pro Signature Edition, we set out to illustrate what happens when Razer’s engineering and design capabilities are pushed to their absolute limits," said Charlie Bolton, Razer's global head of design. "It draws on years of experience in high-performance controllers, delivering tournament-grade responsiveness in a design that feels like a statement piece in hand."

Each Razer Raiju V3 Pro Signature Edition offers a suite of premium accessories for customization, including a premium leather carrying case, alternative thumbstick caps, four swappable back button covers, an anodized aluminium back button replacement tool, a Razer HyperSpeed Wireless Dongle, and a USB Type-A to USB Type-C cable.

FastestVPN has launched its exclusive Black Friday offer for TechRadar readers: a lifetime of VPN access and password management for just $27.

FastestVPN usually rolls out a discount to its Lifetime Plus Plan each November but this is the cheapest and best we've ever seen it. In previous years the exclusive has cost $30 and only included two years of the family password manager. Now, the price is lower, and access to both is free for life.

You'll need to use the code FUTUREBF2026 at checkout to make sure you secure the extra $10 discount. Otherwise you'll find the price stuck at $40.

We don't rate FastestVPN as highly as the likes of others on our Best VPN list but it's hard to ignore at this price, when most cost upwards of $60 for 2 years and don't come with a password manager.

FastestVPN still comes with many of the features you'd expect of a VPN, such as a kill switch, P2P optimized servers, an ad-blocker and some malware protection too. It has servers in over 49 countries worldwide and will allow you to stream from BBC iPlayer, Netflix and free regional services such as 7Plus and 9Now in Australia.

Unlike the name suggests, it's not as quicker as many other VPNs, such as Mullvad and VPN Super Unlimited, as we found in our tests. But FastestVPN is private and secure enough for most use cases.

If you're more focused on VPN performance, the best alternative VPN deal right now is Proton VPN. It's $2.49 per month on a two-year plan, and brings sturdier security, faster speeds, and far more reliable unblocking performance with streaming platforms.

There are also other Black Friday VPN deals from top performers like Surfshark and NordVPN too.

Fancy expanding your Lego collection, or picking up a set or two to give as a gift? You don't need to wait until the end of the month to do so, because it looks like the Black Friday Lego deals are here already. A number of major retailers — including Currys and Argos — have launched their sales events already, with Black Friday Price Promises to reassure you it's safe to buy now.

Included in these sales are dozens of excellent price drops on a huge selection of Lego sets, from happy plants and retro gaming classics, to modern-day animated icons and shapeshifting spooky creator sets (what do you mean, Halloween is over?). I've scoured through all the sales at a range of retailers, and rounded up my favorite Lego discounts below, at prices that range from stocking-filler / Secret Santa to blow-the-budget.

Prefer to browse yourself? Use my quick links below. Alternatively, scroll down to see my top picks.

• Amazon: Discounts on Lego sets for kids and adults
• Argos: Up to 1/3 off, with Black Friday price promise
• Currys: Up to 30% off ahead of Black Friday
• Very: Dozens of deals with Black Friday price promise

The best early Black Friday Lego deals

The UK growth debate often focuses on tax, regulation, and investment incentives. While incredibly important factors to the economy, they can obscure a more immediate constraints sitting inside many businesses; one of the most significant being the payment systems that determine how quickly businesses can collect, move, and reconcile money.

The CBI has made clear in their 2026 Roadmap to Growth that business leaders want to innovate but are being held back by costs and slow delivery. Payment operations are one of the places where that tension shows up most clearly. A business can have ambitious growth plans, but it cannot deliver on them if moving money still depends on manual workflows, slow approvals, and disconnected systems.

While the business case for improvement is often understood by leaders, execution is a different hurdle. Our research found that 91% of business leaders recognize that a payment system can drive overall business growth, but only 27% of firms currently use a modern payment system, defined as having advanced automation and integration features.

These inefficient payment processes ultimately hold back productivity across the wider economy. When companies cannot move money with speed and certainty, growth becomes harder to plan and slower to deliver.

The growth gap inside payment operations

Businesses are increasingly operating in an environment where customers, employees, and suppliers expect transactions to happen quickly and accurately. For example, 79% of UK consumers expect to receive disbursements immediately. Processes designed around batch payments, manual approvals, and fragmented reconciliation are no longer good enough. They may still function, but they slow businesses down.

The cost appears in smaller, repeated points of friction. A delayed supplier payment can weaken a commercial relationship. A payroll correction can damage employee confidence. Manual reconciliation consumes finance teams’ time that could be better spent on analysis, forecasting, and control. These issues not only impact operations but can also have a material impact on growth.

Leading organizations increasingly view payment operations as a source of competitive advantage rather than an administrative necessity. Growth depends on the ability to move money accurately, quickly and at scale.

Open finance raises the standard and sets the direction

If payment operations are becoming a competitive advantage, open finance is raising the stakes even further. The FCA’s open finance roadmap sets out a vision for unlocking faster and more personalized financial services, strengthening competition and supporting wider economic growth.

That future will expose the gap between businesses that have modernized their payment operations and those that have not. Better access to financial data is valuable only if the organization can convert that insight into action. A business may be able to assess cash flow more intelligently, understand customer risk earlier or access finance faster, but if its payment processes still rely on manual intervention, the benefit will be limited. Open finance can point the way to better decisions, but modern payment infrastructure is what allows those decisions to translate into faster growth.

The UK has already shown that data-enabled financial services can scale when common standards, trusted infrastructure, and market adoption come together. Open Banking Limited reported that the UK open banking ecosystem had recorded more than one billion open banking payments by July 2026.

These figures are important because they show open banking is no longer a niche innovation layer, it is becoming expected infrastructure that consumers and businesses use to move money and share account data. The challenge now is to turn that personal finance scale into practical value for businesses.

Changing the game in collections: cVRP moves the dial on open banking

Looking towards collections for an example of this practical value: a key next stage of the rollout of open finance is the move from one-off open banking payments into recurring, automated collections. Most open banking payments to date have been single transactions which are useful for a specific purchase, transfer or account-to-account payment, but less suited to the recurring flows that keep many businesses running.

The UK Payments Initiative (UKPI) has recently launched an industry-led scheme that marks a major development for businesses that collect at scale and the experience they can provide their customers. Commercial variable recurring payments (cVRP) allow businesses in specific sectors to collect varying amounts from a customer's account on a regular basis following a single authorization by the customer, all whilst giving the customer greater flexibility and control.

With cVRP, businesses can receive real-time confirmation through Faster Payments and reduce reconciliation uncertainty. That means better cash visibility, fewer failed collections, and more confidence to grow for UK firms.

The growth opportunity is clear in sectors where collections are frequent, high volume or variable, including utilities, subscriptions, and lending.

Getting payments right is a growth strategy

Payment optimization is not a narrow finance project. It is a practical route to better productivity, stronger relationships and lower operational risk. It helps businesses protect cash flow, reduce errors and free up teams from avoidable administration.

Getting payments right may not be the most visible growth strategy but it can be one of the most immediate wins that also has a significant long-term growth impact for your business.

We've reviewed and ranked the best mobile payment apps.

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

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

• Cybercriminals are stealing AI API keys and cloud credentials to hijack expensive LLM services and infrastructure
• Stolen access is used to run AI workloads, create proxy services, and rack up massive bills for victims
• Google warns “LLMjacking” is evolving into a specialized cybercrime ecosystem with dedicated monetization operators

Threat actors are using large language models (LLM) to support almost every stage of a cyberattack, from researching potential victims to developing malware and automating various attacks. However, using AI is not cheap, and for those eager to use frontier models, costs can stack up relatively quickly.

But cybercriminals are used to stealing things, so why not steal someone else’s tokens, too?

A new report from Google’s Threat Intelligence Group (GTIG) says that different hacking groups are doing exactly that and that the practice - known as LLMjacking - has been steadily growing more popular over the last couple of years.

Most groups start by focusing either on publicly exposed LLM endpoints, or on obtaining valid LLM application programming interface (API) keys. In many cases, the crooks would use exposed or stolen cloud identity tokens (IAM credentials, Service Account keys, GitHub Personal Access Tokens, and active session tokens) to establish an initial foothold into the target organizations’ infrastructure.

From there, they would “systematically work” to escalate privileges, solidify their access, and identify targets for further compromise.

LLMjacking

The endgame is LLMjacking, which GTIG describes as “the monetization and resource-hijacking of foundation models, which actors achieve through token proxying, automated model streaming, and GPU hardware provisioning.” In other words, someone else is using your AI subscription while you get the bill.

The bills can cause quite the headache, too. With cryptojacking, for example, token mining is relatively slow and steady, and a certain amount of time needs to pass for the charges to really stack up. With LLMjacking, however, costs can go into six digits in six minutes.

“Upon acquiring LLM API access keys, UNC6993 attempted to generate additional API keys and enable associated services within victim cloud projects, quickly accumulating over $100,000 in billing charges,” Google says in the report.

“Another threat actor subscribed victim accounts to Pay-As-You-Go agreements, selecting premium foundation models and launching high-volume inference streams. To bypass rate limits, the threat actor deployed a "burn and rotate" strategy, continuously creating and deleting short-lived service accounts while auditing the cloud provider’s billing interface to monitor financial impact.”

In some cases, threat actors also used the acquired access to host their own proxy endpoints and agentic workloads. Google says that one threat actor enabled agents, deployed publicly accessible reverse proxies, and used custom Node.js proxy bundles to expose endpoints compatible with multiple frontier LLMs rebranded under a persona known as D1337-KAIROS.

This actor deployed a containerized backdoor which queried an upstream LLM to reason through instructions and execute commands locally.

A growing ecosystem

The LLMjacking ecosystem is young, but growing fast and maturing even faster. Google says that there is already a division of labor between hacking groups that establish initial access to compromised environments, and operators who manage downstream monetization infrastructure. The latter are called “LLM Mixers” and they operate as a routing gateway that proxies, aggregates, and resells unauthorized LLM access to underground subscribers.

“In one investigation, GTIG observed two distinct operational clusters coordinating to compromise and monetize enterprise LLM infrastructure, demonstrating the sophisticated specialization emerging in this space,” Google said.

By using open source API gateways and custom relay infrastructure, the attackers were able to pool stolen credentials, meter usage, and hide where the requests really originated from.

Google is now sounding the alarm, saying it expects threat actors to “increasingly prioritize” LLMjacking at the expense of other resource theft practices. It urges organizations to increase their visibility of LLM platforms, prioritize detection of modifications to LLM accounts and services, and treat any unauthorized enablement as a critical security event. It also advised organizations to enforce the principle of least privilege and make sure accounts and users only have permissions they absolutely need to get the job done.

We all know Awkwafina as a comedic actress and a prolific voice actor (having voiced characters in numerous animated projects like Raya and the Last Dragon, The Bad Guys and most recently Wildwood), but she's now trying her hand at cooking while also learning more about her rich family history.

Across eight episodes, we'll see Awkwafina — or Nora Lum as she's actually called — travel across the States learning how to cook meals from scratch, with the aim of eventually reviving her family's long-closed Cantonese restaurant in New York.

Thanks to Apple TV, Awkwafina will get the chance to rub shoulders with acclaimed chefs includin Roy Choi, Martin Yan, and Cory Ng, and she will also visit the Culinary Institute of America. She will also be judged fairly harshly by her grandmother, but you can't win them all.

If you're ready to see her try her hand in the culinary world, read on for how to watch Awkwafina: The Unlikely Cook online and from anywhere.

How to watch Awkwafina: The Unlikely Cook from anywhere

For those away from home looking to watch Awkwafina: The Unlikely Cook, you may find yourself unable to watch the show like you normally would due to annoying regional restrictions. Luckily, there’s an easy solution.

Downloading a VPN allows you to stream online, no matter where you are. It's a simple bit of software that changes your IP address, meaning that you can access on-demand content or live TV just as if you were at home.

Use a VPN to watch Awkwafina: The Unlikely Cook online from anywhere:

How to watch Awkwafina: The Unlikely Cook around the world

Where to watch Awkwafina: The Unlikely Cook online in the US, the UK and everywhere else

What you need to know about Awkwafina: The Unlikely Cook

Awkwafina: The Unlikely Cook trailer

Who is involved in Awkwafina: The Unlikely Cook?

• Awkwafina (star and producer)
• Roy Choi (guest chef)
• Martin Yan (guest chef)
• Cory Ng (guest chef)
• Zhan Chen (guest chef)
• Jimmy O. Yang (co-star)

We test and review VPN services in the context of legal recreational uses. For example:1. Accessing a service from another country (subject to the terms and conditions of that service).2. Protecting your online security and strengthening your online privacy when abroad.We do not support or condone the illegal or malicious use of VPN services. Consuming pirated content that is paid-for is neither endorsed nor approved by Future Publishing.

• Cathedral Systems will offer a new perspective into OpenAI vulnerabilities
• Partnering with third parties sets 'an important example' for the industry
• Bug bounty payouts set to increase 4x for critical findings

OpenAI has announced a partnership with Cathedral Systems to bring security researchers from both organizations together in a bid to test its AI infrastructure, identify potential vulnerabilities and improve existing defenses.

The ChatGPT maker hopes that enlisting a third party with add fresh perspectives to its existing research, and it's a clear sign that the company knows future threats could pose a serious issue.

"OpenAI is setting an important example by partnering with outside security firms and using its own models to continuously test and strengthen its defenses," Cathedral CEO Gavin Kliger wrote.

OpenAI is preparing to serious enhance security

The company is also improving its bug bounty program to make it more attractive for individual researchers to get involved. Though exact rewards were omitted from the announcement, OpenAI did say it was prepared to quadruple the payouts for critical findings.

When OpenAI launched its bug bounty scheme, it said it would be looking for applications across software patches, model privacy, threat detection and response, security tool integration and AI agent security.

It's not the first time the AI giant has led a major security initiative – it previously asked a quarter of its developers to set aside some time to find and fix security weaknesses. Naturally, they used Codex to speed up investigations and fixes.

"We’re building on that work because traditional cyber defenses alone are no longer sufficient," the company said, referring to its previous sprint.

Much like Anthropic's Project Glasswing, OpenAI promises to use its own frontier models to strengthen its own systems and those of others.

"Keeping the defender’s window open will take action across the industry, and that work needs to start now," the company added.

• Apple will place usage limits on Siri AI in the future, with the option to pay for expanded access
• What the cost will be or how pricing will be structured remains to be seen
• On-device inference will still be free, but the heavier tasks such as AI image editing will count towards your daily usage limit

In the Apple earnings call for Q4 2026, new CEO John Ternus joined CFO Kevan Parekh and COO Sabih Khan to discuss the results of the previous quarter. Within the discussion the trio quietly confirmed that Siri AI may soon have usage limits.

Now, we’ve put Siri AI through its paces and found it to be a remarkably good life assistant across Apple’s fleet of devices after years of Siri directing tricky questions to ChatGPT. But a usage limit could seriously hinder the effectiveness of your Apple assistant.

Those looking to use Siri as their go-to resource for AI answers and make full use of Apple’s AI photo editing features will have to pay for expanded access. So far, there are no details on how much users will have to fork up at each iCloud+ tier to make full use of Siri AI.

Server-side inference could raise costs

“The mix of [on-device and cloud Siri] is very early days, and so it's hard to predict exactly what the mix of those costs are going to look like,” Parekh explained. “Obviously, on device, you know, it's very very efficient. No real costs associated with that. But on the server side, we have some inference costs.”

Siri AI has had a somewhat limited rollout, being released mid-September in English as part of the iOS 27 update. Siri AI runs on a mix of on-device and Private Cloud Compute, so until Apple gets to grips with how much demand they might expect from Siri AI users the costs aren’t really known.

“I think again, we've just recently launched it and made it available to more users,” Parekh added. “And so we're in the very very early stages of understanding the kind of mix of those costs, as well as further requirements we might have on the compute side.”

Obviously there is some reasoning behind Apple’s decision to use Private Cloud Compute for some tasks. Given that your phone holds some pretty sensitive personal information that Siri AI may need to access, Private Cloud Compute offers a way to process that data in an anonymized, privacy-focused format.

Much of Siri AI’s inference happens on-device, but tasks that exceed the resource restrictions of an iPhone can be offloaded onto Private Cloud Compute. But with that comes processing costs that Apple has to cover.

“As far as agentic capabilities, you know, we think these are promising tools for users,” Tetanus said. “We just want to make sure that users are aware of what they're giving access to. We want to make sure this is done in a way that protects security and privacy.”

“You know, whether it's your contacts, your communications, health data, financial data, all of that, and Siri can tap into that and help you access it in a way that's very secure and private, and that nobody else has access to.”

What will you actually get for free?

As far as using Siri AI for free goes, light tasks that run on-device should remain accessible to all users. But if you have a more complex task or want to make use of Apple’s AI image editing features that need to be performed on more robust hardware, you could end up hitting the daily usage limit.

As Apple disclosed in the footnotes of the Siri AI release, the use of Apple’s cloud models specifically is what will count towards the usage limit. “Daily limits may vary by feature, request complexity, system demand, system policies, and other factors. Expanded access to such features will be available for a fee in the future.”

What these costs look like remains to be seen, but just drafting an email or asking Siri to find a video your friend mentioned in a message a few days ago shouldn’t be too much of an issue.

However, be aware of utilizing any features that require the use of Apple’s AFM 3 Cloud models moving forward, such as the intelligent photo editing tools and Image Playground.

Buying a printer is one thing, but keeping it supplied with ink can quickly become massively expensive, especially if you regularly print documents, homework, or photographs. Epson's EcoTank ET-2980 is one of our favorite cheap ink tank printers, saving a fortune on replacement ink cartridges.

Now, the value-led cartridge-free printer sees a price cut down to $285 (was $330) at Amazon, while in the UK, it's listed under the name EcoTank ET-2950, down to £230 (was £280). It's the same model, though. And the price delivers an even better deal on the machine we called "a bargain of this bottle-fed printer."

• Best for: Households, students, and home office workers who print a lot and want to keep ongoing ink costs down without loss of everyday printing, scanning, and copying capabilities.

• Why it matters: You get refillable ink tanks with up to three years of ink included. There’s also Wi-Fi, Wi-Fi Direct, a 100-sheet paper tray, automatic double-sided printing, and borderless photo printing.

Why we recommend it

The biggest attraction of the EcoTank ET-2980 is Epson's approach to running costs. The included ink can produce up to 6,600 black pages or 5,500 color pages, and each replacement bottle set is equivalent to approximately 90 individual ink cartridges.

Refilling is straightforward, thanks to Epson's EcoFit bottles, which have keyed nozzles designed to help prevent you from putting the wrong color into a tank.

In his four-star review, our expert Jim said it: "printed reliably on everything from envelopes to thick 300g/m² cardstock. It’s not a fast printer, but automatic duplexing is a welcome convenience. Pigment black ink produces crisp, smudge-resistant text, while dye-based colors look bright and punchy. Photographs are slightly soft and oversaturated towards red, but perfectly acceptable for occasional use."

He added that: "The biggest attraction is the ink economy. Epson is putting plenty of ink in the box and replacement bottles are cheap. I think its low running costs, reliable performance and essential duplexing feature make it excellent long-term value."

For more top performing units, see our guides to the best home printers and best ink tank printers.

Price context & historical value

Amazon's current $285 asking price is $44.99 below how much it usually sells for. That's not the cheapest we've seen it sell for in the past though. It was as low as $229.99 earlier in the year, and it's possible it will be reduced more the closer to Black Friday we get.

The British equivalent model, the EcoTank ET-2950, is usually priced at £279.99 and is currently down to £229.99.

If you want to save on other units, see our guide to the best printer deals we've seen this week.

Should you buy it?

✅ Buy the Epson EcoTank ET-2980 if...

You regularly print documents, school assignments, or photographs and want to reduce your reliance on ink cartridges. The ET-2980 also offers automatic double-sided printing, which can help cut paper consumption.

❌ Skip the Epson EcoTank ET-2980 if...

You only print occasionally or need a dedicated business printer with advanced document-handling features. A cheaper cartridge-based model may be more economical for light use.

• See all ink tank printers at Amazon

It was inevitable that there would be a successor to the Bambu Lab H2C, which was no small machine; however, the Bambu Lab H3C takes scale and ability to the next level, although the innovative Vortek Hotend Change System is no more.

Like the H2C, there’s the multi-tool functionality, but as you’ll see from the images, the system is all new, with dedicated tools in the Swaptek Tool Changer rather than the Vortek hotend-changing system, so something closer to the Bondtech INDX tool changers.

An interesting update is the ability to print with different-sized nozzles, so a fine 0.2 mm for fine detail, going to a 0.8 mm for infill and areas where speed is preferable to fine finishes. What I have discovered through the month that I’ve had the printer so far is that this is an evolution of the 3D printing process, refining the techniques and while this technology and its application are still being ironed out, the full potential is apparent.

Adding to the multi-diameter printing is the advanced TPU handling. This essentially enables you to print with thicker 2.85 mm filament using the PrintTool hotend. As the filament is thicker, it supports softer TPU, and again, there’s a learning process behind its use. It’s not quite as simple as popping it in and letting it go, but it is a step in the right direction.

Sitting on top of the machine is the new AMS 3H filament management system that holds up to nine filaments, with two slots enabling integrated drying and bypass feeds for specialist materials.

I’ll admit, as the boxes kept arriving and the H3C started to take shape on the workbench, the machine did seem to dominate the space. However, after squeezing in the H2C next to it, while the new AMS gives a visual impression of scale, the footprint of the two machines is much the same.

The big question for me at the start of this test was whether this new machine and design was a direct replacement for the H2C or something altogether new. It certainly initially looks far more complex, but then, as I became more familiar with the machine, the full tool changer design, once assembled, did seem, a little more simplistic and certainly more reliable as I pushed the print times and complexities.

One point to note is that throughout the test, while I had the machine and seemingly every accessory that I could need to go with it, the actual official machine specifications and final tweaking of the software were still to be released.

What Bambu did supply was a few test models and an extensive set of instructions about how to tackle each of the main new features. There was also a set of complex models that I used to start the test before deviating to some of my own, and then back to the supplied models to finish the test.

I’ll cover more of those tests in the performance section, but what interested me was using this printer for a real practical task. This is, after all, a business-class machine, although the price will put this just in range of the keen enthusiast as well.

The main project for this test was to print and build a set of speakers from Print Your Speakers, with the Supersonic MK3 kit fitting the bill for this test. Essentially, a solid PLA outer with TPU to add damping to the infill, and then a switch of nozzle sizes for the fine detail of the outer wall and a thicker nozzle for the inner walls.

This is where the learning curve of this machine and the Bambu Lab software really comes into play, and while that learning curve is steep.

As the test ran on, the potential of this machine started to grow. While multi-filament and tool changer machines now dominate and Bambu Lab’s lead in the field was quickly fading, especially with the Original Prusa CORE One INDX, of which the H3C’s new system bears more than a passing resemblance. But with support for 2.85 mm TPU and multiple other materials, it now pushes them back out in front and truly offers something unique, although there are still a few refinements needed.

Bambu Lab H3C: Price and availability

The H3C will be available to purchase from the Bambu Lab website from 3 November 2026, following the official launch. Unlike the H2C, there is just one configuration available at launch, with the AMS 3H and all nine PrintTools included.

• H3C AMS Combo: $2,499 / £2,199
• Combo + PLA Palette + TPU: $2,589 /£2,269
• Combo + PLA Palette + TPU + Bambu Station (Enclosed) + E1 Air Purifier:
• $3,049 / £2,699
• Combo + PLA Palette + TPU + Bambu Station (Open): $2,799 / £2,469

Bambu Lab H3C: Design

The Bambu Lab H3C is a 3D printer on another scale, and whilst it doesn’t look that much larger than the H2C initially, once you start to work with it and try to move it around the workshop, you suddenly realise that its colossal 36.28 kg weight and physical dimensions of 493 × 571 × 587 mm make it more of a two-person lift than something you’re able to handle on your own.

That would explain why Bambu Lab also sent a special trolley along with the machine, so it’s easy to access and move around the workshop. The issue at the moment is that there are so many 3D printers coming out that my workshop doesn’t have any space, so this will be something that I will manoeuvre the printer onto once I can find a pathway through the workshop to get the two sections to meet.

As it was, I decided to set this machine up in the photo area, and I’ve already manoeuvred the H2C next to it so that I can do a side-by-side comparison on size and scale, however, again this left little room as the the scale of these machine is rapidly increasing.

As I first lifted the H3C into position, the two machines looked roughly the same size, but as I started to add the huge new AMS 3H on top, I realised that this was an altogether different proposition.

As I unpacked the machine, it was obvious that Bambu Lab had taken a completely different direction to multi tool head printing compared with the H2C, and the H3C has something more akin to the INDX tool changer. You can definitely see the similarities. As I went through the initial setup, what was surprising here was just how easy and straightforward it was, and whilst the machine was large, the actual placing of the AMS on top and positioning all of the nozzles wasn’t at all difficult.

The only thing that you do need to make sure of is that there are nine tools included: eight with 0.4 mm PrintTools for 1.75 mm filament and one dedicated 0.4 mm PrintTool for 2.85 mm filament. Careful reference to the instructions as you’re setting up is essential to make sure that this nozzle goes in the right place.

Before taking a closer look at this new system, it’s worth pointing out that the AMS 3H is quite an evolution on the AMS 2, sitting on top of the machine and really integrating more closely with the design compared with the add-on box of the AMS 2. This, I know, will please a lot of print farms, which, whilst generally fans of the Bambu Lab machines, find the AMS 2 is often a weak point and the part that they have to replace most often. One point here is that, of those nine slots, there’s a separate section within the rack which enables just two of those nine slots to be actively dried.

As we’ve seen before, you can also add more AMS units, enabling up to 24-colour printing, up from the nine possible out of the box.

Once I got everything set up and installed, which, to be honest, didn’t take too long, it was then time to connect the machine to the local network via Wi-Fi 6. As ever, this was as straightforward and simple as with all other Bambu Lab machines. One of the surprises is that a machine of this size and scale, especially aimed at the professional market, doesn’t feature an Ethernet port.

The other point here is that as soon as it connected, it ran through a series of updates before then going through the calibration process. So, while the machine itself is really well designed and easy to set up, the software updating and full calibration process can take up to an hour, which is slightly infuriating when I was quite excited to see what this machine was capable of doing.

After the Wi-Fi connection, calibration and everything else had been run through, it was time to connect through the Bambu Studio software, and again, a few clicks through the beta release that we were given, enabled me to connect quickly to the printer.

However, as always, before running through my own prints and some of those others that had been supplied to fully test out this new machine’s abilities, I ran through the standard test prints on the machine. These are the usual 3D Benchy and variants, alongside scrapers and a few other bits and bobs, as expected. It ran through all of those prints flawlessly, and then it was time to cross over to Bambu Studio.

Bambu Lab actually supplied some very interesting test models, which I go into in the performance section, but before I went into these multi-filament, multicolour designs, I decided to test out some of my more standard prints. This just enabled me to double-check the calibration and also try out a few different materials.

There are two built-in 1080p cameras, with an optional Bird’s Eye camera available as well. The chamber heats up to 65°C, with a 120°C heated bed and a 350°C maximum nozzle temperature.

Leaving the machine just to run through those few prints, I was able to step back and take a proper look at the machine, and it is an evolution of what we’ve seen with the other H series printers, just slightly larger. Again, this is a steel chassis, which gives it structural support and also makes it insanely heavy to lift around the workshop.

The outer is plastic, with a grey and black exterior that looks stylish, just in the same vein as we’ve seen with the other Bambu Lab hardware. On the front, you’ve got a big viewing window so that you can see what’s going on with the print, and whilst the top-mounted AMS 3H does stop you from looking down into the machine, it's integrated rather than being a separate unit, which I personally really like.

As we’ve seen before, there’s a really nice 5-inch colour touchscreen, with an 854 × 480 display resolution, and this is very responsive, giving me full access to all the functions and settings. However, it is getting more complex, especially with those nine toolheads. It does take a little bit of extra time just to orientate yourself with all of those different settings.

Across the front is the main feature of this new 3D printer, obviously the new nine-toolhead rack, and what’s interesting here is that the rack raises and lowers, so if you do want to see what’s going on on the print bed, it’s far easier than with some of the other multiple systems of this style out there. At all times, however, you can see where the influence of the Prusa CORE One INDX version has influenced Bambu Lab’s slight change of direction, although they’ve taken that style of design in a slightly different direction.

The rack features nine individual PrintTool docking positions, and these all automatically dock with tool alignment and offset compensation. What’s interesting here is that there are dedicated filament paths for each PrintTool, meaning that each works completely independently, and there are integrated filament cutters within the PrintTools, again just to help speed things up. When it comes to the extruder and internal mechanisms, they all use hardened steel, with hardened-steel extruder gears and nozzles rated at HRC 55+.

As ever, we see a flexible steel build plate, and this is a decent size, with a 330 × 320 × 325 mm build volume enabling you to tackle all sorts of projects, especially with the huge material support this new printer offers.

Bambu Lab H3C: Features

It’s difficult to do a quick run-through of the features because, over the H2C, things have really progressed, and the new Swaptek nine-tool changing system is a real evolution of design. You can’t help feeling it’s heavily influenced by the INDX system that we’ve seen from Bondtech and featured on the latest Prusa CORE One machines.

However, over that system, there are plenty of changes, and it’s obviously got the Bambu Lab spin, with automatic tool changes that happen in under six seconds. There’s also active preheating that essentially anticipates the upcoming tool changes and heats the next nozzle in advance, making complex models slightly faster to print than on some of those other systems.

One of the features that I noticed early on, and one of the issues with other systems of this style, is that quite often the tool rack gets in the way of being able to see the print area. However, once that new tool is selected, the tool rack actually retracts upwards so that you get a good clear view of the print area, and it also helps to keep the filament tubes out of the way.

Feeding all of those nine new toolheads is the redesign of the AMS system, known as the AMS 3H, that manages the filaments in an integrated box that sits on top of the machine. Like the AMS 2, it's a sealed storage option and helps protect the filaments inside from moisture.

Again, as we’ve seen in the past, it has automatic filament loading and routing that works exceptionally well, and with RFID built in, as long as you’re using Bambu Lab’s own materials, they’ll be automatically recognised and synced with the software, making the whole process that much easier.

One of the surprises is that only two of those nine slots are actively dried, so if you do have filaments that are more prone to moisture than others, then they should sit in those two. There’s a cover that sits over the top, so it’s pretty easy to see which two those are.

As ever, there are also expansion options, so if you have other AMS systems and you want to plug them in, the H3C supports additional AMS units for up to 24-colour printing.

One of the new additions is the FlexFeed extruder that supports both 1.75 mm and 2.85 mm filament, and there’s also a dedicated 2.85 mm filament path for flexible materials such as TPU. While TPU for AMS can just be used directly on the machine, as I discovered when I printed speakers with files from Print Your Speakers, when it came to printing other softer TPU, you have to use one of the bypass feed ports and quickly change part of the installation and feed to the toolheads.

The first time you do it, it takes a little bit of time and working out, but essentially, the soft TPU needs to go into one of the slots that enables active drying. TPU is hygroscopic and collects water like a sponge, so ensuring that it’s nice and dry before it goes through the system is just going to improve and help with the quality of the print.

You then feed that through into the 2.85 mm supporting toolhead, but rather than running through the filament feed tube, you bypass the tube, cutting just a small length that goes into the toolhead and then directly into the extruder. This just helps with the smooth movement of the TPU filament through the system. This means that it can actually cope with TPU all the way down to 75A, perfect for some of the projects that Bambu Lab supplied.

Using TPU in the system is not uncommon, but you still have to reconfigure the system slightly. Bambu Lab has made that slightly easier, and of course, you also have the ability to print with multiple materials and multiple colours.

But what impressed me the most is that with the Bambu Lab PLA palette, which is a new range of filaments, you can print with CMYKW to enable colour mixing using cyan, magenta, yellow, black and white filaments. The end results, as you’re seeing in some of the sample pictures, are quite amazing, enabling far more colourful and complex prints than has previously been possible.

But it’s not just about multiple colour-mixing filaments. You can also use a huge amount of engineering materials because of the hardened nozzles and the actively heated chamber, so utilising ABS, ASA, PA, PC and carbon-filled composites is all perfectly possible.

Further supporting the flexibility over the way you print is that there are multiple nozzle sizes, including 0.2, 0.4, 0.6 and 0.8 mm. But what makes a difference here is that you can optimise it to work using those different nozzles in one print. That means you can speed up the printing of the interior support structure with, say, a 0.6 or 0.8 mm nozzle and get that fine-quality finish on the exterior with a 0.2 or 0.4 mm nozzle. This is something very new.

There’s also been a range of feature enhancements for speed, with a maximum toolhead speed of 1,400 mm/s and acceleration of 25,000 mm/s². To keep up with that high-speed printing, the hotend flow rate of 40 mm³/s is also impressive, although these run to Bambu Lab’s own test conditions rather than what you’re actually going to see in the real world.

Something which I haven’t really been able to look at in any depth is the CraftTool support, and we’re starting to see Bambu Lab not only promote their machines for 3D printing but for other craft uses. This is what really made the H2D such an interesting machine. Potentially, in the near future, the CraftTools will enable automatic switching between compatible printing and crafting tools, so there will be blade cutting, drawing and debossing, but I’ll bring you more on that when I’ve been able to look at it properly.

As I mentioned in the Design section, Bambu Lab and, actually, quite a few of the other manufacturers are starting to take filtration seriously, and here we’ve got the H11 HEPA filter, granulated coconut-shell activated-carbon filter and an integrated chamber exhaust and circulation system. Bambu Lab actually even provided an air filter with the machine, which has been very handy, especially in quite a crowded workshop.

When it comes to features, the H3C is one of the most feature-packed printers to have come onto the market yet, but with all of those features comes an added level of complexity that could cause issues in the long run, and the only way to discover those is to ensure that, over the month of having this machine, it has been running almost constantly.

Bambu Lab H3C: Performance

As these 3D printers get more complex and the models that they’re able to produce become increasingly more elaborate, mixing not only colours but materials, the time it takes to fully test one of these machines is rapidly increasing.

Thankfully, Bambu Lab recognises that whilst machines are speeding up, when it comes to fully testing some of those mixed-filament options, it doesn’t just take hours; it takes days for each print to be produced. In order for you to run through all of the test prints provided, including rerouting different filaments, that all takes a considerable amount of time.

But thankfully, I’ve had this printer for almost six weeks, and through that time, there have only been brief periods where it has remained dormant. Otherwise, filament after filament has gone through, producing everything from pinhole cameras out of carbon-filled filament through to some quite incredible desktop speakers, thanks to the guys at printyourspeakers.com, and of course, the extremely elaborate models that the guys at Bambu Lab provided to test out this printer.

Before I go into that, I’ll just quickly run through the setup process of the machine, as whilst this machine looks extremely complex, the actual setup process is typically Bambu. Essentially, with the machine arriving in two boxes, the machine and a separate box for the AMS, it only took about half an hour before the machine was sitting proudly on the workbench, looking ready to produce its first print.

The instructions are pretty clearly laid out, and as you’d expect, the large 5-inch touchscreen gives you all the information you need. Once everything is plugged in and connected, you then simply load the filament, and it’s pretty much set to go. However, as always, that complexity comes with a large extra step.

First, you download and install the firmware, which all ran through nice and quickly, but then it was on to the full setup calibration. Whilst it doesn’t actually take any involvement from yourself other than clicking a couple of buttons, it does take about an hour, and when you have a machine of this size and potential in front of you, that wait is slightly agonising but unfortunately essential.

Once the machine had run through its copious amount of checks and all the nozzles had been calibrated, it was time to run through the first set of prints, and here I had three different options. Firstly, those pre-installed on the machine in the 64 GB internal storage; the second was my usual test prints; and the third was the additional files that Bambu Lab had provided to fully test out the system.

I opted for the quick run of those prints in storage just to check whether the calibration had worked and everything was printing correctly, and sure enough, print after print came off the print bed without issue, mixing the colours beautifully into pretty good-quality standard prints. All of these that I started with initially were PLA and mainly ends of reels from previous reviews.

Then it was on to my own test prints, and again, the accuracy and precision of the printer really stood out, and I was also impressed with the relative speed. One of the big points here is that the tool-changing system, as I saw with the INDX, vastly increases print speeds when compared with the H2C, which is a machine that I actively always have set up and running.

The speeds with this new system are again slightly faster. A quick test of those print speeds on the H3C, printing the two-tone tree frog in large, took two hours and three minutes, with about eight minutes taken for the preparation time and the rest on the actual model printing. This compared with almost three hours on the H2C, and whilst it was far faster on the preparation and checks at just under six minutes, the print time was closer to two hours and 50 minutes. Finally, comparing it with the H2D, surprisingly, it put in a similar time to the H2C.

As I ran through the test, the speed of the machine was the first thing that really stood out, and with a relatively large print bed as well, I was able to print large projects with absolute ease. Support structures went in nice and easily, and I always use PLA support, although as I started to venture into Bambu Lab’s own models, I realised the standard PLA support was just not delicate enough for what was needed, and PVA support, which could be washed away, would have been a better option.

Through my own models, there were quite a few different projects that I wanted to really give a go. The first was a pinhole camera that I had designed for some lectures and was small. It's a good way to test out dimensional accuracy, as I had to drop in a brass-sheet pinhole lens that I had designed and make it tripod-mountable, with brass inserts used for the screw thread.

As the box also had to be lightproof, dimensional accuracy due to the fitting and removing of a lightproof back was also essential, and using a carbon-filled filament with solid infill just ensured that the whole box would stay lightproof. I also had to feature a sliding shutter mechanism that would cover the brass aperture, so again, quality of surface printing as well as dimensional accuracy was absolutely essential.

After running several iterations of the design, I came up with the final product and tested it in the studio with success, producing small prints onto photographic paper through a 0.3 mm aperture. The quality of the print and success of the project just highlighted the first bit of flexibility with this machine and its design.

The next was to call on somebody with a lot more experience with high-quality and precision engineering 3D design, and the guys at Print Your Speakers provided me with the Supersonic MK3 desktop speaker kit to give this new printer a go. There were several angles that I wanted to fully test that no other printer would be able to do.

The first was obviously to print speakers using PLA for the outer, but then rather than having PLA infill, I wanted it to be TPU in order to help absorb some of the sound. I also wanted to be able to swap between nozzle sizes, all of which took a little bit of experimentation, and at present this is an area of the systems use that I will have to come back to, as at the moment I haven't had any success. I then tested a straight PLA box against the PLA and TPU mixed box, and the difference in audio quality was apparent, with a far deeper sound from the more advanced material-mix option.

Having now tried a variety of different materials and options through the machine, I was pretty confident about setting off with some of Bambu Lab’s own models to fully test out its abilities. The first was the FROSTSPINE Dragon Bust 1015, which used a variety of materials and took a while to organise the different materials that were needed into the AMS and configure the system.

Printing this started to point out the big difference between the old Bambu Lab multi-filament AMS systems and this new iteration. It's far more complex, and along with the printing guides that helped guide you through the setup, there’s definitely a greater skill required to get the most out of this machine.

It took two or three initial attempts before that model rattled off the print bed, taking several days to complete. I then moved on to the hummingbird frame, which was a simple multi-filament PLA mix, and here I really started to see the full potential, being able to mix the colours to create colours that weren’t directly available from a single filament but only once they were blended together, and the results were spectacular.

The only thing was that I accidentally used PLA support rather than PVA support, so once the final model printed, which took several days, it was going to be a long process to break away all of that support. The intricacy and delicacy of many of the models that you’re now able to print on this machine far exceed what you’ve been able to do before.

Through this whole process, the one thing that became apparent was that this is a far more hands-on machine, and there is an obvious learning curve, especially working with all the different toolheads, nozzle sizes, bypass filament feeds and potentially additional CraftTools. This is a very feature-enhanced workshop machine at a price that will appeal to both enthusiasts and professionals.

Bambu Lab H3C: Final verdict

In those six weeks of testing, I feel like I’ve only just started to scrape the surface of what’s possible, but there are several key points that I’ve learnt through this test. The first is that if you just want to get straight into printing, then there are cheaper machines out there which will enable you to do that. So if you only print with up to four filaments, which most people do, then the complexity of this machine might be just a little bit above what you need.

But then, as you start to embrace all its options and capabilities, and if you’re willing to sit down and learn about how to mix materials, how to use those different nozzle sizes together and obviously all those different filament types, then you have a 3D printer that is really quite different from anything that’s come before and sets the tone for what’s coming up in the future.

Whilst there are more than passing similarities with the INDX system, what Bambu Lab has done here is to evolve what it had done with its previous multi-filament systems and, with this new option, create something that is far more advanced.

With that increasing complexity come its own issues, and through the process of this review, I did have occasions where I had to rerun the full calibration as one of the nozzles wasn’t quite sitting right. Then there’s the whole bypass reconfiguration of the filament feed, which does get quite complex, but once you’ve done it once, you basically allocate one toolhead to TPU, still leaving you with eight for all other filaments.

The options and possibilities for this machine are huge, and through this review, I’ve just been incredibly impressed with the overall quality. Each firmware update has fixed some of the issues that I was having early on, such as the RFID not recognising filament reels and, for some reason, the auto-feed system suddenly not auto-feeding. But after the update, those issues were fixed.

I also had, as you might be able to see on the speaker set, a slight issue with surface finish, but then that’s because I was experimenting with some of those more advanced features, and this was the first time that I had configured the whole process myself in the software. Later, small models printed with much better quality, but again, this is all part of that learning process. As it stands, this printer is a definite evolution of what’s possible and an exciting glimpse into the future of 3D printing.

Should I buy the Bambu Lab H3C 3D printer?

Buy it if...

Don't buy it if...

For more models, I've tested the best 3D printers you can get right now.

Apple iPad mini 8.5-inch (2026) review: One-minute review

It's been a long time coming, but Apple has finally addressed more than a few people's wish lists with its latest-generation iPad mini. Yes, it's still the smallest iPad in terms of display size, not price, but this upgrade is really about the sum of its parts.

The jump to OLED is the headline upgrade, and paired with a thinner, lighter build and plenty of performance under the hood, it makes a big, big difference. The 8.5-inch Super Retina XDR display is an absolute joy to use; whether I'm watching a film, playing a game, editing photos, or just scrolling through iPadOS, everything looks fantastic.

If you've been holding out for an OLED iPad mini, this is the upgrade you've been waiting for. It's a major improvement, even without ProMotion, and the improved refresh responsiveness makes everyday scrolling and swiping feel that much more seamless.

The design is another big part of the story. At just 5.33mm thick and 254g, the iPad mini is deceptively slim and light. The mini was already the iPad I find easiest to take pretty much anywhere, but this new model is even more comfortable to hold and use for longer stretches. And while it might look familiar, there are some meaningful changes beyond the thinner build. The new IP67 rating adds welcome water and dust resistance, giving the mini more durability for everyday use, and giving you peace of mind if you're sharing it with family members.

It also packs plenty of power. The A20 Pro chip makes multitasking, photo editing, gaming, and even some of Apple's newer AI features, which demand a fair bit of power, feel effortless. Battery life has been solid in my testing, too, getting close to Apple's 10-hour claim, while the N1 chip brings Wi-Fi 7, Bluetooth 6, and Thread connectivity. If you opt for the Cellular model, you also get Apple's C2 modem.

Of course, all of this comes at a price. The iPad mini starts at $699, so it's certainly not the cheapest iPad in Apple's lineup. The model I've been testing is the 512GB Wi-Fi and Cellular version with the optional nano-texture display, which comes in at $1,099. That's a considerable investment for a tablet this size, but the mini has never really been about offering the lowest entry price. It's for people who want this particular screen size and the portability that comes with it, and Apple has now managed to squeeze a whole lot more value into this little package.

There are still a couple of things I'd like to see, most notably ProMotion and a more exciting selection of colors. But those small issues don't take away from how much I enjoy using this iPad. The OLED display makes a world of difference, the slimmer build is even better, and the combination of performance, battery life, and added durability makes this a seriously capable little tablet. If you're already a fan of the iPad mini, you'll love a lot about this generation.

Apple iPad mini 8.5-inch (2026) review: Specs

Apple iPad mini 8.5-inch (2026) review: Pricing and availability

Apple's new iPad mini starts at $699 / £699 / AU$1,099 for the Wi-Fi model, or $849 / £849 / AU$1,349 if you want Wi-Fi and Cellular. Apple began taking preorders on October 28, ahead of the tablet's official launch on November 6, 2026.

You can configure your iPad mini with 128GB, 256GB, or 512GB of storage. If you want Apple's nano-texture display, you'll need to choose either the 256GB or 512GB model, and that adds $100 / £100 / AU$200 to the price. Apple also includes a polishing cloth in the box with nano-texture models.

• Value score: 4 / 5

Apple iPad mini 8.5-inch (2026) review: Design

• Super light and thin
• This makes for more of an upgrade than you might expect from reading the specs
• Still the most portable iPad in the lineup

The new iPad mini is really, really thin — almost deceptively so — and won't weigh your hand down at all. At just 5.33mm thick and 254g, it's noticeably slimmer and lighter than the previous generation, which measured 6.35mm and weighed 295g.

The iPad mini has always been the most portable tablet in Apple's lineup, offering much of the full-size iPad experience in a much smaller package. That's still true here, but the latest model takes that portability even further. It's easy to hold in one hand while scrolling, throw into a bag for a trip to the cafe, and take pretty much anywhere. It's genuinely pocketable when you're wearing a jacket, and, interestingly, weighs less than an iPhone Duo.

The overall design hasn't changed dramatically, and that's not necessarily a bad thing. The slimmer build brings it even closer to offering that 'sheet-of-glass' look and feel, while also making it more comfortable to hold for longer viewing sessions. The display is larger, too, thanks to the slimmer bezels, so you're getting more screen without sacrificing that compact form factor.

On the front, Apple has moved the 12MP Center Stage camera so it's on the top edge when you're holding the mini in landscape orientation, aligning it with the rest of the iPad lineup. There's still no Face ID; Touch ID is built into the power button instead, and it works quickly to unlock the tablet and authorize purchases. The separate volume buttons sit nearby, with their orientation changing depending on how you're holding the iPad.

Flip it around and you'll find the centered Apple logo, rear camera, and Adaptive True Tone flash. The USB-C port is on the bottom when you're holding the mini in portrait mode, and a microphone sits near the Center Stage camera.

One thing you might notice is what's missing here: the speaker grilles. Apple has done away with those cutouts as part of its new Acoustic Audio system, which I'll get into more in the Audio section below. It makes for a much cleaner look, and the change also makes IP67 water and dust resistance possible — a first for the iPad mini, and a welcome addition. I've tested it with a little water, and the iPad stands up just fine; it's nice to know you can take it to the beach or use it in the rain. It's also some useful added durability for a tablet the whole family could easily use.

The one area where I'm a little less enthusiastic is the choice of colors. The A17 Pro iPad mini came in Blue and Purple, while here you can choose between Starlight and Midnight. I've been testing Midnight, which has a very dark blue look, but it does show fingerprints on the rear. Paired with Apple's new TechWoven Folio, though, it makes for a nice-looking setup.

• Design score: 4.5 / 5

Apple iPad mini 8.5-inch (2026) review: Display

• OLED display is a terrific addition
• Content looks exceptionally great here, with rich blacks and vibrant colors
• Nano-texture pairs really well with OLED and lets you use the iPad mini in any light

It feels like we've been asking Apple to bring an OLED display to the iPad mini for years — it's certainly been a wish of mine — and Apple finally delivered this year. Whether you're coming from the previous-generation iPad mini or older model, this is pretty close to a night-and-day experience.

The 8.5-inch Super Retina XDR display is an absolute joy to use. It's OLED, so blacks are properly inky, colors really pop, and everything looks fantastic. It's a great tablet for running through some tasks, drawing with the Apple Pencil, FaceTiming with family and friends, and especially for watching content.

Admittedly, it's only 0.2 inches larger than the previous display, so you're not getting much extra usable space. But the visuals are much more vibrant, everything looks crisper, and it's a more immersive view. At this size, it's made a real difference in my iPad mini experience.

I found myself editing photos in Pixelmator Pro and Apple's Photos app, drawing in Procreate, and playing games like Mini Metro and Mini Motorways to see how quickly visuals render and how much colors can pop. Watching films like Toy Story 5 or Moana on Disney+ was a blast, too, and the same goes for catching up on Ted Lasso on Apple TV. HDR content, in particular, benefits from the OLED panel, with more pronounced contrast and colors that really jump off the screen. This is a fantastic little movie-watching tablet.

The new mini also gets a brightness boost, with peak brightness reaching 1,200 nits for HDR and 600 nits in everyday use, up from 500 nits on the previous generation. You'll notice that difference indoors and outdoors, especially when watching HDR content. And when you're using the iPad mini outside, that extra brightness paired with NanoTexture makes a big difference.

Which brings us to the optional nano-texture upgrade. Available on the 256GB and 512GB models, it costs an extra $100 / £100 / AU$200, and adds Apple's proprietary, etched coating over the Super Retina XDR display. It's incredibly effective at reducing glare from both natural and artificial light sources, as well as reflections on the screen. Apple has brought this technology to several other products, including the iPad Pro, MacBooks, iMacs, and Studio Displays, and there's also a slightly different implementation on the iPhone Duo.

I've found nano-texture particularly useful in direct sunlight. It effectively mutes reflections and cuts down glare, making the display much easier to see when you're out and about. It also gives the screen a more paper-like feel, which makes using either the Apple Pencil Pro or Apple Pencil USB-C particularly nice. I've found myself bringing the iPad mini along to cafes and meetings to take notes, and its size makes it feel a little like carrying a classic Moleskine notebook.

It's not a cheap upgrade, and I haven't tested the standard glass and nano-texture versions side by side. But paired with a more portable iPad than the Air or Pro, and one that now has extra durability, it makes the mini even more enjoyable to use when you're out and about.

Another benefit of the move from LCD to OLED is better display responsiveness, even if the display here doesn't quite reach the 120Hz standard of ProMotion. Apple describes this as sub-pixel millisecond control, and in everyday use, scrolling and swiping feel a little more seamless than on the previous generation, and it does make the display feel that bit more fluid.

It's not, however, ProMotion. The iPad mini is still limited to a 60Hz refresh rate, rather than the super-smooth 120Hz you'll find on Apple's Pro models. As I've said in previous iPhone and iPad reviews, if you're not coming from a device with a higher refresh rate, you probably won't miss it. It's an odd omission here, especially at this starting price, but I've still found the iPad mini responsive in use, with no noticeable issues in everyday use.

All in all, the move to OLED makes a big difference. The new mini is a joy to watch content on, and even using iPadOS feels better on this display.

• Display score: 4.5 / 5

Apple iPad mini 8.5-inch (2026) review: Cameras

• A welcome move for the front-facing camera
• Still a 12MP main camera on the back

Like every other iPad in Apple's lineup, and most that came before it, the iPad mini has two cameras. Around the back is a 12MP wide camera paired with an Adaptive True Tone flash. It's nearly identical to the main camera on the previous-generation iPad mini, and if you do choose to take photos with the mini — and it's probably the most acceptable iPad for that — you'll get some good shots. I wouldn't expect it to rival an iPhone 18 Pro or even an iPhone Air, though.

The real upgrade is the camera built into the bezel around the front display. It's now a 12MP Center Stage camera, positioned centrally on the side of the iPad when you're using it in portrait mode, so that when you're holding it horizontally, it's in the right place for video calls. And when you pair the iPad mini with Apple's Folio Case or a third-party alternative, that's generally how you'll have it positioned when propped up.

Performance here is great for video calls, whether you're using Apple's own FaceTime or jumping into a Webex meeting, Slack huddle, or a Zoom or WhatsApp call. I looked clear and stayed in focus, but as someone who likes to walk and talk, I particularly appreciated Center Stage. As I moved around, the camera and onboard software worked together to keep me in the shot, using smart panning and zooming to follow my movements.

• Cameras score: 4 / 5

Apple iPad mini 8.5-inch (2026) review: Software

• Runs iPadOS 27 out of the box
• Full multitasking with windows is great on this form factor
• Support for two Apple Pencil models

The iPad mini is running iPadOS 27 out of the box, which certainly brings a lot more to the table than previous versions, particularly anything before iPadOS 26. For starters, the iPad mini supports the new Siri AI beta if you opt in and are in an eligible region. You also get the full windowed app experience that Apple introduced with iPadOS 26.

That means you can have several apps open at once, switch between them, and layer windows on top of each other. The mini's 8.5-inch display isn't the largest canvas for this kind of multitasking, but it feels surprisingly expansive, and can handle several apps at once.

Multitasking felt fast, and I often found myself using the iPad mini for more than just one thing at a time, ranging from streaming content or gaming to productivity tasks like emailing and web browsing. I also used it for creative pursuits, including journaling, drawing, and photo editing, as well as simply settling in with a good read.

What I really love about the size here is that, this iPad's name notwithstanding, this is not an especially small display. At 8.5 inches it's closer to the size of a classic notebook or book, so it actually feels pretty expansive, especially when you're working between different apps. Some days, though, I connected a keyboard and trackpad and plugged the iPad mini into an external display to give myself another window to work with.

That's a particularly welcome addition, because the mini is no longer limited to simply mirroring its display on an external monitor. You can also extend your workspace, which opens up even more possibilities for multitasking and getting things done.

I found myself pulling images and other assets from Photos, Files, and Messages, then dropping them straight into Pixelmator Pro. Moving between apps and juggling different windows felt effortless, and it's a good example of what the iPad mini can help you do beyond the usual browsing, streaming, and reading.

Of course, a lot of that comes down to the hardware under the hood. I'll touch on this more in the next section, but the iPad mini runs Apple's A20 Pro chip, from the same family of processors as those powering the iPhone 18 Pro, iPhone 18 Pro Max, and iPhone Duo. However, performance differs depending on which storage configuration you choose, so it's worth looking a little closer.

• Software score: 4 / 5

Apple iPad mini 8.5-inch (2026) review: Performance

• A20 Pro is very, very fast
• Plenty of a runway for demanding tasks
• You'll need to pay more for the best version of the chip

As I mentioned above, the iPad mini runs Apple's A20 Pro chip, but there's a tale of two chips here. I've been testing the 512GB model with Wi-Fi and Cellular, which gets the full-fat version of Apple's latest silicon.

If you opt for the 128GB or 256GB configurations, you get a five-core CPU (two super cores and three efficiency cores), a seven-core GPU, and a 16-core Neural Engine. The 512GB version, meanwhile, comes equipped with a six-core CPU (two super cores and four efficiency cores), the same seven-core GPU, and a Dual 16-core Neural Engine. That's the same chip configuration as in the iPhone 18 Pro, iPhone 18 Pro Max and iPhone Duo, while the lower-storage models get a slightly more binned-down version.

I don't suspect those extra CPU cores will make a night-and-day difference for most everyday tasks. You might see more of a benefit with particularly processor-intensive workloads or when running more AI locally on-device, but for the vast majority of people even the lower-storage iPad minis should feel plenty fast.

This is the second generation of iPad mini to get a Pro chip option, and it enables the tablet to tackle higher-performance tasks with ease. In Geekbench 7, my review unit scored 3,998 in single-core tests and 11,485 for multi-core. In Geekbench 6, it scored 4,720 and 12,697, respectively. Those are very impressive results, particularly compared with the scores for the previous-generation iPad mini, which scored 2,829 in single-core and 6,938 in multi-core in Geekbench 6.

Those are impressive numbers on paper, and in real-world testing over the past seven days, the iPad mini has been a champ. Apps open incredibly quickly, photo edits happen without a hitch, and even more demanding tasks, like using the Super Resolution option in Pixelmator Pro, and other power-hungry image upscaling tools, felt effortless. Everything feels very, very responsive, and I haven't found myself waiting around for the iPad mini to catch up.

I also put the Apple Intelligence features through their paces, including Clean Up and Expand Images in Photos, alongside multimodal requests with the new Siri AI beta. I tried a few third-party apps with AI-powered features, too. These are the kinds of tasks where having Apple's latest silicon and Neural Engine come into play, and the iPad mini handled my testing without any noticeable slowdown.

• Performance score: 4.5 / 5

Apple iPad mini 8.5-inch (2026) review: Battery & connectivity

• A lot of mileage for a mini iPad
• Impressive results in our battery test
• Apple's C2 modem serves up faster connectivity

Apple continues to promise all-day battery life with the iPad mini: up to 10 hours of web browsing over Wi-Fi or video, and up to nine hours of web browsing over a cellular data connection. Inside is an 18.9-watt-hour rechargeable lithium-polymer battery, and in my testing I've found it lasts close to 10 hours, depending on what I'm doing.

That's a pretty good result, especially considering the upgraded display. I suspect the OLED panel is more efficient, and paired with the A20 Pro, it makes for a nice combination. In Future PLC's lab battery test, the new iPad mini lasted an impressive 14 hours and 24 minutes. That's a substantial improvement over the previous generation, which managed 10 hours and 36 minutes in the same test.

Connectivity also gets a meaningful upgrade. The iPad mini features Apple's N1 chip, which supports Wi-Fi 7, Bluetooth 6, and Thread. If you opt for the Cellular model you'll also get Apple's C2 modem, which the company says offers up to 50% faster speeds compared to its previous modem.

I've been testing the cellular connection on Verizon's network in the US, with access to 5G, and it's performed pretty well. Throughout New Jersey, I tracked download speeds ranging from 600 to 1,200Mbps, with upload speeds around 35 to 62Mbps. That's plenty for streaming, downloading files, browsing, and staying connected when I'm away from Wi-Fi.

• Battery and connectivity score: 4 / 5

Apple iPad mini 8.5-inch (2026) review: Should you buy it?

Buy it if...

Don't buy it if...

Apple iPad mini 8.5-inch (2026) review: How I tested

I've been using iPads since the original launched, so I've seen Apple's smallest tablet evolve quite a bit over the years. I've been testing the 512GB Wi-Fi and Cellular version of the iPad mini (8th Generation with A20 Pro) in Midnight, with the optional NanoTexture finish, for the past seven days.

During that time, I've put it through its paces with a range of apps across entertainment, creative, and productivity, while also exploring many of the features in iPadOS 27. I've also used it with Apple's new TechWoven Folio, alongside both the Apple Pencil USB-C and Apple Pencil Pro, to get a feel for how well the mini works for everyday use, creativity, and getting things done.

First reviewed: November 2026

With memory prices showing no signs of creeping back down to sensible numbers, I've been keeping our eyes peeled for any discounts I can find on RAM - and in general, it's Corsair's Vengeance line that's consistently delivered good reductions on DDR5.

For example, the Corsair Vengeance RGB 32GB (2 x 16GB) DDR5-6000 kit is down to $490 (was $638) at Newegg when you use code BNWF89. If you’re building a new Intel gaming PC and want fast DDR5 that also looks the part, the Vengeance RGB kit gives you both. It's my personal pick for aesthetic Intel builds.

• Best for: gamers and PC builders putting together a new Intel system who want fast memory with customizable RGB lighting.

• Why we like it: 32GB of DDR5-6000 memory, ten individually addressable RGB LEDs per module controlled through Corsair’s iCUE software, one-click XMP 3.0 setup, and a limited lifetime warranty.

Why We Recommend It

Corsair’s Vengeance RGB line is one of the most popular choices for builders who want their memory to match the rest of their lit-up rig, and its iCUE software makes syncing the lighting with Corsair fans and coolers simple. We haven’t reviewed this specific kit, so this pick rests on Corsair’s reputation and the kit’s specs. Its XMP 3.0 profile is designed for Intel 600 and 700 series motherboards, so Intel builders get its full 6000MT/s speed with a single BIOS setting.

Price Context & Historical Value

This is $128 off the $637.99 list price, or $148 off with the code, around 23% in total. That makes it $70 cheaper than the non-RGB Corsair Vengeance DDR5-6000 CL30 kit we covered recently, though that kit has tighter timings and an AMD EXPO profile. Memory prices are still far higher than a year ago, so this is a good deal by today’s standards, not a long-term low.

Should you buy it?

✅ Buy the Corsair Vengeance RGB DDR5-6000 32GB if... you’re building an Intel system and want fast DDR5 with one-click XMP setup or you want customizable RGB lighting that syncs through iCUE.

❌ Skip the Corsair Vengeance RGB DDR5-6000 32GB if... you’re building an AMD Ryzen system, since this kit lists only an Intel XMP profile and no AMD EXPO or you want the tighter CL30 timings of faster kits.

• See all desktop memory at Newegg.

RedMagic 12 Pro: One-minute review

RedMagic has produced yet another phenomenally high-performing phone in the RedMagic 12 Pro, but it no longer screams ‘bargain’ in the way the series used to.

Its design has been slightly slimmed down and lightened up, though it remains a whole hunk of a phone with a slightly gauche look. Mind you, that visible liquid cooling system meets the approval of our inner 13-year-old.

The phone’s display is brighter and faster than that of the RedMagic 11 Pro, and provides one of the cleanest media experiences thanks to the lack of a selfie notch — though that omission leads to issues elsewhere.

As one of the first phones released with Qualcomm’s Snapdragon 8 Elite Extreme Gen 6, the RedMagic 12 Pro is just about as fast as you could hope for, though it may not be able to sustain peak performance quite as well as its predecessors.

Battery life is the biggest improvement here, thanks to a massive increase in capacity. This is a phone that could potentially last you days, provided you don’t hit the games too hard, with 80W charging (both wired and wireless) getting you up to speed relatively swiftly.

But for all of these improvements, RedMagic has fallen into a familiar and oh-so-2026 trap. In raising the price of the 12 Pro by such a considerable amount (£170 / $250 / €200), it no longer feels like the total bargain this series built its reputation on.

Don’t get us wrong, £799 is a very good price for a phone that performs and lasts as long as this, especially in this hiked-up climate. But we’re still talking about a flagship(ish) price here, which can only draw extra attention to the phone’s failings.

For all the subtle progress elsewhere, that’s most typified by a camera system that’s as mediocre as ever. Selfies, especially, continue to be a total write-off on the RedMagic 12 Pro thanks to that aforementioned hidden front camera.

Elsewhere, a paltry two-year OS update promise further dents the brand’s formerly unimpeachable sense of value.

The RedMagic 12 Pro is the most capable gaming phone on the market, but the previous-generation RedMagic 11 Pro remains a much better buy while stocks last.

RedMagic 12 Pro review: Price and availability

• Released on November 18, 2026
• Priced from $899 / £799 (around AU$1,290)
• Another big price bump
• Available globally

Open sales for the RedMagic 12 Pro commence on November 18, 2026. Availability is global, including the UK, US, Australia, and much of Europe.

Following an industry-wide trend, the phone is significantly more expensive than its predecessors. With a starting price of $899 / £799 (around AU$1,290 — RedMagic lists the 'global' price of the 12 Pro in US dollars) for 12GB of RAM and 256GB of storage, it’s a considerable $250 / £170 (around AU$360) more expensive than the RedMagic 11 Pro.

This represents the second major price rise in a year for the brand. The RedMagic 11S Pro only launched in June, and at a lower starting price of $849 / £709 (around AU$1,220).

You’ll pay a lot more for models with higher RAM and storage, too: $999 / £889 (around AU$1,430) for 16GB/512GB, and $1,199 / £1,059 (around AU$1,720) for 16GB/1TB.

• Value score: 3.5 / 5

RedMagic 12 Pro review: Specs

RedMagic 12 Pro review: Design

• Extended liquid cooling system
• Confusing IP68/IP69 and IPX8 certification
• Dedicated gaming mode button and capacitive shoulder buttons
• Improved speakers

RedMagic’s current design language dates back to the RedMagic 8 Pro in late 2022. That was the first RedMagic phone to feature the brand's now-signature flattened-out look, as replete with right angles as it was with semi-transparent elements and RGB lighting.

The RedMagic 12 Pro broadly follows that trend, but there have been a few minor tweaks and refinements in the years since the 8 Pro hit shelves.

It remains a big lump of a phone, measuring 163.03 x 75.96 x 8.55mm and weighing 223g. Compared to last year’s model, however, the 12 Pro is 7g lighter and 0.35mm thinner, together with a fractionally smaller footprint.

I’m somewhat perplexed by the RedMagic 12 Pro’s level of dust resistance. New to this year’s model is the claim of full IP68 and IP69 certification, which is genuine flagship-grade. However, this comes accompanied by the following small print, which I’ll quote directly from the press materials: “IP68 and IP69 ratings apply to the device, while the cooling fan area is rated IPX8 due to its independent airflow channel design.”

So the 12 Pro is completely dustproof... until you get to the open vents on either edge of the phone. Thanks for that, RedMagic.

Those vents do, admittedly, form part of the most extensive cooling system around, leading to a physical fan that spins at a higher 25,500rpm (up from 24,000rpm). This is paired with an even more extensive liquid cooling system, which now winds its way around much more of the RedMagic 12 Pro, drawing heat away from core components.

The lava lamp look of this liquid cooling system continues to fascinate, with tiny clear globules winding their way around a brilliant blue pathway. It’s more tasteful than the blingy RGB LED lighting, which can be configured around the cooling fan, shoulder buttons, and to a lesser extent the cooling goop itself.

The right edge of the phone continues to be crammed with elements, with a set of capacitive shoulder buttons joining the volume rocker and circular power button. One new element sees the old slider switch giving way to another push button.

It serves the same purpose, granting access to RedMagic’s custom gaming UI. But you’ll now need to press and hold the button rather than simply flicking a switch.

It’s less satisfying and intuitive to use, but on the plus side, it now functions much better as a remappable ‘Magic Key’. Repurposing it as a camera or flashlight shortcut feels more natural now.

RedMagic has bolstered the audio output of the 12 Pro’s Dolby Atmos-certified stereo speakers. You now get a pair of 1115E speakers — they're more balanced and fuller-sounding, although they lack the low-end kick of the speakers on something like the Poco F9 Ultra.

• Design score: 3.5 / 5

RedMagic 12 Pro review: Display

• 6.85-inch OLED display
• 185Hz refresh rate
• 1,900 nits maximum brightness

RedMagic has kept the basic dimensions of the 12 Pro Display consistent with its recent antecedents — it’s another large 6.85-inch OLED panel with a 2,688 x 1,216 (1.5K) resolution.

Improvements have been made, however, starting with a move to a new BOE X11 panel material, which has a higher maximum brightness of 1,900 nits (up from 1,800 nits).

It can also hit a higher 185Hz refresh rate, compared to the RedMagic 11 Pro’s 144Hz. Not many games will be able to hit the kind of elevated frame rates to make the most of that spec, but it makes more sense on this gaming-focused phone than any other.

As always, I found RedMagic's default color profile to be a little too punchy, but a quick rummage around in the Settings menu — and a switch to the Standard mode — saw to that particular problem.

Ultimately, the RedMagic 12 Pro's display has all of the key strengths we’ve grown accustomed to with the brand. Its large size and unobstructed view — there’s no selfie notch to get in the way of gaming or video content — set it apart from most other phones and make it a great choice for media playback.

• Display score: 4 / 5

RedMagic 12 Pro review: cameras

• 50MP main camera with OIS
• Mediocre 50MP ultra-wide
• No more macro
• 16MP in-display selfie camera remains awful

RedMagic phones have never been particularly impressive in the camera department. We’ve always accepted this fact, albeit somewhat begrudgingly, because of the brand’s tight gaming focus and aggressive pricing.

The RedMagic 12 Pro’s cameras don't improve matters, and they're a little harder to defend given the phone's price bump.

Its camera system is led by a 1/1.56” 50MP OmniVision OV50J sensor. Image quality in good lighting is fine, though it's nothing you can’t get (or even improve upon) with a decent mid-range phone like the Google Pixel 10a.

Night shots, meanwhile, reveal noisy skies and a distinct lack of detail pulled from the shadows.

Portrait shots are weirdly horrendous, with an oddly fake-looking bokeh effect made worse by a terribly imprecise edge blur around the subject.

You’re not going to hear me complaining about the loss of the RedMagic 11 Pro’s 2MP macro camera, but it’s technically another cut.

With no dedicated telephoto, zoomed shots are forced to crop in on that main sensor. The results look somewhat soft even at 2x, becoming noticeably grainy at 3x, and downright unusable at 10x.

There’s also a humble 50MP ultra-wide camera with a 1/2.87" OmniVision OV50D40 sensor. It yields slightly better results in terms of sharpness, though the tone is noticeably cooler and the exposure falls short of the main camera's equivalent.

Matters degrade further with the RedMagic 12 Pro's 16MP front camera, which once again suffers badly from its positioning under the phone’s display. While this positioning is an advantage for clean media playback, it renders selfies a fuzzy mess, producing sickly skin tones and smudged details. They’re a write-off.

Video, at least, is moderately impressive on the RedMagic 12 Pro, with support for 4K/120fps enabling reasonably high-quality slo-mo footage. You take your wins where you can get them.

I get that photography isn’t the focus here, but the simple point is that a $899 / £799 smartphone should be taking better pictures in late 2026.

• Camera score: 2.5 / 5

RedMagic 12 Pro review: camera samples

RedMagic 12 Pro review: performance

• Snapdragon 8 Elite Extreme Gen 6 chipset
• 12GB or 16GB, or 24GB of RAM
• Extended liquid and active cooling system

The RedMagic 12 Pro is one of the first phones on the market to use Qualcomm’s new Snapdragon 8 Elite Extreme Gen 6 chipset. While this 2nm chip is the fastest Android chip on the market at the time of writing, it doesn’t bring about a seismic shift in performance.

Benchmark tests suggest that the RedMagic 12 Pro is barely any faster than the RedMagic 11 Pro and RedMagic 11S Pro (which share the same chip) in CPU terms, according to Geekbench 6.

GPU tests point to a much bigger improvement in GPU terms, with Geekbench 6 GPU scores that are between 16 and 28% faster, depending on the test.

That’s good to hear from a gaming-focused smartphone. Less positive were my 3DMark stress tests, which suggest the RedMagic 12 Pro is worse than its predecessors when it comes to sustained load.

Its peaks were higher, but its lower scores towards the end of the 20-minute WildLife Extreme Stress Test were quite a bit lower. This resulted in a stability score of 54% compared to 66% on the RedMagic 11 Pro.

Putting aside the possibility that this is a simple driver issue, it would suggest that, while the RedMagic 12 Pro is capable of running faster than its predecessors, it’s not capable of maintaining that flat-out speed quite as well. That’s surprising, given RedMagic has packed in an even more extensive liquid cooling system, plus a large vapor chamber and a faster cooling fan.

Benchmarks aside, the RedMagic 12 Pro performs impeccably in demanding games, running Wuthering Waves at relatively smooth frame rates with high graphical settings.

Ultimately, if you’re after the fastest-performing Android phone on the market right now, you won’t do any better than the RedMagic 12 Pro. Viewed from this angle, that $899 / £799 asking price doesn’t seem so bad at all.

• Performance score: 4.5 / 5

RedMagic 12 Pro review: software

• Fluid yet bland Android 17-based UI
• Game Space gaming UI
• Two Android updates
• Five years of security updates in Europe, three in the US

One of RedMagic’s main weaknesses has always been software, and some credit is due in this department on the RedMagic 12 Pro.

The company has cleaned up its act considerably, to the point where RedMagic OS 12 looks and feels much like most other custom Android UIs from Chinese manufacturers.

Instances of random Chinese text and troublesome bugs are now relatively rare — though I did observe a slight glitch with the DisneyPlus app in picture-in-picture mode. Gone, too, are the ugly screen-dominating widgets.

What’s left is slightly bland and generic, and relatively free of frills or flair, but it works quickly and reliably. Google Feed is placed to the left of the home screen, in all its increasingly AI- and ad-blighted glory.

RedMagic OS still has a bit of a bloatware problem. There are unwanted preinstallations for Booking.com, TikTok, Instagram, and AliExpress, as well as three rubbish games and a duplicate web browser app. But none of that is particularly unusual.

More problematic is the relatively poor level of software support. RedMagic promises a mere two full Android OS updates and up to five years of security updates for Europe. The US only gets three years of the latter.

As ever, the unique selling point here is Game Space UI. Long-press the red button, and this interface will offer you a hub for running, tweaking, and optimizing your games. It’ll also let you crank up the processor or map those capacitive buttons to specific controls within games via a drag-out overlay.

It’s all very busy and nerdy, but that’s just what the doctor ordered with a gaming phone of this caliber.

• Software score: 3 / 5

RedMagic 12 Pro review: Battery life

• Huge 8,400mAh battery
• Multi-day usage
• Comprehensive 80W wired and wireless charging

RedMagic has managed to massively increase the size of the 12 Pro’s battery, despite its slimmed-down chassis. It now runs on a massive 8,400mAh cell, up from 7,500mAh in the RedMagic 11 Pro and its subsequent S variant.

If you don’t stretch the phone's legs with too many high-end games or a lot of 4K video recording, you could well get through three full days before needing to recharge. On one day of light-to-moderate usage, with 3 hours 40 minutes of screen-on time (and no gaming), I was still left with 72% in the tank.

This hefty battery is, of course, built to allow you to incorporate a couple of hours of heavy-duty 3D gaming into your day, but even then, you should be able to get through a full day without running too low.

When it comes time to recharge, 80W wired and wireless charging keeps things reasonably rapid. It’s not hugely impressive, but given that large capacity, a 50% charge in 30 minutes seems pretty good.

A full charge from empty took me about 72 minutes.

Do note, however, that I was able to use a compatible Nubia charger, and that no such provision comes in the box.

• Battery score: 5 / 5

Should you buy the RedMagic 12 Pro?

Buy it if...

Don't buy it if...

RedMagic 12 Pro review: also consider

RedMagic is the only dedicated gaming phone maker on the market, but there are alternative options to consider.

How I tested the RedMagic 12 Pro

• Review test period = 2 weeks
• Testing included = Everyday usage, including web browsing, social media, photography, gaming, streaming video, music playback
• Tools used = Geekbench 6, 3DMark, native Android stats, Nubia 80W power adapter

I tested the RedMagic 12 Pro for two weeks, using it as I would any other gaming phone. I played demanding titles like Wuthering Waves for extended periods, watched movies and TV shows on apps like Netflix, and browsed the internet and social media during casual-use days.

I tested the phone's power credentials using benchmark apps including Geekbench 6 and 3DMark, and used a compatible Nubia 80W power adapter to measure its charging speed.

First reviewed: October 2026

Apple's Mac mini is a fantastic option if you want a powerful Mac without losing much desk space. We love the older M4 model, and wondered during tests whether it might be the best Mac for most people (it's certainly one of the best mini PCs we've ever reviewed).

Now, ahead of Black Friday, the new Apple Mac mini M6 is down to $779 (was $899) at Amazon, while UK readers can grab it for £824 (was £899). The latest 2026 model takes that familiar five-inch design and upgrades it with the M6 chip, making it even more powerful.

• Best for: Anyone wanting a compact but capable Mac for everyday productivity, multitasking, programming, photo editing, or creative work without paying for a more expensive Mac Studio.

• What we like: This is the latest upgrade of one of our favorite desktop computers. One of the Mac mini's biggest advantages is that it packs plenty of processing power in a package that measures just 5 x 5 inches and stands a mere 2 inches tall.

• What you need to know: The older and more affordable M4 Mac mini is still a worthwhile investment for general work and light creative tasks, though it lacks the power of the new model.

Today's best Apple Mac mini (M6) deal

Why we recommend it

The M6 chip is designed to handle everything from routine office applications to more demanding creative tasks. Its hardware-accelerated media engine supports H.264, HEVC, ProRes, and ProRes RAW, which is particularly useful for video editing and playback.

The M6 features Neural Accelerators in its GPU cores and a dual 16-core Neural Engine, designed to speed up supported machine-learning applications without relying entirely on cloud processing.

We haven't reviewed the M6 version of the Mac mini yet, but we're seriously impressed with what it offers.

For more picks, we've rounded up the best mini PCs we've tested.

Price context & historical value

At $779, the Mac mini M6 is $120 cheaper than its usual $899 asking price, a hefty 13.3% reduction.

The same configuration has previously sold for $769, although the current offer is just $10 higher, and definitely worth considering.

Should you buy it?

✅ Buy the Mac mini M6 if...

You want a modern Mac desktop with enough performance for demanding everyday applications, creative projects, and multitasking. It's also a sensible option if you already own a suitable monitor, keyboard, and mouse.

❌ Skip the Mac mini M6 if...

You already have a recent Apple silicon Mac mini, like the M4 version, that handles everything you need. The M6's improvements may not justify replacing a perfectly capable machine, particularly if your work doesn't benefit from the newer processor.

• Warhammer 40,000: Dawn of War 4 creative director Jan Theysen says the RTS game draws inspiration from Dawn of War 2 and 3's mechanics
• Theysen says, "There are definitely elements" in there from the first game, such as hero mechanics and UI features
• While core inspiration comes from the first Dawn of War, King Art felt it needed to find the right balance between old and more recent mechanics

Warhammer 40,000: Dawn of War 4 creative director Jan Theysen has confirmed that while the upcoming game took core inspiration from the first Dawn of War, King Art also borrowed elements from Dawn of War 2 and 3.

In an interview with TechRadar Gaming, Theysen described how the game manifests the classic real-time strategy (RTS) experience of the 2004 game and that, when the studio was looking for a starting point, Dawn of War "was more this classic RTS with base building, with big battles" and its "north star" for inspiration.

The King Art co-founder also revealed that while Dawn of War was the basis for Dawn of War 4, the studio also took features from the second and third games and "there are definitely elements" in there such as hero mechanics and UI features.

"A lot of the hero stuff comes from Dawn of War 2, of course," Theysen said. "I mean, the heroes are not as powerful or not as important as they were in Dawn of War 2 or 3, but a lot of the upgrade mechanics and the different paths you can send your heroes down; you can optimize them in different ways and so on, like meteor range or stuff like that.

"There is definitely a lot of Dawn of War 2. I also think the look and feel is kind of maybe more Dawn of War 2 than it's Dawn of War 1. Dawn of War 3, we liked aspects of the UI, for example. So there are some elements like the unit cards at the bottom of the screen and so on. That is something that we, even for Iron Harvest or previous RTS [games], already took from Dawn of War 3."

Theysen's reasoning for borrowing from the sequels comes down to Dawn of War being over 20 years old, and King Art felt it needed to find the right balance between old and new.

"Dawn of War 1 is a great game, and I love it, but it's also over 20 years old," he added. "So stuff happened in between [game releases] and wherever it's possible, we want to get the best version. But what we also said was, when in doubt, if there's not a really strong argument to do something differently, then let's just do it."

Similar to what Relic Entertainment did with its Dawn of War sequels, Theysen said, "Just because there's no reason to change stuff, don't fix it when it's not broken."

Warhammer 40,000: Dawn of War 4 launches on PC on December 3.

• John Ternus has warned that: "What we're seeing with memory pricing right now is unprecedented."
• The Apple CEO cautioned that RAM costs are going to climb higher in this final quarter of 2026
• It sounds even bleaker than Apple's previous statements on the RAM crisis, and follows many other warnings that 2027 is set to be a miserable year price-wise

Apple is warning of an 'unprecedented' rise in the cost of RAM, adding to the cacophony of alarm bells that have been ringing about the price of memory chips (and other PC components besides) recently.

The Guardian reports that in a conference call following Apple's fiscal Q3 results — which were better than analysts expected, driven by the iPhone, Mac mini and sales in China — Apple's new CEO John Ternus touched on the subject of the RAM crisis.

Ternus, who recently took the reins from Tim Cook, cautioned bleakly that: "What we're seeing with memory pricing right now is unprecedented."

He then observed that memory costs would climb even higher in the final quarter of 2026 compared to the rest of the year, during which, as you've doubtless seen, we've already witnessed a steep climb in pricing.

Despite its financial results outperforming analyst forecasts, Apple informed investors that these higher memory costs are going to erode some of its profit margins, which meant the company's stock dropped.

Apple raised the prices of Macs earlier this year, as well as iPads and more recently iPhones, and Ternus acknowledged that this had "provided some offset" for Apple in terms of the ongoing cost crisis around components.

Analysis: Ternus takes a turn for the sterner

It's an ominous utterance from Ternus, and sounds more severe than the warning Apple issued back before the Mac (and other) price hikes happened in June 2026. Back then, when Cook was still chief executive, he warned about "significantly higher memory costs" that would "drive an increasing impact" on Apple's products, which, as noted, came to pass.

Now we're hearing the word "unprecedented", which sounds a lot more concerning than "significant"; although in fairness, this is referencing what's already happened this year as much as what's to come. But we're clearly being primed for more price-hike misery by Ternus here, and it's difficult to see how this won't have a further impact on Macs, iPhones and iPads, if not in this quarter, then in 2027.

We don't, of course, know to what extent profits might be eroded by more expensive memory chips, or how this might be balanced by consumers having to foot that bill. Only time will tell on that score.

Mac mini sales were mentioned in the report — which quoted Ben Barringer, the head of technology research at investment management firm Quilter Cheviot — as being responsible for part of Apple's boosted earnings for the most recent quarter, and that's down to the increasing demand for local AI.

The Mac Studio is major part of that picture, too, and this particular PC just popped up as an example of how ridiculous RAM pricing has just become at the very top end of the Apple product tree. If you want a Mac Studio with 512GB of RAM, you'll now pay more than double the cost of a Mac Studio that has 256GB of memory, meaning that it's cheaper to buy two of the latter computers (while getting two CPUs/GPUs in the process).

It's a bizarre situation, but this is what we've arrived at with memory upgrades, which were always seriously pricey with Apple, even before the RAM crisis kicked off (that goes for storage, too).

And this situation is only going to get worse, seemingly, according to Ternus, which is also, of course, the view of the wider industry and various analysts. I can't begin to count the warnings we've had in the past couple of months about the RAM crisis becoming even more dire, so at this point it's just a question of how much worse things might get.

I fear that some categories of tech and PC components are going to become increasingly unaffordable next year, and I recently wrote about why Black Friday could be a good time to move on certain purchases.

• Reddit wants to overturn Australia’s social media ban for under-16s
• It is challenging the law in the Australian High Court
• Government says it’s defending the law from "sustained ideological attack"

In late 2025, Australia became the first nation in the world to ban social media for under-16s — and as you would expect, that hasn’t gone down particularly well with the app makers affected. Reddit, for example, has issued a stinging rebuke to the Australian government and is in the process of challenging the ban in court.

A lawyer representing Reddit called the ban “ill thought out” (via Reuters) and said it was a “profound ‌restriction on political debate.” The attorney was speaking at the Australian High Court, where Reddit is seeking to contest the country’s new law.

Together with a libertarian advocacy group, Reddit claimed in court that the social media ban was “a very substantial restriction on what young people can now do” and that it could stifle political debate. It added that the restrictions may stop users creating content and responding to posts.

Although Reddit said it supported the aims of the law, it branded the legislation “inept,” saying that it did not have any specific means to keep minors off social media and failed to regulate other platforms where cyberbullying might take place, such as on gaming websites. And the fact that users who were not logged in could still view huge amounts of material undermined the law, Reddit argued.

Could Reddit overturn the ban?

Since the Australian law came into effect, several studies have shown that large numbers of teenage users targeted by it are still able to access restricted material. Sometimes, that might be as simple as using a VPN service to disguise your location, while at other times, social media firms are simply not taking action to restrict usage by under-16s.

Despite Reddit’s criticism, the Australian government has said it will defend its legislation. In court, the state’s lawyer, Joanna Davidson, said that children were more vulnerable to harmful online content because their brains were not fully developed compared to those of adults.

And speaking outside the High Court, communications ⁠minister Anika Wells ​said: “We ​are here to defend a fundamental tenet of Australian democracy, which is an Australian’s right, and particularly an Australian child’s right, to be free from harm. We seek ​to defend that from sustained ideological attack.”

The government has doubled the maximum fine for non-compliance with the law to AU$99 million (about $69.5 million / £51.5 million), increased the discovery powers of the eSafety Commissioner, and said it is preparing enforcement action against some of the largest social media platforms operating in Australia.

Whatever side you take on this matter, it’s clear that neither the Australian government nor Reddit are preparing to back down anytime soon. And with other nations considering similar laws, many people around the world will be watching closely.

• Elon Musk mocks Sam Altman's recent claim that people shouldn't own RAM
• "The people deserve to have their RAM too!" Musk's meme on X says
• Altman has been keen to secure more memory for OpenAI recently

Elon Musk has mocked OpenAI CEO Sam Altman following the latter's declaration that owning a PC with RAM is 'quite terrible', as the components should be powering the next generation of AI data centers.

In his typically irreverent manner, Musk shared a meme on X, the social network he owns, showing a version of a scene from Inglorious Basterds with a German WWII officer played by Christoph Waltz apparently hunting for people hiding sticks of DRAM.

"The people deserve to have their RAM too!" Musk's meme declares.

Stopped clock

Altman made his claim during an appearance on the YouTube channel Cheeky Pint, with his beliefs causing widespread condemnation and confusion across the industry.

"From a global allocation perspective, actually, it's quite terrible to have everybody have a bunch of the world's limited supply of DRAM sitting in their own computer and being used 5% of the time," Altman said.

"You want maximum utilization in the cloud if you're trying to deal with this very supply-constrained environment. And somehow many people's intuition is the opposite of that."

Altman's comments come as consumers face rising prices for any kind of computing hardware as the demand for RAM and other components gets passed down.

In recent months, we've seen the likes of Intel CEO Lip-Bu Tan warning of continued shortages and price rises, noting, "AI hardware will stay in short supply for the next 2 to 3 years", a trend he said was "driven by persistent constraints in the underlying tech supply chain."

Dell CEO Michael Dell has also voiced his belief that the memory shortage will get worse through 2027, and that these RAM woes are going to be prolonged for quite some time.

Top DRAM manufacturers Samsung and SK hynix have also warned that AI-driven memory shortages will last until 2027 and beyond, with their supplies rumored to be down to fewer than 10 days of memory inventory.

Interestingly, Altman reportedly recently met with about 50 senior Samsung executives in South Korea, as he looked to secure more hardware for OpenAI to expand its AI capacity.

But no matter what happens, one thing that (probably) won't happen is Altman breaking into your house to steal your PC's RAM...probably.

• Octopus Energy, Ampace, and CATL are joining forces
• They're promising to 'transform' the smart home battery landscape
• We're expecting next-gen tech like sodium-ion batteries

It's a boom time for home energy solutions — from DIY solar kits to smart home batteries — and three of the biggest names in the industry are joining forces to build the next generation of home energy storage solutions.

Energy supplier Octopus and battery manufacturers CATL and Ampace have announced a 'Memorandum of Understanding' (MoU) to work together to produce the best possible combinations of hardware engineering and software intelligence.

Octopus will bring tech such as the smart tariff, mobile app, and customer ecosystem platforms, while CATL and Ampace supply the battery chemistry and hardware nous. Initially, the initiative will be focused on European markets.

At this early stage we don't have details of any actual products, but expect advances in cutting-edge tech like sodium-ion batteries, high-density cell designs, and long battery lifecycles — all of which should benefit customers and reduce electricity bills.

Looking to the future

According to Octopus founder and CEO Greg Jackson, combining the efforts of the three companies will "transform home energy storage", and offer "next-generation batteries where hardware and software work in complete harmony".

Jason Zhu, the President of Energy Storage Systems at Ampace, is equally enthusiastic: the partnership will "fundamentally reshape how energy is delivered and consumed" as systems become "smarter, more connected, and more accessible".

Of course we'll have to reserve judgment until actual products and offers arrive, but Octopus Energy, CATL, and Ampace all have strong track records. The products that we've seen from these companies, we've liked.

The joint press release emphasized the ability for home batteries to charge up during off-peak times and share power back to the grid at times of highest demand, which should mean significant savings for customers — something everyone can support.

In July 2026, OpenAI disclosed that several of its cyber-capable AI models, operating during internal evaluations, circumvented controls intended to isolate them from the internet and compromised parts of OpenAI’s research infrastructure and Hugging Face’s systems.

A fuller report published the following month revealed that the agents had exploited shared infrastructure, communicated through unauthorized channels, collaborated with one another and accessed several external systems. OpenAI described the incident as a “warning shot” demonstrating that capable AI agents can find and exploit weaknesses across interconnected systems when safeguards are insufficient.

Despite the alarming headlines, this should not be viewed solely as a failure of artificial intelligence. It offers an important security lesson for organizations preparing for a future in which autonomous systems play a larger role in cybersecurity operations, business processes and software development.

The most valuable takeaway is not just that an AI model “escaped”, rather the lesson is that even controlled environments can contain unintended paths to sensitive systems, credentials or data. Once such a path exists, a sufficiently capable and persistent agent may find it.

The positive side of early disclosure

Every significant advancement in security has resulted from discovering risks before attackers could exploit them at scale. In this case, the activity occurred during controlled evaluations rather than a criminal campaign. The organizations involved were able to identify weaknesses in containment, access controls and environmental design before similar techniques could be widely adopted by cybercriminals.

The incident also demonstrated something important about modern AI systems: persistence. When confronted with obstacles, the models did not try to stop. They continued pursuing their assigned objectives and searched for alternative routes.

Security teams perform penetration tests and red-team exercises because human attackers rarely follow a single path. Advanced AI systems are becoming capable of behaving similarly, exploring multiple options and adapting when obstacles appear. The earlier organizations understand this reality, the better prepared they will be.

The real issue was not one model

Much of the initial discussion focused on the behavior of OpenAI’s models. Subsequent disclosures have shown that the underlying challenge is more nuanced. Anthropic identified three incidents in which Claude models reached the public internet during cybersecurity evaluations and gained unauthorized access to real organizations. OpenAI also disclosed separate cases involving third-party testing environments in which model activity extended beyond its intended boundaries.

These incidents, though similar, were not identical. In some cases, models exploited vulnerabilities to circumvent isolation. In others, configuration errors left routes to the internet open while the models had been told they were operating within simulations. However, the common lesson is that instructions, policies and safety guardrails cannot compensate for weaknesses in the surrounding infrastructure. The instructions described one set of boundaries while the technology enforced another.

Security professionals have encountered this problem for decades. A firewall rule may be intended to block access while an overlooked network connection creates an alternative path. A user may be prohibited from accessing data while excessive permissions provide an unintended route. Autonomous systems create the same challenge at greater speed and scale.

An optimizing agent does not necessarily think like a human attacker. It pursues the objective it has been assigned and evaluates the available options. Organizations should therefore assume that autonomous systems may pursue their objectives through any path the environment leaves open.

Security must move beyond AI guardrails

Much of the discussion around AI governance has focused on whether a model should answer a particular prompt or perform a particular action. Prompt-level guardrails remain important, but they are no longer sufficient on their own. Security teams must continuously assess who is authorized, which systems and data they can reach, and what actions could create risk. Each layer should be tested independently rather than relying on a single refusal mechanism to govern them all.

The OpenAI incident illustrates this clearly. The problem moved beyond models possessing advanced capabilities, to one where capabilities, connectivity, credentials and autonomy came together in environments where containment mechanisms were insufficient.

Organizations must therefore consider AI security as a combination of identity, authorization, containment and oversight.

The conditions for trusted autonomy

In order to combat this, defenders need to align themselves to a new framework that is shaped by frontier AI: Assume Autonomy. The framework begins with the expectation that adversaries and defensive controls may operate through automated, non-human agency. It is built around context, constraint, transparency and reversibility.

Context means understanding what an agent can access, which identities it can use and how those systems connect to other resources. Constraint requires access to be limited through technical controls rather than instructions or expected behavior alone.

Transparency provides visibility into what an autonomous system is doing while it is happening. Reversibility ensures teams can stop workloads, revoke credentials, isolate systems and recover quickly when unexpected behavior occurs.

For high-capability agentic systems, governance must address these areas with evidence rather than assumptions. This requires autonomy stress testing, independent red-team exercises focused on containment failures, testing of identity and network boundaries, audited isolation controls, runtime authorization for high-impact actions and rehearsed rollback procedures.

Responsible access to powerful AI

Incidents like this often prompt calls for tighter restrictions on advanced AI models, but the reality is more complicated. Attackers can switch between models, providers or open-source alternatives when they encounter restrictions. Defenders, particularly smaller organizations, may not have those options. If advanced defensive capabilities become slow, expensive or heavily restricted, the result could be reduced defensive readiness without a corresponding reduction in offensive capability.

The objective should be responsible access. Verified identities, strong authorization controls, immutable logging, human accountability and time-limited access can provide meaningful safeguards without unnecessarily restricting defenders.

Preparing for autonomous systems

Organizations cannot improvise security once an autonomous system begins acting unexpectedly. The necessary controls must already exist.

This includes deny-by-default network access, workload-specific identities, short-lived credentials, segmentation between evaluation and production environments, telemetry that can correlate activity across systems, and rehearsed containment procedures. Organizations should assume that a capable autonomous system will discover every reachable path that advances its objective and ensure that no such path grants unintended authority.

These incidents should be viewed as proof that security architecture must develop alongside rapidly advancing autonomy. Organizations that adapt now will be better positioned to harness the benefits of AI tools while keeping risk under control. Those that rely on instructions alone may discover that technology follows the paths we leave open, rather than the rules we intended it to follow.

We've featured the best antivirus software.

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

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

Business leaders, governments, and tech giants are locked in a high-stakes race for AI supremacy, fueled by debates over the destructive power of superintelligence. In this sort of arena, it’s essential that enterprises stay grounded and resist the urge to start a race of their own, deploying ever-larger models simply for the sake of it.

In the context of enterprise AI, there is no prize for launching the most powerful model first, building the highest number of agents, or forcing AI into every workflow. True success is defined by solving genuine business problems and delivering outcomes that justify the underlying cost and complexity.

While the prospect of a single, omnipotent large language model (LLM) is inspiring, today's business realities demand a more measured approach: every problem has a right tool, and often, that tool is a simple one.

AI doesn't have to mean LLMs

Before generative AI dominated the conversation, businesses were already using machine learning for recommendations, forecasting, fraud detection and estimating things such as the lifetime value of a customer.

We have had years to understand these technologies, including how to deploy and monitor them and make predictions at scale quickly and cheaply. An LLM may be capable of tackling some of the same problems, but that doesn't always make it the better option. More sophisticated models can cost considerably more to run and introduce variability that is unnecessary for a straightforward prediction.

Sometimes AI isn't necessary at all. If a process follows known rules and needs to produce the same result every time (moving data between systems on a schedule or triggering an action when a predefined condition is met, for example), conventional automation may do the job perfectly well.

Long before agents arrived, businesses handled these tasks with pipelines, schedulers, and rules. If that approach is reliable, fast, and inexpensive, adding an AI agent doesn't make the process more intelligent; it can simply make it more complex.

Generative AI earns its place when creativity or interpretation are required: understanding language or images, for example, or handling a process where the next step changes according to context. Agents can go further by using that understanding to interact with systems and take actions, opening up processes that were previously extremely difficult to automate.

The future is likely to combine conventional automation, traditional ML, generative AI and agents, rather than one replacing everything that came before it.

The AI honeymoon is ending

These distinctions matter more as enterprise AI moves beyond experimentation. A proof of concept can demonstrate what a model or agent is capable of, but production introduces running costs, security, monitoring and the work required to keep increasingly autonomous systems under control.

An engineer once gave me an analogy that captures this well. Imagine growing a tree that you need to maintain in a particular shape. When a branch grows in the wrong direction, you cut it.

Then another appears elsewhere. The difficult part is finding all the branches. Developing an agent can be similar. Obvious behaviors can be controlled relatively quickly, but real-world use exposes edge cases that weren't visible during development.

The more freedom a system has to interpret information and make decisions, the more important evaluation, monitoring and guardrails become. Control needs to extend beyond model behavior to the data a system can access, the actions it is authorized to take, the information it retains and the point at which a human can intervene.

There is also cost to consider. The most capable models can be expensive at scale, so additional capability needs to create enough additional value to justify it. As experimentation gives way to scrutiny of ROI, businesses will become less interested in whether something can use AI and more interested in whether AI actually leads to an improved commercial or operational outcome enough to justify the cost.

Let the data decide

Avoiding over-engineering starts with the business need and the data, rather than choosing the technology first.

Data engineers and analysts are crucial because they understand where information comes from, how reliable it is and what it can actually tell you. AI and ML engineers can put mathematical models on top, but sophistication cannot compensate for data that doesn't support the task.

The data and desired outcome will often point towards the appropriate approach. Fast, reliable predictions from structured numerical data may be better suited to traditional ML.

Language, images and other unstructured information can make generative AI valuable, while processes that require human level understanding to trigger actions across different systems may justify agents.

And sometimes conventional software will remain the best answer. Starting with “we need an agent” and then looking for a problem to give it is the wrong way around.

The right people matter as much as the technology

Increasingly capable AI has also created an assumption that businesses will inevitably need fewer engineers. AI is certainly changing productivity: experienced engineers can complete work faster with these tools, while junior engineers who learn to use them effectively can become productive more quickly.

But businesses still need people who understand what is happening underneath. I like to compare it with smartphones. Most of us use a phone effectively without being able to explain the hardware, operating system and networks behind it. That doesn't mean the underlying engineering skills have disappeared. Somebody still needs to understand those systems.

AI is likely to develop in a similar way. Many people will become highly capable users, while businesses will still need engineers with the deeper technical knowledge to build, control and fix the systems underneath.

That means continuing to invest in junior talent too. Senior engineers eventually move companies, change roles or retire. Organizations still need mentorship and knowledge transfer so today's juniors become tomorrow's senior engineers.

The opportunity isn't people or AI, but the two together. Capable engineers equipped with these tools can achieve considerably more than either could independently.

The next AI advantage will be restraint

The wider battle over superintelligence and increasingly powerful models will continue, but it shouldn't dictate how enterprises make technology decisions. To make this technical restraint viable in practice, engineering leaders must actively manage executive and board-level fear of missing out, turning top-down pressure for a visible ‘AI strategy’ into discipline, value-driven execution.

Some business problems will justify sophisticated AI tools. Agents will make previously impractical forms of automation possible, while generative AI can transform how organizations work with language, images and unstructured information. Others will be solved perfectly well by traditional ML, conventional automation or straightforward software.

Knowing the difference will become increasingly valuable. Competitive advantage will come from knowing where AI materially improves an outcome, where a simpler technology is more effective and how to move the right systems into productions securely and economically.

That requires good data, strong engineering talent and enough technical judgement to resist complexity simply because it is available. In the next phase of enterprise AI, restraint will not signal a lack of ambition, instead it will be a sign of technical maturity.

We've featured the best AI website builder.

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

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

• Owl Labs argues that tracking workers' activities doesn't show how they're working
• 82% of workers believe employees should have to disclose the use of monitoring tools
• The biggest opportunity to improve productivity comes from upgrading hybrid meeting tech

New research has claimed almost three in five (59%) UK workers say their company is using monitoring software to track their activity both across basic attendance but also computer activity like mouse movements, keyboard usage and meeting participation.

And a further four in five (82%) believe that employers should have to disclose their use of monitoring software – something that's not currently required by law.

But despite a rise in workplace monitoring software, likely driven by a surge in post-pandemic hybrid working, the report from Owl Labs argues that tracking doesn't actually provide any meaningful insight into work at all.

Is your employer tracking how and when you work?

According to the data, the most commonly monitored activities are login/logout times (32%), number of meetings attended (22%), screen and mouse movement (17%) and keyboard usage (15%) – however, workers are already finding workaround to falsely appease tracking metrics.

For example, more than a quarter (28%) of UK workers admit so-called phantom meeting attendance, where they mark a meeting as 'attended' or even attend it without paying attention, and then catch up later via notes and summaries. It's something that managers (33%) are more likely to do than non-managers (15%), the report found.

The study also pointed to an increase in AI avatar interest, with 50% of managers and 21% of non-managers considering using an avatar to attend meetings on their behalf – further making meeting attendance metrics less important.

Hybrid work's impact on workplace tracking

Owl Labs also revealed some losses in efficiency as a result of hybrid working, but ironically, having workers placed across multiple locations isn't the issue. It's the tech they're being handed out.

Around three in four (73%) UK workers reported losing time because of technical difficulties during hybrid meetings, with audio problems generally the most significant. Teams spend on average six minutes to get a hybrid meeting started.

The report implies that spending money on better AV tech to support hybrid meetings may create a better outcome than simply deploying software to see who's attending those (suboptimal) meetings.

What does tracking software mean for workers?

Owl Labs CEO Frank Weishaupt urges companies to be more transparent about what they're tracking with their workers, helping improve trust between workers and higher up.

"With the Government reviewing how workplace monitoring should be regulated, now is the time for employers to be open about what they track and why," Weishaupt added. "Trust and reliable technology are what keep people productive, and what keep them in the business."

We've always liked testing Western Digital's WD Red Pro line of NAS HDDs, which emphasize reliable for long-term storage. During reviews, we've found it difficult to argue with the resilience of this drive.

For a limited time, the WD Red Pro 18TB hard drive is $700 (was $855) at Newegg when you use code BNWF49. That equals almost 4 cents per GB. While not the cheapest high-capacity drive (especially as component costs soar), for sheer reliability, the Red Pro range comes recommended by us.

• Best for: NAS owners, home lab builders, and small businesses that need high-capacity, always-on storage.

• Why we like it: 18TB of capacity in a single drive, 7,200rpm speeds with a 512MB cache, a 300TB-per-year workload rating for NAS systems with up to 24 bays, and a five-year warranty. Other capacities are available, from 4TB to 26TB.

• What you need to know: In general, we recommend sticking to a specific brand and model for your NAS system, as they're designed to work in harmony with others.

Today's best NAS drive deal

Why we recommend it

WD’s Red Pro line is a long-standing favorite among NAS users, built for the vibration and constant activity of multi-bay systems that run around the clock.

We haven’t reviewed the 18TB model, but we did test out the 14TB version in our 4-star review. It's not the fastest, but being designed for reliability over the long term is arguably more important if you're spending this much on a HDD.

For most users, 18TB strikes the sweet spot between high-capacity storage and lower initial investment compared to the 22TB to 26TB offerings.

Price Context & Historical Value

Currently, it feels like every component is spiking in price, and this HDD is no different. Just a few years ago, this same model sold for $300. Still, in today's market, it represents a good deal rather than a historic low. And the $155 off gives it a solid saving on one of WD’s highest-capacity NAS drives. The five-year warranty is a real advantage over standard desktop drives, which often come with only one or two years of coverage. If you need the space now, it’s worth grabbing before the code expires.

Should you buy it?

✅ Buy the WD Red Pro 18TB if... you need high-capacity, always-on storage for a NAS or home server, you want a drive rated for heavy workloads in multi-bay systems, or you value a five-year warranty.

❌ Skip the WD Red Pro 18TB if... you only need a simple backup drive for one PC, since an external drive is easier and cheaper; you’re building a quiet setup, since 7,200rpm drives are audible; or you can wait in case hard drive prices ease.

• See all external drives at Newegg.

The CMA has published roadmaps for Big Tech before, just not for Microsoft. With £400 billion in annual public sector procurement and millions flowing to Microsoft’s cloud, the CMA can’t wait until February 2027 to act.

The topic of public procurement is becoming a major issue for UK economic policy. Sarah Cardell, the CMA's Chief Executive, delivered a speech at a Parliamentary event at the House of Commons on 8 September, introducing two reports by the CMA published that day: Public Procurement in the National Interest and Rigged Bids, Real Costs.

The headline conclusion is, perhaps unsurprisingly, that UK public procurement could be leveraged far more effectively to shape markets and improve outcomes for citizens, businesses and taxpayers alike. But a more nuanced point, and arguably the most important one, is that the biggest risks arise when short-term cost savings are prioritized at the expense of long-term value and competition.

Cloud computing is one of the starkest illustrations of this trade-off in practice. Buying Microsoft is, in the immediate term, the path of least resistance: it is already embedded in most public-sector IT estates, integration is straightforward, and the near-term cost of switching or diversifying looks far higher than the cost of simply renewing. That short-term logic is precisely why Microsoft has wound itself so tightly around the machinery of UK public-sector cloud infrastructure.

Each individual procurement decision may look like sensible, low-risk value for money at the time it is taken but cumulatively, those decisions have entrenched a single supplier so deeply into public-sector data, identity and security architecture that it now poses a very real threat to competitive procurement in public-sector AI. The apparent saving today is a major liability for tomorrow, once every AI rollout has to be built on top of infrastructure only one vendor fully controls.

A market out of balance

The CMA's cloud market investigation reached a clear verdict in July 2025: competition wasn't working well. Microsoft and AWS each hold 30–40% market share with Microsoft's share still growing, and barriers to entry including data-center costs, economies of scale, incumbent product breadth. have been deemed likely to keep it that way.

Apparently, fewer than 1 per cent of customers switch cloud providers annually, held back by egress fees and technical friction. Most tellingly, the CMA found Microsoft withholds certain licensed products from AWS and Google that directly weaken those rivals; a problem with clear implications for public-sector buyers seeking competitive procurement.

A new AI market, losing balance

However, as Microsoft rapidly rolls out new AI tools across UK enterprises and the public sector, the fundamental issue of cloud infrastructure lock-in is being overlooked. The true barrier to a competitive AI market isn't just the stickiness of new daily workflows, but the interconnected stack comprising software, data, identity, security and the foundational grip Microsoft maintains through its legacy enterprise software, specifically Windows Server and SQL Server.

Once organizations are penalized for running these essential core workloads on rival clouds, their entire data gravity shifts permanently into Microsoft's ecosystem. Building subsequent AI processes and agents around these locked-in data structures makes opting for competing, innovative AI solutions financially and technically unviable.

The CMA has already diagnosed how Microsoft uses these core licensed products to weaken rivals; they must urgently address this foundational lock-in before it permanently closes off public procurement for enterprise AI as well.

Progress, but is it too cautious?

The CMA hasn't been idle. On 31 March 2026, the CMA accepted voluntary commitments from Microsoft and AWS rather than the binding designations recommended by its own investigators. These commitments included extending Microsoft's free egress-switch window from 60 to 180 days and a new cloud interoperability request mechanism, with a six-month review promised.

It simultaneously opened a strategic market status (SMS) investigation into Microsoft's entire business-software ecosystem including Windows, Office, Teams, and Copilot, encompassing over 15 million UK users with a decision due around February 2027.

That's a fair, if cautious, step, and critics such as the Open Cloud Coalition have welcomed it while pressing for swift action if commitments aren't met. Microsoft can reasonably argue these were offered in good faith to head off heavier intervention, but we need to start seeing these promises come into fruition quickly before further entrenchment occurs. Compare an AI rollout announced in June and largely complete by October against a regulatory clock that’s usually measured in years.

A roadmap for competition and social value

The CMA is now dealing with two closely connected issues: how public procurement can deliver greater value for taxpayers and society, and how Microsoft's market position can limit competition and choice. An interim roadmap could help address both at once. It would give public bodies clearer guidance on what to consider when procuring cloud and AI services, while setting out how the CMA intends to tackle the underlying lock-in that can narrow those choices.

This would support Andy Burnham's wider focus on delivering social value in public procurement. A more competitive and open cloud market would not just benefit businesses. It could give public bodies greater flexibility, reduce the risk that short-term savings lead to expensive, irreversible dependence on one supplier, and deliver better taxpayer value at the same time.

Certainty helps everyone

The roadmap should cover non-discriminatory licensing on rival clouds; interoperability extending to identity, permissions and agent portability; public-sector contracts with built-in exit plans; and independently verifiable evidence, not just documentation, that voluntary commitments are producing real switching and multi-cloud adoption. It should also provide broader transparency about how the CMA will assess whether these measures are working in practice.

Microsoft, its customers and its competitors all have an interest in predictable rules. Microsoft would know which integration practices remain legitimate; public bodies could invest in procuring AI at scale knowing it will not close off tomorrow’s choices, and rival cloud and AI suppliers would know if building interoperable products for the UK market is commercially worthwhile.

The CMA has diagnosed the problem, taken a first cautious step, and opened the right investigation. It should now set out the destination, the milestones and the tests for success before another year of AI adoption makes Britain's cloud and workplace-software markets harder to reopen. Without a clear roadmap or tests for success, we risk anti-competitive behavior leaking from the cloud market into deeply embedded AI work environments, at a far higher cost for the UK public sector to rewind.

We've featured the best cloud database.

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

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

For small business owners, an unexpected payment might look like good news. A new customer has placed an order, an invoice has been settled, or an unfamiliar transaction has landed in the account. But when money arrives without an obvious explanation, the smartest response isn't to celebrate, It's to investigate.

Payment scams exploit the everyday pressures of running a business. Entrepreneurs are busy, cash flow matters, and customer relationships depend on responding quickly and professionally. Fraudsters understand this, using unexpected payments, urgent messages and convincing instructions to persuade business owners to make decisions before they've had time to check the facts.

And unfortunately, the risks are becoming harder to ignore as criminals use increasingly sophisticated digital tools to make these communications more convincing and their scams easier to scale. For small businesses without dedicated fraud teams, knowing what to look for, and having the right safeguards in place, can make the difference between a suspicious transaction and an extremely costly mistake.

Why do "random" payments happen?

An unexplained payment isn't automatically fraudulent, but it should always deserve scrutiny. Scammers can use unexpected transfers as the starting point for several different schemes, exploiting the assumption that money received is money earned.

One common tactic is the refund scam, where a fraudster sends money to a business, apparently by mistake, and then contacts the owner asking for it to be returned. They may apologise profusely, invent an urgent personal emergency or insist that a quick refund is needed to resolve an accounting error. The catch is that they ask for the money to be sent through a different channel, such as a bank transfer or gift cards.

The original payment may have come from a stolen account or compromised payment card, so if the legitimate account holder later disputes the transaction, the initial payment can be reversed, leaving the business out of pocket for the money it sent back.

Another possibility is testing stolen payment cards or hacked accounts. Criminals may use compromised payment details to make small purchases or transfers, checking whether the credentials still work. An apparently legitimate transaction could therefore be part of a wider fraud operation, and if the real cardholder disputes the payment, the business may face a chargeback, fees or the loss of goods already supplied.

Then there are phishing setups, in which the payment itself is little more than bait. Once the recipient contacts the sender, the fraudster may provide a link to a fake payment portal, request login credentials or ask the business owner to confirm sensitive information. What begins as a question about an unexplained transaction can become an attempt to take over an account.

The common overall thread is manipulation - the scammer wants the business owner to act quickly, bypass normal procedures or share information they would ordinarily keep private that they can now exploit or sell on to the highest bidder.

The warning signs every business owner should know

Although payment scams can take different forms, they often share a recognisable set of red flags. Spotting them early gives businesses the opportunity to pause before money or sensitive information changes hands.

Look out for these warning signs:

• An urgent request for a refund, where the sender insists the money must be returned immediately, perhaps citing a deadline or personal emergency.
• Pressure to use a different payment method, where the scammer asks you to return money by bank transfer, gift card or cryptocurrency instead of using the original payment platform.
• Instructions delivered outside official channels, such as unexpected emails, messages and links which direct you to alternative websites or ask you to disclose account details.
• Payments that don't make business sense. A round-number payment arrives that doesn't match an invoice, order or product price, or the sender's details don't match the customer record.

The key principle is simple, but usually effective: never assume that money received is money safely yours to return or spend. Check the transaction independently through your payment provider or bank, and never follow refund instructions supplied by an unverified sender. If a payment looks suspicious, contact the provider through its official website or app and ask how to handle it - never initiate a separate repayment yourself.

Building a stronger defence against payment fraud

Avoiding these scams doesn't require a dedicated fraud department, but it does require vigilance alongside consistent processes and a willingness to verify unusual requests rather than act on instinct.

One method is to ensure you always keep business and personal finances separate. Dedicated business accounts make transactions easier to reconcile and suspicious activity easier to identify, and mixing personal and company payments can obscure unusual transfers and potentially expose personal finances to business-related fraud.

Another useful protection is understanding what your normal transaction patterns look like. Keeping track of typical order values, customer locations, payment methods and refund volumes should be common practice - but although a transaction that falls outside these patterns isn't necessarily fraudulent, it may deserve a closer look. Reconcile payments against invoices and orders regularly, and investigate discrepancies promptly.

Securing the accounts that control your money will also go a long way. Make sure you always use strong, unique passwords for banking, payment platforms and business email accounts. Enable two-factor authentication (2FA) wherever available, and turn on alerts for large transactions, password changes and new logins, and where possible, limit account access to employees who genuinely need it.

Giving your team clear rules is another key step, as anyone who handles customer payments should know that unexpected refunds require verification. Establish a policy that prohibits returning unexplained payments through alternative channels, and encourage staff to check suspicious messages with a colleague before responding. Regular training can help employees recognise fake invoices, phishing links and attempts to impersonate customers or suppliers.

Finally, recognise that payment fraud often begins outside the payment platform itself. A fake invoice might arrive by email, a fraudulent website might imitate a trusted service, or a phishing message might steal the credentials needed to access a legitimate account. That's why good financial procedures need to be backed by effective cybersecurity.

Why Bitdefender Ultimate Small Business Security is a valuable partner

For small businesses looking to bring these safeguards together, Bitdefender Ultimate Small Business Security offers a practical way to strengthen protection without having to build an in-house security operation.

The service combines antivirus and device protection with email security, password management, VPN access, digital identity protection for all employees and business asset exposure monitoring. Its scam protection features are particularly relevant to payment fraud, helping users identify and block suspicious messages, links and websites.

Its Scam Protection Pro tool allows employees to check a suspicious message or link before responding, clicking or paying, offering an additional layer of scrutiny when an unexpected payment request or refund instruction doesn't feel right. Web and email protection can also help guard against phishing attempts designed to steal credentials or direct employees towards fraudulent sites.

Centralised management makes it easier for business owners to oversee protection across employees and devices, while support for Windows, macOS, Android and iOS reflects the reality these days that key business activity rarely takes place on a single machine or ecosystem.

Financial controls, and suspicious payments should always be verified with the relevant bank or payment provider, but combining those controls with cybersecurity tools designed to detect digital scams can make it much harder for fraudsters to exploit a moment of confusion.

For start-up owners entrepreneurs, the lesson is clear: an unexplained payment is a reason to stop and check, not an invitation to act quickly. With consistent procedures, informed employees and a comprehensive security solution such as Bitdefender Ultimate Small Business Security, small businesses can reduce their exposure to payment scams while keeping their attention where it belongs - on serving customers and growing the business.

Try our free 30-day protection built for busy founders, not IT teams

Commercial engagement in biopharma used to center largely on direct conversations between sales reps and physicians, or pharmacists. As therapeutic areas have grown more specialized, more teams have become involved in engaging with healthcare professionals (HCPs), including key account managers, medical science liaisons, brand marketers and service teams. But these functions often work separately, leaving HCPs with disjointed experiences rather than interactions that build on one another and widen their understanding.

Biopharma companies are also engaging with a much broader network of HCPs, patients, health systems and administrators. Many of these stakeholders are limiting their contact with the industry, and are increasingly using AI to inform their decisions. As clinical complexity grows and access to HCPs becomes more restricted, companies need commercial models that are more responsive to customers’ needs.

To respond to these expectations, biopharmas are turning to Agentic Commercial, connecting AI with field teams, digital channels and approved content to create more consistent, personalized HCP experiences. Built into industry-specific applications, agentic capabilities help teams capture real-world insights, uncover treatment barriers, provide trusted answers and deliver relevant content across human and AI-led interactions. This leads to more meaningful engagement, better-informed HCPs and faster access to the right medicines for patients.

Field operations that turn conversations into real-time insight

An Agentic Commercial model transforms sales teams from promotional distribution channels into strategic sources of customer intelligence, evolving CRM from a system of record into a system of guidance.

Reps conduct nuanced HCP conversations every day, but traditional reporting systems restrict insights capture to rigid picklists and drop-down menus. Critical qualitative context — the 'why' behind a physician’s decision or the localized barrier to access — goes unrecorded.

Agentic call reporting uses agentic voice and free-text capabilities, with built-in compliance checks, to help reps capture richer details from HCP conversations. These interactions generate Commercial Evidence — actionable, real-world insights that reveal intent, content gaps, and treatment barriers facing physicians and patients.

Commercial leaders gain immediate visibility into real-world friction points. Early implementations show that 65% of Agentic Call Reports identify actionable treatment barriers, enabling field teams and medical specialists to address clinical concerns with precision.

Digital experiences built around immediate answers

Field teams build trusted human relationships, but they cannot be everywhere at once. Digital channels such as brand websites and email should extend the dialogue and enable more timely responses to questions, yet too often, physicians and patients must navigate websites or lengthy documents to find what they need.

Conversational AI transforms static digital experiences into real-time conversations. Using only medical, legal, and regulatory (MLR)-approved content, agents provide trusted, compliant answers to natural language questions in real time.

These interactions also generate valuable Commercial Evidence. Instead of showing only what users clicked, AI reveals what they asked, where they struggled, and the information they could not find. Commercial teams can identify content gaps, uncover emerging treatment barriers, and trigger timely follow-up from the field.

Approved content that gives reviewers more time for higher-value work

Continuous engagement depends on a steady stream of timely, compliant content. Yet, with global pharma content expenditure reaching $30 billion annually and 29% more assets created year-over-year, content volume is outpacing human review capacity.

In an Agentic Commercial model, accelerating content speed to market is essential. Agentic MLR acts as an end-to-end expert reviewer — understanding document context, mapping specific regulations, and executing targeted agent reviews against approved labels, claim libraries, and brand SOPs. This shifts human reviewers from line-by-line proofreading to high-impact work.

By accelerating content review while maintaining compliance, commercial teams can respond faster to customer questions across field and digital channels. This delivers a more cost-effective approval process, with content leaders anticipating 38% of the MLR process to be AI-driven by 2028.

Combining AI capabilities with human expertise

Agentic Commercial brings AI and commercial teams together within the same applications, following shared processes, governance and compliance requirements. By taking on repetitive tasks and analysing large volumes of data, agents give people more time to apply their scientific expertise and judgement, shape strategy and build trusted relationships.

Bringing field teams, digital engagement, approved content and AI together allows biopharmas to create a more connected and personalized experience across channels. As therapies become more complex and access to HCPs more limited, Agentic Commercial can help teams engage customers with greater precision, support better-informed decisions and ultimately help patients access the right medicines sooner.

We've reviewed and ranked the best Electronic Health Records software.

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

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

• Google leads efforts to dismantle 1.3 million device residential proxy botnet built through malware preinstalled on Android hardware
• Researchers say manufacturer Videostrong intentionally shipped malware and sold botnet access to cybercriminals
• Google seized infrastructure, blocked the company from Android certification, and strengthened device security checks

Google has revealed its role in taking down a massive residential proxy network and the malware that built it - with the unusual twist that the developer was a seemingly legitimate business with multiple products and countless trusting customers.

In a blog post, the company said that together with its partners, it dismantled a backdoor called XCPE. This malware was apparently used for ad fraud, served as a dropper for additional malware, and used for assimilating devices into an illegal residential proxy network.

In total, more than 1.3 million set-top boxes alone were part of the network, the company said.

No supply chain attack

Malicious residential proxy networks are nothing new - they emerge and get dismantled on a regular basis. But this one was unique, as it came preloaded on Android devices built and sold by a certified Android Original Equipment Manufacturer (OEM) and Original Design Manufacturer (ODM). In other words, people were thinking they were buying a legitimate product from a legitimate, Google-vetted company, but were instead getting malware-laden hardware that abused their trust and sold their bandwidth to cybercriminals.

In this case, not only did the company know what it was doing, but it was actively developing the malware and even worked on hiding it from Google and everyone else.

“A device manufacturer intentionally infecting their users with malware is an unacceptable breach of trust, and represents a serious risk to those users,” Google said, not hiding its disappointment with the manufacturer.

Videostrong ended up selling two things: On the surface, it was selling Android-powered smartphones and similar devices that contained malware that built a botnet. Under the table, it was selling access to that botnet to cybercriminals.

Google was joined by a number of other companies and agencies in this operation, including Japan’s National Police Agency, Lumen, HUMAN Security, Plume, Synthient, and TrendAI.

Threat actors are always in need of residential proxy botnets. They can use them to launch powerful Distributed Denial of Service (DDoS) attacks, hide their addresses as they operate, conduct large-scale credential stuffing campaigns, bypass geographical restrictions, scrape websites at scale, and more. We don’t know how much Videostrong made from selling access to the botnet, or who used it.

Taking down the C2 domains

To disrupt both the malware and the network, Google took down the domains and servers the company used to control the botnet and install malware modules, took down C2 domains that were used by over a dozen malware families loaded as modules via the XCPE backdoor, and banned Videostrong and all future subsidiaries and affiliates from building and selling Android devices in the future, or supplying software to other certified vendors.

Google also said it had improved its firmware integrity checks for certified Android TV devices to prevent similar incidents involving other manufacturers.

“Google requires OEMs go through a comprehensive suite of compatibility and security tests to ensure a consistent and safe user experience across the Android ecosystem,” it stressed. “However, Videostrong employed obfuscation techniques to exploit a gap in our tests and load malicious code after the device's first boot. As a result, Google has implemented changes to its existing review process to better safeguard against similar activity in the future.”

This is not the first time an electronic device came with malware off the shelf. Last month, security researchers Bitdefender found Android phones from multiple manufacturers, including Doogee and Cubot (both Chinese firms) being sold with preloaded malware. However, in those cases, the supply chain was compromised and the malware was included without the OEM’s knowledge, or consent.

OpenAI chief scientist Jakub Pachocki seemed to issue a warning on September 6.

In “An Alien Mind,” he argued that current alignment and monitoring techniques may soon prove inadequate for the pace at which machine intelligence is advancing. He also described a future in which AI takes a growing role in improving AI, compressing the time available for human supervision and institutional response.

Much of AI safety research concentrates on behavior. For example, what an AI system is permitted to do, how its actions are monitored and where humans retain approval. Yet every action begins one step earlier, with a judgment about what information deserves belief.

That earlier step is incredibly important.

AI acts on information whose pedigree is often obscure

Before an AI sends a payment, rewrites software or recommends a decision, it first assembles a picture of the world. That picture can draw on search results, corporate records, research papers and material produced by other AI systems.

The web was built for circulation. Information travels easily, while its ancestry often fades as it passes from one publication to another. A claim can begin in a research paper, appear in a news story, be condensed into a blog post and surface months later in a forum thread. By the time an AI encounters the statement, the route back to the original evidence may be difficult to reconstruct.

Authority also gets compressed. A peer-reviewed study, a corporate announcement and an anonymous post can all arrive as text. An AI system can rank and combine them into a polished answer, while the reader sees little of the difference between documented evidence, advocacy and repetition. Disagreement adds a further complication.

Credible sources regularly reach different conclusions, especially in finance, medicine, politics and emerging technology. AI is usually rewarded for producing a useful answer, which creates pressure to reconcile conflict into a single sentence. Uncertainty becomes easy to erase precisely when it deserves to remain visible.

The consequences grew substantially once AI acquired the ability to act. An agent can place a trade, approve a purchase, alter software or send a message on someone’s behalf. An uncertain claim can therefore become the premise for an irreversible action.

Human oversight depends on better evidence

The effectiveness of human oversight depends on the reviewer having superior information or a reliable means of checking what the AI used. In many settings, the human reviewer has access to the same web that the AI searched, only at a fraction of the speed. The AI can scan thousands of pages while the person responsible for approval works through a summary and a handful of links. As capability rises, review risks becoming a procedural signature rather than an independent examination.

Pachocki writes that OpenAI’s ability to rely on chain-of-thought monitoring is diminishing as reasoning becomes more complex and as AI becomes more capable of manipulating its own reasoning process. He calls for safety thresholds that can be enforced by third-party auditors, government agencies or international bodies.

Those safeguards govern capability and behavior. Reliable oversight also requires an evidentiary record showing what information produced the conclusion, where each claim originated, how trustworthy the source is and where credible disagreement remains. Without that record, the reviewer is judging the answer more than the evidence behind it.

AI inherited the web’s weakest habit

AI absorbed an information culture built around speed, reach and replication. Search engines, social feeds, aggregators and reposts made access extraordinarily efficient. They also encouraged a world in which the same statement could appear hundreds of times while tracing back to a single uncertain source.

Generative AI intensifies that effect because AI systems now produce material that later AI systems can ingest. A weak claim can be generated, scraped, summarized and republished until repetition gives it the appearance of consensus.

This creates a provenance problem with compounding consequences. Every additional generation can separate a statement further from the evidence that gave rise to it. Once an autonomous agent treats that statement as reliable, the information chain becomes part of the decision chain. While for consumers, this can mean a poor recommendation, for a financial institution, insurer or government agency, the same weakness can affect money, access, security or public services.

Autonomous AI needs a system of provenance

Take finance. Serious financial activity depends on records that establish who acted, under whose authority, on which terms and with what audit trail. Institutions spend heavily on those records because trust requires more than a plausible outcome. Autonomous AI needs a comparable discipline for information.

Claims should carry provenance that survives copying and transformation. A person or an agent should be able to trace a statement back to its origin, inspect the evidence behind it and see how the statement changed as it travelled. Credible disagreement should remain intact.

Where reputable sources diverge, the system should preserve the competing views and the degree of confidence behind each one. For an autonomous agent, uncertainty can serve as a trigger for further research, human approval or a narrower course of action.

Accountability should travel with information as well. A claim from an identifiable institution with supporting evidence deserves different treatment from an unattributed assertion repeated across dozens of pages. The aim is to give AI a richer sense of evidentiary quality before it commits resources or acts on someone’s behalf.

Governance belongs at a shared level. Provenance standards, verification methods and accountability rules need independent scrutiny and participation from many institutions, rather than dependence on the preferences of a single provider.

Final thoughts

Pachocki argues that current alignment and monitoring techniques may soon constrain the pace at which labs can responsibly scale AI tools, and he expects voluntary slowdowns until shared safety thresholds emerge. He also warns that continued progress could carry AI into recursive self-improvement, with AI driving a larger share of its own development.

That leaves every company building autonomous systems with a fundamental task. Establish what the agent knows before it acts and make that knowledge auditable.

Behavioral safeguards remain essential. So do permissions, audits and human approval. Their value rises sharply when the information beneath every decision carries provenance, context and accountability. The next phase of AI safety, therefore, begins with the evidence an agent sees, the authority it assigns to that evidence and the uncertainty it preserves before taking action.

Autonomy built on verifiable information gives humans a stronger basis for control. Opaque information leaves supervisors judging decisions whose foundations remain difficult to inspect.

Pachocki is right to warn that the window for preparation may be narrow. The industry should use that window to build systems that can show where their knowledge came from before those systems are trusted to act on the world.

We've featured the best AI chatbot for business.

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

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

OpenAI is preparing to combine the number of subscribers who can use ChatGPT and its Codex coding assistant. Instead of separate measurements, Plus, Pro, and Business subscribers will have a single gauge for both, according to the company's updated documentation. More complicated tasks will consume more of that allowance, although ordinary Instant conversations won't count against it.

The change comes as OpenAI prepares to bring ChatGPT Work's more autonomous capabilities into the main ChatGPT experience. Users will be able to ask ChatGPT to carry out complicated projects directly from a conversation. That convenience also means combining the resources available for their use, and I suspect plenty of paying customers are going to be unhappy about that particular trade-off.

OpenAI says most subscribers shouldn't find themselves waiting for their usage limits to reset more often than they do today. The company hasn't disclosed exactly how large the new allowances will be, making that reassurance rather difficult to evaluate. It's asking customers to trust that the subscription they're already paying for won't become less useful once the rules change.

OpenAI has made it clear it wants ChatGPT to become a comprehensive platform for both ordinary consumers and people using AI to run substantial professional projects. I can see the appeal of putting everything on one platform, but I think the company is underestimating how irritated people become when a familiar service starts reorganizing itself around customers with very different needs.

One subscription, too many expectations

For most people, ChatGPT is still a relatively straightforward assistant. They probably aren't especially interested in how many tokens a software development agent consumes while building an application. That makes the new shared allowance awkward. A person who uses ChatGPT for research and occasionally experiments with Codex could find those activities competing for the same resources.

Instant chat remaining outside the allowance is an important protection, and it means ordinary conversations shouldn't suddenly become impossible because someone used Codex too enthusiastically. But more demanding reasoning tasks will count, and those are precisely the capabilities that help justify paying for a subscription.

The eventual allowance could be generous enough that most people barely notice the change, and some users might benefit from spending more of their resources on whichever tool they prefer. But without clear figures, customers have no meaningful way to determine whether they're getting a better deal.

The reaction among existing users suggests OpenAI has already stumbled over that problem. "Worst decision ever by OpenAI" is the header and general sentiment on one Reddit thread. "Def unsubscribing for this one, it was the main appeal to have a 20$ sub next to the Claude's 200," one user wrote.

The anger reflects a broader worry that OpenAI is increasingly prioritizing professional AI tools at the expense of people who simply want to use ChatGPT without worrying about how much computing power they've consumed.

The frustration is understandable, particularly given OpenAI's recent adjustments to its Pro plans. The company reversed a proposed reduction in Pro 200 usage limits after subscribers complained, while also increasing allowances for other plans. That's welcome responsiveness, but it hardly inspires confidence when another substantial change to usage accounting is already approaching.

Business first

The more interesting question is what OpenAI ultimately wants ChatGPT to become. Its increasingly ambitious Work capabilities are designed to carry out substantial assignments, moving the product closer to a professional productivity platform. That's a big departure from the friendly chatbot that attracted millions by making AI accessible through ordinary conversation.

OpenAI seems determined to combine two rather different experiences, and the danger is that ordinary users end up navigating a service increasingly designed around professional workflows.

The company has good commercial reasons to pursue this direction. Businesses are willing to pay substantial amounts for AI that can perform valuable work, while consumer subscriptions face much greater pressure to remain affordable. Combining the experiences could encourage people to experiment with advanced capabilities, potentially making more expensive subscriptions attractive.

But if I pay for access to advanced reasoning, I'd like to know roughly how much I can use it without wondering whether another task has eaten into my allowance. Having to monitor token consumption feels like the sort of administrative chore AI was supposed to eliminate.

OpenAI should publish understandable allowances and explain how different activities affect them before the December changes arrive. Customers shouldn't need to conduct their own experiments to determine whether their subscriptions have quietly become less generous.

Making ChatGPT more powerful is worthwhile, but making people feel they're getting less for their money would be an exceptionally foolish way to celebrate that progress.

Meta might be building a 4.5ft-tall, 106lb humanoid robot that could look you in the eye, mimic human expression, and be a help around the home — a description that would fit almost any consumer-grade, humanoid robot from almost half a dozen companies. The report from The Information adds, though, that this bot will be " a physical embodiment of Muse."

Rarely has anyone put a finer point on why companies like Tesla, Meta, Unitree, Figure, X1, and Boston Dynamics are now racing to deliver home companion, assistant, and janitorial robots into homes.

These robots will essentially be Trojan horses for the increasingly unpopular AI systems these robotics companies, and their frontier and LLM partners, desperately need consumers to adopt to justify the hundreds of billions of dollars they are pouring into every part of the AI sector.

In the US, we sit on the precipice of a midterm election in which AI and data centers are, mostly indirectly, on the ballot. Voters are all but asking for an AI litmus test for candidates: where do they stand on the rapid expansion of data center construction? Obviously, it's not just the data centers. Voters are concerned about AI's role in misinformation, infrastructure security, as well as its role in the rapid deconstruction of the job market and workplace.

In short, AI is, for many (though certainly not all), if not a villain, then certainly something to be examined critically.

If more than half of Americans are at least "concerned" about AI, companies like OpenAI, Meta, Claude, and others may soon face a backlash that results in people pulling away from the use of chatbots and AI agents like Meta Muse.

AI and robotics — perfect together

Perhaps robotics faces some of the same challenges. Science fiction has played its role in making many fear the robot uprising. Still, social media is lately filled with humanoid robotics feats of dexterity, dancing, and even terrifying fighting skills. People might be worried about Terminator, but can't help but be intrigued by a tireless android ready to take on all your least favorite chores.

It can be easy to forget that AI is the thing that's supercharging robotics development or that Nvidia, the most valuable company in the world, is the one most responsible for the chips powering the AI inside many data centers.

Partnerships between AI and robotics companies are everywhere. Meta's rumored robot is reportedly using Nvidia's Jetson Thor chip. That platform can also be found in Figure AI robots and Agility Robotics, among others. Google's DeepMind is partnered with Boston Dynamics. Over the years, Amazon has bought its share of robotics companies, often redirecting them to work in its vast supply chain and inventory operations.

For any of these companies, though, I think it's safe to assume that the potential home robotics revolution is an attractive target for their AI aspirations.

Even Apple has reportedly looked to combine robotics with Siri AI. Its new HomeView 7-inch desktop display will arrive with Siri AI baked in, and a rumored follow-up might include an animated head that can look at and track you as you, say, move about the kitchen and consider your next Siri AI prompt.

What's really inside your robot

If any of these companies succeed in integrating home-targeted AI systems and delivering them in useful humanoid robotics, it will not only be a game-changer for consumers, but will also open the door for what I assume is the ultimate goal: locking consumers into long-term AI and robotics subscription programs.

After all, most of these robots will be far too expensive for any consumer to buy outright, but monthly leasing fees that might wrap in model access and even AI tokens would likely appear more approachable for most consumers.

Instead of paying for access to Meta Muse, Claude, ChatGPT, Alexa+, or SiriAI, they'd be paying for cleaning and assistant services that come with their friendly neighborhood humanoid maid.

Getting real about humanoid robots

Assuming that's the plan, it suffers from one major flaw: the readiness of any of these humanoid robots for the home.

Don't be misled by those aforementioned viral videos. Boxing robots, for instance, are almost entirely teleoperated. Most of the best stuff many of these robots do is still via remote control (or is canned programming and not true autonomy).

I'm not arguing whether these robots can walk, run (quite fast), and dash up a flight of stairs, but they remain unpredictable and, I think, potentially unsafe in the average home.

That will, of course, change. Five years from now, the first safe and reliable humanoid robots may be ready to clean your home, walk your dog, and even look after you and your children. And you'll be paying pricey monthly fees that will funnel back to AI companies building more data centers and developing AIs that are, themselves, building better models.

The robots are coming, and they're bringing the future of AI with them. Just be ready to pay for it.

OpenAI, Google, Microsoft and Anthropic have agreed to support a new way for website owners to stop their content being used to train AI models. The companies are backing a shared opt-out signal in a website’s robots.txt file, which tells AI crawlers whether they're allowed to use its content for training.

Like a modern day Library of Alexandria, the internet is the single largest source of human knowledge that exists. Countless people have spent their lives cataloguing and uploading the entire history of humanity for the betterment of everyone.

AI companies are a tinderbox in this regard. To train and develop better capabilities for their models, AI companies are repeatedly scraping every corner of the internet to create archives of training data. Compounded by changes in search and how people seek information, AI threatens to become the single point of access to information.

But just because you have a website, it doesn’t mean you want it constantly scraped by bots stealing your answers to life’s difficult questions so people can access them elsewhere.

Luckily, you’re not alone. Outrage from users and website hosts across the internet has been fueled into developing methods to keep AI companies from harvesting data from your websites.

Like layering slices of Swiss cheese, these methods work best to keep out scrapers when used in a defense-in-depth manner. But some malicious bots will still find ways to navigate through the holes. Here’s a few ways to keep the honest bots at bay:

1. Robots.txt

This is the crux of OpenAI’s method to honor anti-web scraping markers. The ‘robots.txt’ text file sits in the root directory of your website and provides guidance on how bots should behave. This can be a useful way to provide a digital map to search engine crawlers and can also prevent your site’s servers from becoming overburdened by bot requests.

But by adding just a few characters to this file, you can block the majority of bot traffic from accessing your website. Of course, this unfortunately only applies to scrapers that actually abide by the rules of your robots.txt file.

OpenAI has set something of a precedent for other AI companies here, adding that, "We encourage other AI providers to honor the signal."

There are a few ways to manage which specific bots you would like to allow on your website, but to block AI traffic you can take the following steps:

• Open your robots.txt file, or create one if you don’t have one using a text editor
• Add a section to the file using the heading ‘# AI crawlers’
• To block all AI crawlers, add a line below stating ‘Disallow: /’
• To block a specific AI crawler, such as ChatGPT, place ‘User-agent: ChatGPT-User’ on one line, with ‘Disallow: /’ underneath
• You can prevent AI crawlers from accessing specific pages by adding the web address after the ‘/’. For example, I could stop AI crawlers from accessing my author page by using the rule ‘Disallow: /author/benedict-collins’
• Update your robots.txt file in the source files of your website

2. User-only viewing

Alternatively, you could take the planned Reddit approach to blocking AI traffic by requiring all users to have an account in order to access your website. This is an excellent way to filter out human traffic from bot traffic because user accounts usually need to be verified with an email or phone number - something mostly humans have.

But this comes with its own caveats, and Reddit is finding out the hard way in feedback from users. Not requiring an account gives users from all over the web an easier search experience. They can find the information they need without any obstacles, and without handing over personal information to yet another website.

Having to log in or create an account to view web content can cause people to ‘bounce’ - they click on your website, see they need to create an account, and immediately back out of your site to seek the information they need elsewhere. So tread carefully if your traffic matters to you.

3. IP limiting

AI bots may descend en-masse using a single IP address to access your website. Third-party services offer this as an option to prevent your website’s servers from being bombarded with quick-moving traffic, but it can come with a monetary cost.

This method can be used to set up a timer on specific IP addresses that are detected as bot traffic. For example, if a single IP is sending multiple access requests, you can adjust the settings so that only 5 requests from the IP can be made in a 10 second period.

But beware that some AI bots or web scrapers will use rotating IPs, hopping between them to avoid a block from a single IP.

4. Captchas

A slightly archaic method of human verification that most people have a love/hate relationship with. I don’t mind counting the number of buses or bicycles, or aligning the correct jigsaw piece with its place, but some people see it as a humiliating ritual to prove their humanity.

What’s worse is that some AI models are getting good enough to solve captchas themselves.

It can help to filter out relentless attempts from web scrapers and less capable AI bots, but it could also cause users to bounce - especially those with a particular fear of ClickFix attacks who may see a complicated captcha as a way to hijack their device.

So use these methods cautiously for they is a double-edge sword. You can stop your data from being harvested by AI bots but it may also harm the experience of your users.

• Rapid drug discovery using AI's assistance is adding pressure to an already restricted macaque supply
• A US startup plans to breed 30,000 monkeys over two decades
• China's export ban cut off a major source of American research animals

Artificial intelligence is creating new demands for pharmaceutical research, increasing pressure on all kinds of resources needed to develop and test potential treatments - including animal test subjects.

Macaques, commonly used in medical experiments, have become considerably more expensive, with reported prices rising from approximately $3,000 in 2019 to more than $20,000 today.

Now , Palantir cofounder Joe Lonsdale is reportedly supporting Safer Human Medicine, a startup seeking to establish a large-scale breeding facility in the US.

Rising demand meets restricted monkey supplies

The proposed facility would eventually accommodate 30,000 monkeys, with the company planning to increase its breeding population gradually over approximately 20 years.

Its initial stock would comprise between 500 and 1,000 macaques obtained from breeding facilities across Asia.

The company intends to expand domestic production and reduce American laboratories’ dependence on imported animals for pharmaceutical research and testing.

Artificial intelligence has intensified the pressure by enabling researchers to develop potential medicines more rapidly and generate more drug candidates for testing.

Each promising candidate may then require laboratory studies to assess toxicity, safety and biological effects before human clinical trials.

Some experimental treatments require animal studies before human trials, and macaques are frequently selected because of their biological similarities to humans.

However, the supply problem predates the recent expansion of artificial intelligence applications within pharmaceutical research and drug discovery programmes.

China, previously responsible for approximately 60% of America's laboratory monkey imports, prohibited exports in 2020, forcing researchers to seek alternative suppliers.

American laboratories subsequently turned to Cambodia, but investigations in 2022 uncovered allegations involving wild-caught monkeys falsely represented as captive-bred animals.

Breeding limitations and animal welfare concerns

Expanding captive populations is not an immediate solution because female macaques generally reach reproductive maturity around four years of age.

Each female typically produces one offspring annually, limiting how quickly breeding operations can increase their populations to meet laboratory requirements.

China now reportedly charges approximately $26,000 per macaque, roughly nine times the price recorded in 2019.

American laboratories reportedly import approximately 20,000 monkeys annually, leaving researchers exposed to international supply restrictions and substantial procurement expenses.

Safer Human Medicine's proposed operation would therefore require considerable time before its intended population could supply a significant share of domestic demand.

Nevertheless, the plan has already attracted criticism from animal welfare advocates concerned about keeping thousands of primates for experimental purposes.

A critic on X described the proposed breeding operation as “morally indefensible,” and others have called for judicial intervention against what they consider animal abuse.

Their objections could lead to legal challenges, potentially delaying construction, breeding activities, or the company's plans to establish domestic production capacity.

The eventual progress of this project will depend not only on breeding capacity and commercial demand, but also on regulatory scrutiny and potential court proceedings.

Via Ole Lehmann on X

• Russian concerns about Ukraine drone strikes have prompted a review of data center resilience to strikes
• Defense measures include anti-drone nets, signal jamming hardware, and even camouflage
• The Digital Development Ministry has also proposed moving facilities underground

Ukraine’s increasing targeting of Russian data infrastructure has resulted in Moscow considering new defensive measures against drone strikes.

The Digital Development Ministry has proposed a list of countermeasures and precautions against attack, ranging from jamming hardware and camouflage to building subterranean data centers.

With experts warning that a data center outage could leave a facility value wiped by up to 87%, Russia is taking steps to protect is cloud and AI facilities. But such action could increase costs for each facility by around 10%.

Pushing up the potential cost of future data centers, safety measures include protection against shockwaves for servers, power supplies, and cooling systems.

Drone strike risk

Throughout 2026, Ukraine has stepped up its retaliatory strikes against Russian industrial assets, most recently with five data centers targeted in October. Among these was country’s largest facility, based in ‌Balakovo and owned by Sberbank.

Others targeted include three data centers owned by Yandex. At the time, Sberbank First Deputy CEO Alexander Vedyakhin told reporters that "We confirm that the data processing center was attacked. At the same time, all of Sberbank's systems and services are operating normally and without disruption."

The Digital Development Ministry recommendations were seen by Russian publication Kommersant, which noted that the ministry had confirmed “The development of recommendations has been completed, the document [...] sent to the relevant organizations.”

It appears that the document was initially circulated on October 21 2026, possibly in response to the ‌Balakovo data center strike, and certainly as a reaction to the wider concerns. Kommersant indicates that Russian businesses and citizens are concerned about Ukraine’s targeting of internet infrastructure.

Months of downtime

Not all data centers are considered to be equal within the plans, which divide the facilities into three categories. K1 indicates strategic targets, K2 is economic targets, while K3 is a signifier of a corporate site.

While the exact strategies for each category is unknown, it seems likely that the K1 targets would be subjected to the largest defensive measures. While it might be costly to install shockwave protection, anti-drone nets and signal jammers are likely more affordable elements.

The cost of installing shelters and other protections is preferable to the alternative, according to Roman Chustuzian, senior investment director of KAMA FLOW, who told Kommersant: “Even if shelters, grids, sensors and duplication of connections add 5-10% to the cost of the object, it is not comparable with the loss of 87% of its cost and months of downtime.”

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

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

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

Quordle today (game #1744) — hint #1 — Vowels

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

Quordle today (game #1744) — hint #2 — repeated letters

Quordle today (game #1744) — hint #3 — uncommon letters

Quordle today (game #1744) — hint #4 — starting letters (1)

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

Quordle today (game #1744) — hint #5 — starting letters (2)

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

Quordle today (game #1744) — the answers

The answers to today's Quordle, game #1744, are…

• TRIAL
• STAID
• LEVEL
• BONUS

A tricky five-vowel game that I had an incorrect hunch would also include a rare letter.

Instead of a J for “jewel”, my old friend the letter V took a starring role for the far more everyday LEVEL.

Daily Sequence today (game #1744) — the answers

The answers to today's Quordle Daily Sequence, game #1744, are…

• BLAND
• CLOSE
• MAMBO
• TODAY