TechRadar News

• Nvidia stays king of AI chips, buying millions of HBM from Samsung in 2026
• Google’s TPU push makes it Nvidia's closest rival for HBM
• Amazon, AMD, and Microsoft fight for scraps while Intel fades into obscurity

High-bandwidth memory is essential for AI and cloud computing, offering high-speed data transfer and efficiency for demanding workloads.

The HBM space is currently dominated by South Korean memory giant SK Hynix and its chief rival (and neighbor) Samsung, although other players, like Micron, are also looking to grow their share of the market.

Tech giants like Nvidia, Google, Amazon, and Microsoft are among the biggest purchasers of HBM, and an internal Samsung roadmap spotted by ComputerBase provides a fascinating insight into their projected shopping lists for 2026.

Google in second place

As you would expect, AI darling Nvidia remains the top purchaser of HBM from Samsung, with demand expected to grow from 5.8 million units in 2024 to 11 million units by 2026.

This increase is fueled by Nvidia's expanding lineup of AI accelerators, as the H200, equipped with 144GB of HBM3, is projected to need 1.7 million units in 2026. The B200 and GB200, using 192GB of HBM3e (8hi), will require 3.15 million units. The B300 and GB300, with 288GB of HBM3e (12hi), are forecast to take 2.7 million units, while the GB300A, using 144GB of HBM3e (12hi), is estimated to take 1.8 million units. The R100, expected to sport 384GB of HBM4 (12hi), will need 1.5 million units.

Based on the roadmap, which ComputerBase says "probably dates from the beginning of 2024", so is therefore not fully up to date, Google is expected to become the second-largest buyer of HBM, driven by its investment in Tensor Processing Units (TPUs).

Google's total HBM demand is projected to grow from 2.4 million units in 2024 to 2.8 million units by 2026. TPU v7p alone is expected to require 1.7 million units in 2026. Alphabet has allocated $75 billion in capital expenditures this year, a portion of which will likely be spent on HBM procurement and AI infrastructure.

Amazon’s demand for HBM is also growing steadily, as the company’s Inferentia and Trainium chips are projected to need 1.3 million units of HBM by 2026. Trainium 3 alone is expected to require 900,000 units that year.

AMD is largely seen as Nvidia’s closest rival, but it trails far, far behind. The company’s total HBM demand is projected to reach 820,000 chips in 2026. The MI400, expected to launch in early 2026 with 384GB of HBM4 (12hi), will be responsible for 400,000 units that year.

While Microsoft integrates HBM3 in its Maia AI chips, its total HBM demand is expected to reach just 240,000 units in 2026, indicating that it remains a minor player in this space.

Intel is also struggling to compete. The roadmap shows that its Gaudi3 accelerator is losing traction, with chip demand declining from 128,000 units in 2025 to none currently forecast in 2026 (although bear in mind that might have changed since this roadmap was created).

Of course, these numbers only cover HBM supplied by Samsung. Factoring in memory supplied by SK Hynix and Micron could paint a different picture, albeit one in which Nividia is still comfortably the top purchaser of HBM.

(Image credit: Samsung) You might also like

• Samsung to concentrate on enterprise market with focus on HBM
• Samsung confirms next generation HBM4 memory is in fact Snowbolt
• Samsung scientists are working on a new type of memory

• ViewState code injection attacks can lead to remote code execution, Microsoft warned
• Many devs are not generating their own machine keys for ViewState
• There are thousands of publicly available keys cybercriminals can use

Cybercriminals are abusing a weakness in ASP.NET websites to remotely execute malicious code, according to Microsoft’s Threat Intelligence team, which has published an in-depth analysis on the new method.

In the article, Microsoft explained threat actors were injecting malicious code through a method called ViewState code injection attacks.

ViewState is a feature in ASP.NET websites that helps remember user input and page settings when the page is refreshed. It stores this information in a hidden part of the webpage so that when the user interacts with the page again, it can reload the saved data without losing anything.

Accepting malicious code

As it turns out, many developers are using machine keys (security codes designed to protect the website’s ViewState data) that they find online, rather than generating their own. These machine keys are intended to prevent tampering with ViewState, which tracks data on web pages as users interact with them.

However, if developers can find these keys, so can criminals. When they do, they can use them to inject harmful content into a website’s ViewState. Because the machine key is the same as the one the website expects, the server decrypts and processes the malicious code, allowing attackers to run their own commands on the server. This can lead to remote code execution, Microsoft warned.

The researchers found more than 3,000 publicly disclosed keys that can be used in these attacks. In some cases, the researchers added, developers might unknowingly push these public keys into their code, as well.

To prevent these attacks, Microsoft advises developers to generate their own machine keys, avoid using default or publicly available ones, and secure sensitive data by encrypting parts of their configuration files.

Upgrading to a newer version of ASP.NET is also recommended, as is using security features such as the Antimalware Scan Interface (AMSI).

Microsoft also provided instructions on how to remove or replace the insecure machine keys from the server's configuration files and removed examples of these keys from its public documentation to discourage the insecure practice.

You might also like

• Hackers hide malware into website images to go unnoticed
• We've rounded up the best password managers
• Take a look at our guide to the best authenticator app

• Sony has seemingly removed a number of "spam" PS4 and PS5 games from the PS Store
• This follows recent comments made by developers criticizing the
  "slop" games featured on the digital storefront
• Developer RandomSpin has had most of its games removed, including Bodycam Shooter, Supermarket Simulator Pro, and more

It looks like Sony may have quietly removed a number of "spam" shovelware games from its PlayStation Store.

This comes after a recent investigation from IGN which featured interviews from developers criticizing the lack of quality control across the PlayStation Store and the Nintendo eShop in comparison to the Xbox store, as well as the never-ending list of "spam" and "slop" games, and poor discoverability.

Now, a number of these low-quality PS4 and PS5 games have seemingly disappeared from the digital storefront (via True Trophies).

Over on PSNProfiles, players have found evidence to suggest that developer RandomSpin - which released over 40 titles in just 2024 alone - has had most of its games removed, including Bodycam Shooter, Supermarket Simulator Pro, Backrooms Brotherhood, and Backrooms Inside The Escape.

These titles in particular all seemed to use recycled assets, along with AI-generated images, to copy other popular games currently available, like Reissad Studio's Bodycam or Supermarket Simulator from Nokta Games.

It's unclear if the sudden delisting of these games was made by Sony as a result of the recent criticism, and it could be purely coincidental. Sony hasn't commented on the removal at this time.

In other news, Sony has announced that its first State of Play livestream of 2025 will air this week on February 12, 2025. The showcase will run for 40+ minutes and offer "news and updates on great games coming to PS5", which hopefully means we'll learn the release dates for Death Stranding 2: On the Beach and Ghost of Yōtei.

You might also like...

• Sony will release an Astro Bot PS5 bundle in March according to new leak
• Grand Theft Auto 6's release could mean an uptick in PS5 and Xbox Series X sales, according to Take-Two boss
• An all-digital PS6 doesn't seem possible, according to ex-PlayStation boss Shawn Layden: 'I think it would be hard for them to go fully disc-less'

• New research examines battery degradation in different devices
• Design might make more of an impact than you expect
• Understanding the issues might lead to longer-lasting tech

It looks like it’s official: batteries may well be running out of juice at an unpredictable, somewhat alarming rate. And we’re not talking about the bunny-branded AAs you stick in the remotes of one of the best TVs.

Instead, new research suggests internal batteries are suffering from escalating battery degradation due to real-world factors that can’t easily be measured in a lab, with batteries in wireless earbuds used as the big test case example.

All lithium-ion rechargeable batteries will degrade and lose their maximum charge over time. but using X-ray infrared tech, international researchers at the University of Austin Texas (as reported by SciTechDaily) are trying to get to the bottom of why some devices devices may drop charge faster than you expect.

The quest to discover why lifetime battery length is now seemingly diminishing at a faster rate in buds was inspired by Yijin Liu’s frustrations with his headphones. After only wearing the right one, the associate professor at the Cockrell School of Engineering discovered “that after two years, the left earbud had a much longer battery life.”

This then prompted Liu to spearhead fresh research that has since been published in Advanced Materials (a weekly scientific journal that has been going for over three decades).

Batteries not included

(Image credit: The University at Austin)

According to the team’s findings, it appears real-world factors, such as sudden temperature changes and air quality can fundamentally damage long-term battery life in your favourite tech. And that’s despite the fact internal batteries are normally tested under extreme lab conditions.

Other internal components can also have a negative effect on earbuds battery health; such as the positioning of internal mics and other circuitry causing subtle conflicts with your buds’ battery chemistry.

The fact we all use wireless earbuds and our smartphones in vastly different environments under varying degrees of stress levels has led the researchers to rethink how they can redesign electronic devices to withstand a greater variety of real-world conditions.

“They [electronic devices] could be exposed to different temperatures,” says Guannan Qian, who posted the first paper as part of Yijin Liu’s study. “One person has different charging habits than another, and every vehicle owner has their own driving style,” which “all matters” according to Qian.

Cue a battery (sorry, not sorry) of experiments where Liu’s team joined forces with various international labs to learn more about real-life battery degradation. Working with the likes of Brookhaven National Laboratory’s National Light Source II (phew!), and France’s European Synchrotron Radiation Facility, their goal has to be to unearth the secrets of how and why batteries are reacting differently to real-world conditions compared to lab environments.

So what does this mean for you and the best earbuds going forward? Well, advanced X-ray imaging seems to hold the key. According to Brookhaven’s National Laboratory, physicist Xiaojing Huang believes that they must “understand the differences between lab conditions and the unpredictability of the real world and react accordingly” in order to “discover and develop new types of batteries,” according to Huang.

That's easier said than done, of course – solid-state batteries (generally considered to be the next big thing) remain elusive. But if Liu and his fellow energy-obsessed researchers can make progress in their experiments in how real-world factors affect – and in the case of earbuds in particular – diminish battery health, we can maybe at least make current batteries last longer. Or acknowledge more than ever the need for easily replaceable batteries, such as those in the Fairbuds.

You might also like

• The best earbuds of 2025: wireless and wired buds for all budgets
• Xiaomi tipped to set impressive new phone battery benchmark in 2025
• The best power bank 2025: top portable chargers tested

• iOS 18.3.1 and macOS 15.3.1 launched this week
• Some users who had previously disabled Apple Intelligence have found it re-enabled on the new software update
• Easy to disable again, but it's still an annoying bug

Have you turned off Apple Intelligence? Well it might've been turned on again when you updated to the latest iOS 18.3.1 or macOS 15.3.1.

Some users are reporting that the new software updates that released yesterday with security updates has been re-enabling Apple Intelligence, despite users having turned off Apple's AI prior to updating.

The issue was brought to our attention via a blog post by developer Jeff Johnson who says, "Some people who had previously disabled Apple Intelligence in macOS 15.3 and iOS 18.3 saw it re-enabled after updating to macOS 15.3.1 and iOS 18.3.1 today. In fact, I personally have two different Apple silicon Macs running macOS Sequoia, and after I updated both Macs to 15.3.1, Apple Intelligence was re-enabled on my MacBook Pro but not on my Mac mini."

Other users have also reported an issue were the 'Welcome' screen essentially turns on Apple Intelligence and removes your previously registered preferences.

It's fair to assume that this bug should be fixed promptly, although it's worth remembering before you see Apple Intelligence features like Notification summaries, Mail categorization, and ChatGPT in Siri re-enabled on your devices.

AI should always be opt-in

(Image credit: Apple)

Apple, of course, wants users to use Apple Intelligence on compatible devices but fortunately there's the built-in option to disable the AI tools. While this particular bug isn't that inconvenient and can be rectified by simply heading to Settings>Apple Intelligence & Siri>Toggle off, it's still something that might catch you off guard.

We expect Apple to fix this issue swiftly and have reached out to the company for comment. Apple Intelligence continues to improve daily and we're expecting to see even more AI tools when iOS 18.4 releases later this year. If you've not used Apple Intelligence in a while, this might be the nudge you need to give it a try again.

You might also like

• The biggest Apple Intelligence upgrade of the year is expected to land in April
• What is Apple Intelligence? The new AI powers coming to your iPhone, iPad and Mac explained
• Apple Intelligence: here's a full list of compatible iPhones, iPads and Macs

Independent auditors have confirmed that one of the best VPN providers never stored your data.

Security experts at leading auditing firm Deloitte inspected NordVPN's server configuration and relevant IT systems to ensure data related to users' activities is never logged as stated in its privacy policy. The independent audit found the VPN's infrastructure to be "properly prepared," meaning the provider cannot know anything about online user activity when connected to the service.

NordVPN regularly puts its security and privacy infrastructure under independent scrutiny as a way of "showing our users that we care about their privacy." This is the fifth time, in fact, the popular VPN provider has proved its no-logs claims with an external audit since 2018.

NordVPN 5th no-log audit

Deloitte auditors conducted a series of privacy and security analysis between November 18 to December 20, 2024.

This included inquiries with NordVPN employees, an inspection of all types of VPN servers the provider offers (standard servers, double servers, obfuscated servers, onion over VPN, and P2P servers), their configuration, and their deployment, as well as a review of all the relevant privacy settings and procedures.

After this thorough analysis, experts concluded that the VPN is indeed designed in accordance with its no-log claims as "the configuration of IT systems and management of the supporting IT operations is properly prepared."

If you're subscribed to NordVPN, you can access the full audit report by logging in to your Nord Account and tapping on the report section in the user control panel.

Commenting on the findings, CTO at NordVPN, Marijus Briedis, said: "The trust we earn from our customers underscores everything we do in the cybersecurity industry. To maintain that trust, we not only strive each year to innovate and develop world-leading cybersecurity products, but we also fully commit to our promise not to monitor or record our users’ online traffic."

As mentioned earlier, this is the 5th time the team successfully put its no-log privacy policy under external scrutiny since 2018 – with the latest assessment conducted in 2023 – showing the company's continuous commitment to privacy.

"Having this assurance reaffirmed by independent, globally respected researchers for the fifth time demonstrates that privacy isn’t just a buzzword at NordVPN – it’s in our DNA," Briedis concluded.

The importance of using a no-log VPN

Using a trustworthy no-log VPN is your guarantee that no personal information or usage data can ever be collected and then linked to you or your online activities.

This is why a strict no-log policy (bonus points if its regularly independently audited) should be your main priority when choosing a virtual private network service, especially if your main use case is better online privacy.

Privacy isn’t just a buzzword at NordVPN – it’s in our DNA

Marijus Briedis, NordVPN

It's worth noting, however, that even no-log VPNs need to collect some basic data in order to work properly. These include the number of users connecting to the same server as well as the email address associated with a user's account. Nonetheless, these details won't be enough to identify you online or anything you do when connected to the VPN app.

Not having your usage data stored on the VPN servers is advantageous as, even in the case that a hacker or government manages to breach the provider's security network, your sensitive information won't be there.

The real-life importance of a no-log policy has already been proven on a few occasions across the VPN industry. In one such instance, Swedish authorities were left empty-handed after an inconclusive police raid on Mullvad's servers failed to find any stored users' details.

I like to imagine someone swimming in the Gulf of Mexico when Google unceremoniously renamed it to the Gulf of America. There they are, backstroking across the pristine green and blue waters as the giant words float down from the sky and land on top of them. They're distressed and confused, quickly kick out from under the huge letters, pull out their iPhone, and open Maps to switch it back to the Gulf of Mexico.

That's right, there's now technical disagreement on the name of a major body of water across two hugely popular map systems: Google Maps and Apple Maps.

Here's how it started:

Donald Trump takes the oath of Office as US President on January 21, 2025, and immediately announces his intention to rename the Gulf of Mexico to the Gulf of America, along with renaming Mount Denali back to Mount McKinley, and other geographic ambitions like annexing Canda, taking over Greenland, and reclaiming the Panama Canal.

While some of these ideas seemed far-fetched or ridiculous, Google soon announced it would be changing the name of the Gulf of Mexico in Google Maps. As of February 10, 2025, the change is live on all versions of Google Maps that, at least, Americans can see.

The hard truths of cartography have gone wobbly so anything is on the table.

Apple, however, has resisted, or perhaps they've postponed the change until a future iOS 18 update.

I've asked Apple and Google for comment. In particular, I want to know how Google made the decision and what technical hurdles exist when you're trying to region-set map labels for the world.

Someone is opening Google Maps somewhere in Mexico and still seeing "Gulf of Mexico." If they slide into the water and swim toward their American friend paddling near Texas, at what point does their Google Maps view change? Perhaps they'd consider switching to Apple Maps for a singular view of this 617,800 square mile, watery expanse.

Enter confusion

Google's choice is not without consequences. By so quickly acquiescing, does that mean that our US Google Maps experience becomes a moveable feast? Should we expect Mount Denali to switch over to McKinley (it's still Denali on Google Maps)? Will Google preemptively put an asterisk next to Canada or Granada? (*Label under interest of current US administration).

From a technical standpoint, I do wonder if those traveling from, say, the UK to the US for a holiday in the Gulf would find their Google Maps switching when they arrive in the States.

Since maps are now so changeable, perhaps Google (and Apple, too) can introduce some customization tools so we can all set our labeling preferences. I'd like to continue calling that water zone "The Gulf of Mexico." Can I set that somewhere? Right now, no, but the hard truths of cartography have gone wobbly so anything is on the table.

Apple holds – for now

A tale of two gulfs: Apple Maps is on the left and Google Maps is on the right. (Image credit: Future)

Apple has not commented publicly on why they're resisting (if that's what it is). Still, considering that Apple CEO Tim Cook (along with other tech leaders like Google CEO Sundar Pichai and Meta CEO and Founder Mark Zuckerberg) was at Trump's inauguration, it might be fair to assume that the lack of a relabeling in Apple Maps is less about resistance than it is about tech timing. Apple might very well be planning to make the change, but, as with so many things, Apple would do it on its own timeline.

What about the children?

For the first time, your choice of map apps might be less about features, utility, and turn-by-turn navigation and instead about which world view they share. If you were a fan of the Gulf of Mexico, you might choose or stick with Apple Maps. If you like splashing America's name as far and wide as possible, you might now be a big fan of Google Maps.

As for those new to mapping, like school-age kids, this split might present a problem, especially if two major maps cannot agree on the name of one of the largest bodies of water in the world. I wonder how teachers will try and likely fail to explain away the confusion.

The great thing about digital maps is that they dispelled the confusion of old maps that could only show so much detail and eventually fell apart in your hands; your phone and its maps app are a world of accurate navigational information in your pocket. But now that trust is a bit fractured. One map might give you accurate directions to the Gulf of Mexico, and the other could confuse you by switching it to the Gulf of America and simply assume you're up to speed.

What a mess.

You might also like

• A recent Android Auto update has messed with Google Maps
• The Galaxy S25's best software feature gets Google Maps support
• Google Maps has been updated with an important design change

• Google adds NotebookLM Plus to its Google One AI Premium subscription
• Students get a 50% discount
• Google One AI Premium gives you access to the best Gemini models

Google is now offering a 50% discount on its Google One AI Premium subscription for students, but the best bit is that it is throwing one of my favorite AI tools, NotebookLM Plus into the bundle for everybody at no extra cost.

As soon as I discovered NotebookLM it was a complete game changer for me. It creates a podcast based on whatever sources you feed it, like PDFs, YouTube videos or articles. The AI-voiced podcast sounds so real it’s hard to believe you’re not listening to two real people discussing whatever subject you’ve fed it.

Speed up learning

Initially I thought it was just a cool gimmick – listening to AI voices that sound indistinguishable from humans is inherently cool – but lately I’ve realized what a boost NotebookLM can be when you’re learning a new subject. NotebookLM’s audio overviews can really speed up your learning process, especially if you are a slow reader, or you just don’t enjoy reading. Instead you can get a podcast containing exactly the information you need to absorb that can be listened to in situations where reading wouldn't be possible. For students, who have to absorb an incredible amount of information very quickly, they can be invaluable.

NotebookLM is free to use, but limited, while NotebookLM Plus, which is now part of the Google One AI Premium subscription, has more features. The differences between NotebookLM and NotebookLM Plus are five times as many audio overviews, notebooks, and sources per notebook. With Plus, you can also customize style and tone, and share your creations.

Google One AI Premium gives you access to the best Gemini features. (Image credit: Google)

As well as NotebookLM Plus, you get all the other benefits of a Google One AI Premium subscription, including Gemini Advanced (which has higher usage limits and access to more LLMs, including Deep Research), Gemini in Gmail, Docs and more. Not to mention a whopping 2 TB of storage.

A standard Google One AI Premium subscription costs $19.99/£18.99/AU$32 With NotebookLM now being added to the bundle at no extra cost. Students in the US will be able to save 50% off an AI Premium plan, making it just $9.99 a month for 12 months, which is an incredible deal.

You may also like

• Google fumbles Gemini Super Bowl ad's cheese statistic
• Yes, Google's new Gemini 2.0 Flash is much better than the old 1.5 model – here's what I found
• If you want to know who will win the AI wars, just watch these two Super Bowl ads from Google and ChatGPT

• The iBasso Nunchaku DAC lets you access two amp modes
• It's USB-powered and houses Raytheon tubes
• It's priced at $299 and comes with a built-in OLED display

What a day to be an obsessive audiophile. I’ve you ever gone to bed drifting off wishing you had an analogue-ready portable amp for some of the best wired headphones, your sweatiest sound dreams have come true. That’s because iBasso Audio has revealed a portable USB-C DAC (digital to analogue converter) with dual amp modes, including a tube amp option, and a fetching OLED display, to really stand out among the best portable DACs.

The Chinese audio specialist recently released the iBasso Nunchaku portable DAC (via Notebookcheck). Now brace yourself for the audiophile-baiting part. The compact little USB DAC has two selectable amps you can switch between and a pair of Cirrus Logic CS43198 DAC chips.

From early glimpses at this little digital amplifier, it screams quality. Not only is its premium-looking body made from aluminum, it also rocks the aforementioned OLED display, with the 0.96-inch screen designed to shield one of the Nunchaku USB’s duo of tempered glass sides.

(Image credit: iBasso)

Initially available in red or silver, this attractive portable amp has a comprehensive feature set. It supports both high-res PCM and DSD51 music stream, and if you really want to delve further into the sonic weeds, the promised signal-to-noise ratio is 130dB. Adjustable filters also allow you to tweak the sound profiles of whatever audio device you pair with the iBasso Nunchaku.

Switching over to its analog amp, the signal-to-noise ratio measures in at around 107dB, which should create enjoyable audio courtesy of the Nunchaku’s (wait for it…) dual Raytheon JAN6118 tubes, which are tested and paired specifically for each unit. iBasso makes clear there is a likelihood of a 'microphonic' effect from tapping or movement on the case transferring through the tube amps (there's a reason they're not commonly used on portable devices!), but these will be reduced as much as possible by its Poron polyurethane shock-mounted tubes.

You've got 4.4mm balanced output or 3.5mm headphones connections to make the most of everything here, so it should be a great bridge between laptops or phones and elite headphones.

Considering it has an OLED screen and has clearly been constructed with premium materials, it’s not surprising this palm-friendly amp is currently priced at $299 (about £240 / AU$475), which seems positively reasonable given the complexity and audio chops.

You might also like

• These luxe wireless headphones stick a tube-amp hi-fi on your head, and offer really hardcore Hi-Res Audio support
• iFI Go Link Max review: one of the best budget USB-C DACs ever
• 3 of the best wireless DACs to upgrade music from your smartphone

• Lee Enterprises confirms suffering a cyberattack
• It was forced to pull parts of its IT infrastructure offline
• Printing of many newspapers were disrupted as a result

American media giant Lee Enterprises has suffered a cyberattack which forced it to pull parts of its IT infrastructure offline - meaning dozens of newspapers and media outlets across the United States could not operate normally.

In a new 10-Q form filed recently with the US Securities and Exchange Commission (SEC) the company said that it suffered a data breach that affected operations.

“On February 3, 2025, the company experienced a technology outage due to a cyber incident affecting certain business applications, resulting in an operational disruption,” it was said in the filing. “The company is actively investigating the incident, implementing recovery measures, and assessing the potential impact on its operations, financial condition, and internal controls.”

No material impact (yet)

Lee Enterprises did not share further details about the attack, nor did it want to discuss it with the media.

Since the company services more than 70 media outlets across the States, many were affected by the attack, TechCrunch reports. Among them are the Post-Dispatch and Casper Star-Tribune. The latter, for example, said “many of Lee’s newspapers initially were not able to build pages and publish, though the company has been working to print and deliver back issues.”

The publication also obtained a copy of a letter the company sent to its employees, in which it was said that a data center hosting applications and services was offline, pulling systems for subscriber services with it.

There is no timeline for recovery just yet, but Lee said in the SEC filing that the attack did not impact the firm materially. “However, the evaluation remains ongoing,” it added.

Although not a rule, generally a company is forced to shut down its IT infrastructure during a ransomware attack. These attacks also often result in data theft and extortion attempts. We expect further details in the coming weeks.

You might also like

• Hundreds of US news sites hacked to send out malware
• We've rounded up the best password managers
• Take a look at our guide to the best authenticator app

• Apple releases update for iOS and iPadOS devices with security patch
• It claims the update fixes a bug disabling USB Restricted Mode
• The bug was being abused in the wild, Apple says

Apple has released a new patch for iOS and iPadOS devices to fix a recently discovered flaw - which normally would be nothing extraordinary, had Apple not described the patched vulnerability quite as dramatically as it did.

In a security advisory, the company said it was releasing iOS 18.3.1 and iPadOS 18.3.1 to address CVE-2025-24200, a flaw plaguing many of its iPhones and iPads which could allow a malicious actor to run a “physical attack” that disables USB Restricted Mode on a locked device.

USB Restricted Mode is a security feature that prevents data transfer through the Lightning (or USB-C) port when the device has been locked for more than one hour. This helps protect against hacking tools that try to bypass passcodes or extract data via USB connections.

Breaking into locked iPhones

Apple said it fixed the issue with improved state management, but added: “Apple is aware of a report that this issue may have been exploited in an extremely sophisticated attack against specific targeted individuals.”

The wording here, although not specific, suggests the vulnerability might have been used by law enforcement and other state-sponsored agencies to unlock iPhones from individuals of high interest.

Apple has a long history of conflict with the US government. While the latter demanded, on a few occasions, that Apple hand over access to iPhones seized from alleged terrorists and other criminals, Apple vehemently declined, arguing that such a move would undermine the privacy of all users and thus ruin the brand itself.

As a result, the US government hired third-party cybersecurity agencies that claimed they had working methods of breaking into locked iPhones. As TechCrunch reported recently, Amnesty International documented a series of attacks by Serbian authorities where they used Cellebrite, an Israeli digital intelligence company known for its phone forensic tools allegedly used to extract data from locked and encrypted smartphones, to unlock the phones of activists and journalists in the country, and then install malware on them.

Via TechCrunch

You might also like

• United Healthcare data breach may have affected 190 million Americans
• We've rounded up the best password managers
• Take a look at our guide to the best authenticator app

• An upcoming Google Pixel update could bring a new way to organize chats from your favorite contacts
• The new Besties app supposedly allows you to combine incoming communications from various apps
• References to WhatsApp have been found in code, and we expect Google Messages will be compatible by default

Google is releasing a new app exclusively for Pixel phones that will allow users to bring together messages, calls, and group chats from various messaging services.

The new Besties app will allow users to centralize messages from chosen contacts across Google Messages, WhatsApp, and presumably other messaging and calling apps.

An APK teardown by Android Authority revealed the Besties app in code from a supposedly upcoming update – Besties appears to be maintaining a database within Android, which WhatsApp has been shown to be able to access.

In fact, it seems that WhatsApp is only able to access Besties due to a special permission given by Google, which suggests a degree of collaboration between the two.

It tracks that other messaging services will be able to access the database too – though there’s no way to say for certain, I’d like to see Discord, Microsoft Teams, and other Meta services like Facebook Messenger and Instagram included in the new Besties app.

One app to rule them all

The new Besties app could be a new reason to pick up a Google Pixel 9 over an iPhone 16 or Samsung Galaxy S25 (Image credit: Future)

If Besties makes it to release, it’ll mark another step in Google’s ongoing work to make communication easier across platforms and services, which also includes its championing of RCS texting between Android and iPhone.

In theory, there’d be nothing stopping a user from adding all of their contacts to the Besties app and using it as an aggregator for all their mobile communications.

This doesn’t seem to be the intended use case, though, and I imagine most users will be a touch more selective with who gets to added.

I’m personally a fan of Google’s Pixel-exclusive apps, which add a sense of flair and exclusivity to the best pixel phones.

In previous years, Google Pixel devices were considered to run a “clean” or “stock” version of Android, which I’m sure appealed to minimalist users, but useful, specific apps and features like Besties, Screenshots, and Now Playing give the Google Pixel 9 series and its predecessors a certain appeal against rivals like the iPhone 16 or Samsung Galaxy S25.

Anyway, enough of what I think – What do you make of this rumored Google messaging tool? Would you use the Besties app? Let us know in the comments section below.

You might also like

• I tried the Oppo Find N5, the world’s thinnest folding phone – and the Galaxy Z Fold 6 doesn't even come close
• iOS 18.4 could be the biggest iPhone upgrade ever – here’s why
• The iPhone SE 4 could land this week, and case photos have already emerged

• Sid Meier's Civilization 7 dataminers have discovered a fourth Age that could soon be added to the game
• The unannounced Atomic Age would join Antiquity, Exploration, and Modern
• Firaxis has teased its future plans, but didn't confirm the release of a fourth Age

Sid Meier's Civilization 7 dataminers have uncovered a fourth Age that could see the game's Ages system expanded in a future update or expansion.

As spotted by IGN, a user over on the Civilization subreddit shared evidence that the Atomic Age will be added to the game to join Civ 7's current three Ages: Antiquity, Exploration, and Modern.

Right now, the game is divided into these three Ages that players can progress through, then the game will end just before the Cold War. Redditor 'ManByTheRiver11' said the game files featured something called the "age-atomic-shell", suggesting a fourth Age is on the way.

"So a person in our country's community found this file inside the Civilization 7 folder called 'Asset cloud.env'. And there were a lot of interesting data inside," the Redditor explained.

Alongside the new Age, they also listed names of leaders who could soon be introduced, like Edward Teach, aka Blackbeard, the 16th-century Moroccan privateer Sayyida al-Hurra, and Maori Women's League President, Whina Cooper.

There were also suggestions of new civilizations as well, including Iceland, Ottoman, Pirate Republic, Maori, and Tonga.

A fourth Age, alongside these new features, hasn't been announced by Firaxis just yet, however, the developer did tease its future plans in a recent interview with IGN, but couldn't offer any confirmation of the recent leaks.

"You can imagine the possibilities with this, honestly," said Civilization 7 executive producer Dennis Shirk. "The way that the design team set it up so that each age is chockfull of systems, visuals, units, Civs, all specific to that age, and what you could do with that and where you could take it… we can't talk about the specifics. We can just talk about it in generalities. We're excited for where this is going to go."

You might also like...

• Best PC games 2025: the must-play titles you don’t want to miss
• Marvel Rivals Season 1 .5 will add Human Torch and The Thing later this month as NetEase walks back its plans for a mid-season rank reset
• Grand Theft Auto 6's release could mean an uptick in PS5 and Xbox Series X sales, according to Take-Two boss

• A gigawatt nuclear energy data center could open soon in France
• “Power is the number one consideration” for AI
• Models need more energy for training than ever

French President Emmanuel Macron has announced €109 billion in private investments to improve French AI infrastructure.

Key to the investment is one gigawatt of nuclear energy power, which will be destined for the high-performance computers and data centers needed to power emerging AI tools.

Announced as part of the AI Action Summit in Paris, the power will add to France’s already extensive nuclear infrastructure, which consists of 57 reactors across 18 different plants.

France will use nuclear energy to power AI

Around a third of the country’s entire energy consumption comes from nuclear energy, and its infrastructure is so extensive that it exported surplus energy to other countries last year. Using nuclear energy to power its AI data centers seems like the natural progression.

Macron added: “Plug, baby, plug” – a reference to US President Donald Trump’s “Drill, baby, drill” plan to expand oil drilling operations across the States.

Josh Parker, Nvidia’s senior director of corporate sustainability, said at the Summit: “Power is the number one consideration for getting access to AI and bringing our systems online” (via WSJ).

The first wave of the project will consist of 250 megawatts of AI-destined nuclear energy by the end of 2026. FluidStack, the company behind the project, aims to start work in the third quarter of 2024. It’s expecting to get most of its high-performance chips from Nvidia – currently the world’s second-most valuable company with a $3.27 trillion valuation.

Energy demands are only rising, too. Separate WSJ reporting reveals that today’s leading AI models used 30 megawatts of energy for training. That could rise to five gigawatts by the end of the decade.

The latest investments add France to the race of countries and companies bidding big on nuclear and other green energy sources to power future data centers.

You might also like

• We’ve listed the best AI writers you can use today
• Access high-performance chips with the best cloud computing services
• Amazon says it is now the top corporate purchaser of renewable energy in Europe

• There's an upgrade to the 70mai Omni dash cam, delivering 4K / 60p video support
• The main camera can rotate 360-degrees for road or cabin views
• It can be paired with a 1080p rear camera for dual channel recording
• Prices start at $279.99 / £268

70mai has unveiled the Omni 4K dash cam, a next-gen upgrade of the world's first 360-degree dash cam, and it delivers a host of meaningful improvements. Chief among them is 4K / 60p video support – that's a big jump from the original Omni's 1080p recording.

There's also a new 1080p rear camera, meaning support for dual-channel recording, although with both cameras in action the front rotating camera's recording capabilities drop to 4K / 30p.

We loved the original model and awarded it 4-stars in our 70mai Omni Dash cam review. That model currently sits in our best dash cams guide, although we fully expect it to be replaced by this latest 4K upgrade when our in-depth review is complete, coming soon.

Price-wise, the 70mai Omni 4K Dash Cam starts at $279.99 / £268 for the front camera alone, while a bundle that also includes the rear cam, hard wiring kit and 256GB micro SD memory card will cost $406 / £389. Sales are set for March, with a specific date TBC. We'll update this page with Australia pricing and availability when we get it.

The main camera can rotate 360-degrees, enabling you to choose between road or cabin recording. It's not a twin camera that sees road and cabin simultaneously. (Image credit: 70mai) A pricier, comprehensive dash cam upgrade

That boost in video resolution and frame rates is the headline new feature in 70mai's new rotating dash cam. At the time of writing, the Viofo A329 is the only other dash cam to offer 4K / 60p video for highly detailed and smooth recording, effectively doubling your chances of clear capture should an incident occur, and we called that dash cam a 'premium performer in our 5-star A329 review back in October. We expect similar performance from the 70mai Omni 4K Dash Cam.

You can now also hard wire it with a 1080p rear camera for dual channel recording, meaning your front and rear views are both covered, although that's a pricier package. The main camera's 360-degree capabilities are its party piece, although it's important to note that it's a single camera that can rotate 360-degrees for road or cabin coverage, not both at the same time. You have to choose whether you want your journey covered, or your in-car karaoke sessions, now with a slightly reduced 130-degree field-of-view to the original's 140-degree FOV!

Still, the rotating camera has some neat features. When in parking mode, it can track nearby objects as they move around the car, such as a person that's taken an interest in your vehicle, and any potential bumps from other vehicles, powered by 70mai's new AI Motion Detection 2.0.

70mai has swapped out the Omnivision OV02K sensor of the original model for the popular Sony Starvis 2 sensor. It's not clear how the sensors compare, though I did a little research to discover they share the same 2.9 µm pixel size and potentially similar low light performance. 70mai has its own Night Owl vision feature to improve night-time clarity, reducing noise and glare, plus Lumi Vision tech for visibility in near-total darkness.

Other shared features include Wi-Fi 6 5GHz and 4G support for mobile app connectivity, streaming and recording downloads, plus built-in GPS and ADAS support for the likes of pedestrian detection, lane departure alerts and forward collision warnings.

Battery life should however be better in the new 4K model thanks to a new supercapacitor unit versus the original 300mAh lith-ion cell. The main camera's screen size has been increased too, from 1.2 to 1.4-inches. You can discover more about the Omni 4K Dash Cam at 70mai's website. All in all, this is a comprehensive if pricier upgrade, and we'll share our full verdict very soon – trust us, this is one to look out for.

You might also like

• The best dash cam 2025: finest car cameras for every budget
• I bought this dash cam for under $20 and its Full HD video quality was anything but
• The world’s best dash cam gets a much cheaper and lighter sibling – meet the Nextbase Piqo
• Viofo’s powerful new dash cam brings several world-first features - including smoother 4K video

• Leaked pictures of Google Pixel 9a cases show the four colors we'll reportedly see the phone in
• These cases also further suggest the Pixel 9a will have a redesigned camera block
• We should know how accurate these images are soon, as the Pixel 9a will reportedly launch on March 19

There’s just over a month until the rumored March 19 launch date of the Google Pixel 9a comes around, and as is so often the case with phones, most of the details have already leaked. However, while we’d previously heard a list of the colors the Google Pixel 9a might be sold in, we hadn’t actually seen most of those... until today.

Android Authority has shared pictures of what it claims are official Google Pixel 9a cases, which it obtained from “a source inside Google”, and the images show cases in four shades, matching the colors we’re expecting the Pixel 9a itself to be sold in.

The shades in question are Porcelain (white), Obsidian (black), Peony (pink), and Iris (purple), so as well as giving us a look at these colors, this leak is further evidence that these are the colors we’ll see.

(Image credit: Android Authority)

It’s worth noting though that according to a previous rumor, only the 128GB model of the Pixel 9a will be sold in all four of these shades, with the 256GB one apparently only set to be available in Obsidian and Iris – at least in some regions.

No more raised cameras

In any case, colors aside there’s not too much to say about these cases, as the design of the Pixel 9a itself has already leaked numerous times.

Still, these cases match up with what we’ve seen, including a flat cut-out for the camera lenses, suggesting that the raised camera bar from other recent Pixel phones will be absent here.

It’s a design that we’re not entirely sold on, but perhaps it will look better in the flesh. We should find out soon if the rumored March 19 launch date is accurate, but don’t be surprised if there’s a final flurry of leaks and rumors ahead of that.

You might also like

• Google Pixel 9a leak suggests it’ll fix a design quirk of its predecessor
• Best Google Pixel phones: these are the Google phones to buy right now
• Google Pixel 9 review: a fantastic phone* (*added fees may apply)

• Marvel Rivals Season 1.5 has a release date and will add Human Torch and The Thing
• NetEase Games has reversed its decision to implement a rank reset mid-season
• The U-turn comes after "a wealth of feedback from the community"

NetEase Games has announced Marvel Rivals Season 1.5 will be released on February 21, 2025.

The studio shared the news via the latest Dev Talk blog post, where it also confirmed that the remaining members of The Fantastic Four, Human Torch and The Thing, will be added to the roster in the mid-season update, joining Mister Fantastic and Invisible Woman who were added in the first half of Season 1.

No details were shared about the heroes' abilities in this post, but we do know that "some major balance adjustments" will also be introduced, which are "expected to shake up the battlefield during the second half of Season 1."

The post also announced that competitive ranks will be reset with this update, however, following player feedback online, NetEase has now walked back this decision.

"After the release of Dev Talk 10 regarding the seasonal rank adjustment, we received a wealth of feedback from the community," NetEase said.

"A common concern was the pressure associated with having a rank reset every half-season, which has made participating in competitive mode less enjoyable. In light of the community's input, we have decided to make some changes to optimize the rank system."

Thank you for all your passionate discussions and feedback. There wil be NO RANK RESET when the second half of the season begins! Check out the details here! https://t.co/zM4IWsVRw0 #marvelrivals pic.twitter.com/7Cpn3aiwUxFebruary 11, 2025

Instead of a rank reset, players will retain their ranks and scores from the end of the first half of Season 1. To earn new rewards, players will need to complete 10 matches in Competitive mode and meet the relevant conditions by the end of the season.

In addition, when the first half of Season 1 ends, players at Gold rank will receive costume rewards, along with new Crests of Honor, which will recognize players at Grandmaster, Celestial, Eternity, and One Above All (the top 500), each with its own unique design to showcase the players' skill level.

"We strive to make Marvel Rivals the best game it can be, and the community is the driving force behind this mission," NetEase said, adding that more details about Season 1.5 will arrive soon.

You might also like...

• Marvel Rivals codes for January 2025 and how to redeem
• The best multiplayer PC games 2025: top competitive titles to play today
• Grand Theft Auto 6's release could mean an uptick in PS5 and Xbox Series X sales, according to Take-Two boss

Microsoft has ditched an online document that was ostensibly about catering for those wanting to uninstall the Edge browser, but in actual fact, was an attempt to dissuade those thinking of removing the app to stick with it.

We reported on the ‘uninstall document’ last week, which came in the form of a web page that, from search results, looked like it pertained to removing Edge from your PC.

However, when actually navigating to the page, it turned out that the document was all about comparing Edge to Google Chrome, and extolling the virtues of the former.

When this hit the headlines, there were a fair few people left unimpressed by this heavy-handed attempt at persuasion – unsurprisingly – and so now Microsoft has canned the web page, as Windows Latest noticed.

When you try to visit this page now, you’re redirected to Microsoft’s Edge portal, which is still all about promoting the benefits of the browser – but without any pretence that this might help you remove Edge (and no mention of Chrome anywhere, either, it should be noted).

(Image credit: Microsoft) Analysis: A misleading tactic that needed to be canned

As you’re likely aware, Microsoft has a long history of prodding folks to use Edge, dating back to, well, the inception of the browser. The Windows operating system has long played host to various nudges, or outright adverts, pushing Edge, so it’s not exactly a surprise to see Microsoft having tried yet another angle on persuasion.

The misleading nature of the search result, compared to the contents of the web page (when it was still active), was the real problem here – and I’m not surprised Microsoft has quickly dropped this one.

What made it worse was the lure of clicking through to ‘uninstall Edge’ when for most folks (outside of Europe), there’s no official way to completely remove the browser from a Windows 11 installation. That’s because other bits of the operating system rely on code within Edge, or this is Microsoft’s stance on the matter, anyway.

You can delete all the shortcuts and so on to the browser, and ignore it, but you can’t actually remove the app as such. It’ll still be sat there lurking in the background. (Furthermore, I should clarify, you can try to ignore it, but it’ll pop up again and again with some regularity in the form of suggestions in various parts of Windows that you should be using Edge, and to make it your default browser).

One of the more frustrating elements about all this is that Edge is a genuinely good product – it tops our list of the best web browsers, in fact – but with constant attempts to promote it, Microsoft is more likely to drive folks away from the app, not into the fold.

You might also like

• Microsoft Edge is better than Google Chrome, but I can't bring myself to switch
• Has the rush to upgrade to Windows 11 just begun? New stats show a marked uptick as Windows 10 End of Life looms later in 2025
• How to use Microsoft Edge

• Elon Musk and other investors offer to buy OpenAI for $97.4 billion
• OpenAI CEO Sam Altman hits back with $9.74 billion X valuation
• Others have criticized OpenAI’s plans

A group of investors led by Elon Musk has offered $97.4 billion to buy OpenAI’s nonprofit arm - but OpenAI CEO Sam Altman to hit back with a sarcastic bid offer to buy X.

Musk’s lawyer, Mac Toberoff, confirmed he submitted the bid to OpenAI’s board yesterday (via Wall Street Journal), further complicating the relationship between Musk and Altman.

The pressure adds to the already ongoing battle between the two, with Musk accusing OpenAI of betraying its nonprofit mission and partnering up with Microsoft.

Musk offers to buy OpenAI

Altman has set out plans to convert the business into a for-profit operation while splitting a $500 billion investment with the likes of Oracle and Softbank to improve AI infrastructure – something that Musk says neither OpenAI or Project Stargate’s backers can afford to do.

Meta has also complained, in the form of a letter to California’s attorney general, about the for-profit business plans.

“It’s time for OpenAI to return to the open-source, safety-focused force for good it once was,” Musk said in a statement provided by Toberoff.

Altman hastened to share with colleagues on an internal Slack channel: “Our structure ensures that no individual can take control of OpenAI…These are tactics to try and weaken us because we are making great progress.”

The OpenAI CEO also took to X, the Musk-owned social media platform, to post: “no thank you but we will buy twitter for $9.74 billion if you want.” Must replied with “Swindler,” and called his rival “Scam Altman” in another post.

It’s yet to be observed how the negotiations play out, but a possible OpenAI-xAI merger could be on the cards if the bid succeeds – an event that Altman deems highly unlikely.

You might also like

• We’ve listed the best AI tools and best AI writers
• Check out our roundup of the best social media management tools
• OpenAI signs deal with Oracle, Softbank for $500bn expansion across US

• A new study from the BBC says AI chatbots are unable to accurately summarize news
• The study asked ChatGPT, Gemini, Copilot, and Perplexity to summarize BBC news articles
• 51% of responses had 'significant issues' and 19% introduced factual errors

A new study from the BBC has found that four of the world's most popular AI chatbots including ChatGPT are inaccurately summarizing news stories.

The BBC asked ChatGPT, Copilot, Gemini, and Perplexity to summarize 100 news stories from the news outlet and then rated each answer to determine just how accurate the AI responses were.

The study found that "51% of all AI answers to questions about the news were judged to have significant issues of some form." and "19% of AI answers which cited BBC content introduced factual errors, such as incorrect factual statements, numbers and dates."

The study showcases multiple examples of inaccuracies that showcased differing information to the news it was summarizing. The examples note that "Gemini incorrectly said the NHS did not recommend vaping as an aid to quit smoking" and "ChatGPT and Copilot said Rishi Sunak and Nicola Sturgeon were still in office even after they had left."

Inaccuracies aside, there's another crucial finding. The report found that AI "struggled to differentiate between opinion and fact, editorialised, and often failed to include essential context."

While these results are unsurprising considering how often we see issues with news summarization tools at the moment, including Apple Intelligence's mix-ups that have led Apple to temporarily remove the feature in iOS 18.3, it's a good reminder not to believe everything you read from AI.

Are you surprised?

From the study, the BBC concludes that "Microsoft's Copilot and Google's Gemini had more significant issues than OpenAI's ChatGPT and Perplexity,"

While this research doesn't necessarily give us much more info, it validates the skepticism towards AI summary tools and emphasizes just how important it is to take information from AI chatbots with a pinch of salt. AI is developing rapidly and large language models (LLMs) are released almost weekly at the moment, so it's to be expected that mistakes will happen. That said, from my personal testing I've found inaccuracies and hallucinations to be less frequent now in software like ChatGPT than it was just a few months ago.

Sam Altman said in a blog post yesterday that AI is progressing faster than Moore’s law and that means we'll continue to see constant improvements to software and how it interacts with the world around it. For now, however, it's probably best not to trust AI for your daily news, and if it's tech-based you may as well stick with TechRadar instead.

You may also like

• I tried ChatGPT Search and now I might never Google again
• If you want to know who will win the AI wars, just watch these two Super Bowl ads from Google and ChatGPT
• OpenAI's Deep Research smashes record for the world's hardest AI exam