TechRadar News

• All Gemini app users can now upload documents for analysis
• Documents only for now, but this may soon change
• Google adds Deep Research to Workspace, but only on certain tiers

Taking your workload to the next level could soon get even more intelligent thanks to a new update from Google and its Gemini AI platform.

The company has announced all Gemini users will now be able to upload documents to the platform and receive AI-powered improvements and customization options.

But that's not all - it also revealed its Gemini Deep Research platform is arriving on certain Google Workspace tiers, giving users deeper and smarter insights to complex topics.

Gemini document upload

In a post on X, Google Gemini outlined what users can expect from the new feature. To upload a file, users should tap the plus sign when asking Gemini a question, and select "files". Anyone signed up will be able to upload multiple files, including Google Docs, PDFs, and Microsoft Word word processor documents, whether stored on Google Drive or on your device.

Once uploaded, users will be able to benefit from a number of features, including getting summaries, giving personalized feedback, and gaining actionable insights, all of which Google says can streamline your workflows and save you time.

The feature is limited to documents for now, however users with a Gemini Advanced subscription can still upload spreadsheets or code files.

(Image credit: Google Workspace)

Elsewhere, following a wider consumer rollout earlier in February 2025, Google has made its Gemini Deep Research service available on select Google Workspace tiers.

Users with Gemini Advanced, which is available for Google Workspace Business Standard, Business Plus, Enterprise Standard, and Enterprise Plus tiers, will be able to access the service, which offers a range of analysis and recommendation options.

"Deep Research explores complex topics on your behalf and presents its findings in a comprehensive, easy-to-read report," a Google Workspace Updates blog post read, highlighting how the service, "brings hours of research to your fingertips in mere minutes."

This includes tasks such as industry research, helping understand emerging trends to inform market analysis, competitive research around other firms, customer research on prospective leads, and helping educators with grant writing, lesson planning, class projects, or presentations.

"Thorough research can take a lot of time between planning, searching, browsing, analyzing, and reporting. It can be tedious work and often difficult to even know where to start. Deep Research can save hours of work by browsing the web on your behalf, analyzing information in real-time, and developing comprehensive research reports in minutes to get you up to speed on just about anything."

Users will also be able to choose from several models depending on what version of the Gemini app you have, with 1.5 Pro with Deep Research now available alongside the existing 2.0 Flash, as well as newly-added 2.0 Pro Flash Thinking Experimental, and 2.0 Pro Experimental.

You might also like

• We've rounded up the best document management software around
• Google Meet will now use Gemini to suggest "next steps" after your team meetings
• Google's Gemini AI platform can now help put that final polish on your work emails

• Shuhei Yoshida has discussed PlayStation PC ports in a new interview
• He said that they can help reach a new audience, especially where consoles are not popular
• They also generate money that can be invested in future games

Former Sony Interactive Entertainment president Shuhei Yoshida has discussed the rise of PlayStation PC ports in a new interview.

Speaking on gaming podcast Sacred Symbols (spotted by Push Square), he said that "releasing on PC does many things, it reaches a new audience who do not own consoles - especially in regions where consoles are not as popular."

He went on to highlight the market in China as an important factor: "China is a huge PC game market [...] and China is a growing but very small console market. In order to reach the audience in countries like China then it’s crucial to release on PC."

On top of this, he argued that reaching a new audience may create new long-term PlayStation fans. "The idea is that those people may become fans of a particular franchise, and when a new game in that series comes out, they may be convinced to purchase a PlayStation," he said.

He also claimed that there were simple economic reasons behind it. “It also adds additional income, because porting to PC is way cheaper than creating an original title," he remarked. "So, it’s almost like printing money and that helps us to invest in new titles now that the cost of games has increased"

Right now you can buy (take a deep breath) Until Dawn, God of War Ragnarok, Helldivers 2, Ghost of Tsushima, Marvel's Spider-Man: Miles Morales, Marvel's Spider-Man Remastered, Horizon Forbidden West Complete Edition, The Last of Us Part 1, Returnal, Uncharted: Legacy of Thieves Collection, Sackboy: A Big Adventure, Ratchet & Clank: Rift Apart, God of War, Days Gone, and Horizon Zero Dawn on PC.

That's an awful lot of PS5 games and there's no sign that this trend will be slowing down any time soon, at least if this interview is anything to go by.

You might also like...

• Original Xbox designer says the idea the Series X is 'more powerful is not helpful today'
• Exoborne is on track to be the most approachable extraction shooter on the market, but will that be enough for it to succeed?
• I tested the Lexar NM790 with Heatsink for a month, and it’s easily the best value PS5 SSD right now

• Strava has rolled out upgrades to its platform
• Tweaks include improvements to Leaderboard Integrity and Flyover sharing
• Its AI feature is also leaving beta thanks to positive feedback, with some new features in tow

Strava, the best fitness app for runners and cyclists, has rolled out a series of updates to its platform that bring improvements to its AI-powered coach, as well as leaderboard integrity and Flyover sharing tweaks.

Starting with AI, last year the company introduced Athlete Intelligence, a new AI-powered health coach that gives you insights into your progress. It delivers unique readouts for that activity and analysis based on your performance to give you a more instant sense of progress.

Strava has added virtual run and ride data, as well as power insights and segment analysis to Athlete Intelligence, and, excitingly, the feature is now out of beta. The company says this is based on positive feedback from both runners and riders, who reported more accurate insights with fewer mistakes.

Strava clamps down on leaderboard anomalies

If you've ever seen a Strava leaderboard score that was obviously clocked in a car or on an e-bike, we've got good news. Strava is now using machine learning to automatically analyze activities on Strava. It can identify unrealistic speeds, GPS errors, and other anomalies to stop skewed leaderboard results and remove "outliers and potential rule violations."

Strava says the next step will be to prevent bike rides from appearing on run leaderboards, and ride leaderboards from being plagued by scores achieved using the best electric bikes.

Strava is also upgrading Flyover off-platform sharing, so subscribers can now upload their Flyover directly to Instagram Stories or via a shareable link to other platforms.

Finally, subscribers can now compare their past efforts side-by-side to track improvements and progress. Your progress summary chart will include a comparison mode to contrast recent activities with past time ranges of training.

You may also like

• Strava's new Heatmaps are here to make outdoor exercise safer this winter
• Strava could soon stop working with some of your third-party apps – here’s what you need to know
• Apple Watch dominates Strava running devices for 2024, even beating out Garmin

• A new Nintendo Switch OLED bundle is coming very soon
• It features Super Mario Bros. Wonder and a Nintendo Switch Online subscription
• It's expected to launch in early March, ahead of Mario Day

The Switch 2 might be out this year, but did you think that'd stop Nintendo from releasing yet another Switch OLED bundle? Not likely.

Spotted by Nintendo Life, the new Nintendo Switch OLED bundle packs in a digital copy of Super Mario Bros. Wonder as well as three months' worth of Nintendo Switch Online subscription time. Nothing out of the ordinary for a Switch OLED bundle, then, and it's available to pre-order at Best Buy for $349.99.

It appears that this Nintendo Switch OLED bundle is set to launch on March 1, but the Best Buy store page is currently saying it'll ship as early as February 26. Of course, this may differ depending on which part of the US or Canada you find yourself in.

You might also wonder (hehe) why a new Nintendo Switch OLED bundle is launching, especially as the Nintendo Switch 2 is likely just a few months away? Well, it's not an uncommon practice for console makers to create new bundles to shift leftover stock. Plus, with Nintendo recently passing the 150 million units mark for the entire Switch family, it seems there's plenty of lemon still left to squeeze.

If you've yet to jump aboard the Switch train and don't really feel like waiting for the next console, this is still an excellent bundle to grab. Super Mario Bros. Wonder is easily one of the best Nintendo Switch games. A beautiful and creative side-scroller, its multiplayer support also benefits from the two Joy-Con controllers you're getting in the box.

You might also like...

• 10 years later, Xenoblade Chronicles X Definitive Edition still has one of gaming’s best open worlds, but is it truly definitive?
• Avowed is a captivating journey through a dreamlike fantasy world that I simply can’t get enough of
• Best GBA games - essential GameBoy Advance titles you need to play

There's been a recent flurry of AI tools that can go online and do things on your behalf. OpenAI's Operator is a powerful and expensive AI Agent. At the same time, the more makeshift Browser Use offers a thriftier option for those who want AI to automate their online errands. The latest example drawing attention is Proxy 1.0, created by Convergence AI. Proxy boasted of being able to multitask online and handling everything except final approval for you.

I wanted to try it out, but you only get five sessions a day for free, so I decided to mimic the test I did with Browser Use. After linking it to my Google account, I was presented with the Proxy interface and started my test run. Unlike Browser Use, though, I could submit the prompts quickly as I didn't have to wait for one to finish before sending the next.

Purchase Proxy

(Image credit: Screenshot of Proxy 1.0)

As with Browser Use, I started with a shopping advice request: "Navigate to Amazon, Best Buy, and Walmart and search for 'MacBook Air M2'. Extract the product name, price, and stock availability from the first five results on each site. Compare the prices and identify the lowest one. If discounts or coupons are present, record them. Provide a final summary with the best deal and where to buy it."

Proxy eventually offered a couple of the best options at each outlet and its own reccommendations. It took a little bit longer than Browser Use and certainly slower than doing it manually, but it was nice to have it running without me having to keep an eye on it.

AI travel agent

(Image credit: Screenshot of Proxy 1.0)

I next asked Proxy for help with travel planning with the prompt: "Search for a round-trip flight between New York and London leaving Dec 15, 2025, and returning December 21, 2025. Select the cheapest option and extract details, including price, airline, and departure time.

This was a bit of a mixed bag. While Proxy did find a shockingly cheap flight of about $430, it put it in British Pounds for some reason. Still, the small airport flight with a stop in Iceland at that price is not bad.

Meteorologist AI

(Image credit: Screenshot of Proxy 1.0)

I went back to the always helpful test of finding and explaining the weather forecast and what I should wear by asking Poxy to: “Check the 7-day weather forecast for New York City on weather.com and summarize temperature trends, rain chances, and any severe weather warnings and then suggest how to dress for it.”

The AI did fine pulling up the forecast and extracting details about what to expect. The AI went somewhat bland in its suggestions about what to wear, recommending an umbrella on a rainy day and warm clothing most days to deal with the cold. But, it doesn't have to be exciting to be useful when it's the weather.

Meteor

(Image credit: Screenshot of Proxy 1.0)

On its own, Proxy 1.0 is amazing. It doesn't have to dazzle you to convince you of its usefulness. The AI doesn’t pretend to be human and make conversation. It just gets things done. It's not perfect, and I had to experiment with wording my prompts to get the best answers, but not enough to be overly frustrating. I can see it being very appealing if I were someone's personal assistant or had to juggle a complex home life. For busy people who don’t want to waste time navigating to and through websites, it's valuable.

Browser Use felt more like an open-source science project. It’s powerful, but it requires a fair bit of setup and tinkering, making it more high-maintenance, though you get great results if you're willing to put in the effort. That DIY flexibility is excellent if your interests and skills trend that way, especially if you want a more conversational interaction, but I prefer the streamlined Proxy.

OpenAI’s Operator is supposed to be the ultimate hands-free, natural-language-powered web assistant. But at $200 a month, it's not for anyone who casually plays with AI tools. There are reports that it also frequently asks for help when it hits an obstacle, which sort of defeats the purpose of having an AI do the work for you.

Proxy 1.0 sits somewhere in the middle of both those alternatives. Compared to how you can almost see Browser Use's scaffolding, you get a complete product without breaking the bank like Operator. I also like how it's more proactive in overcoming obstacles rather than immediately asking for help. The pace at which Proxy completes tasks can be a little annoying. I found I could complete the prompts quite a bit faster than Proxy, but the point is with the AI, I didn't have to.

Anything that lets me offload my least favorite online chores is worth keeping around, especially as it's likely to keep getting better. If Convergence AI keeps improving Proxy, I might even start trusting the AI agent with my most sacred digital task: remembering to cancel free trials before I get charged.

You might also like

• OpenAI's first AI Agent is here, and Operator can make a dinner reservation and complete other tasks on the web for you
• I discovered a surprising difference between DeepSeek and ChatGPT search capabilities
• ChatGPT Tasks can start taking over your calendar and remind you to finish your to-do list

• A clinical research organization's dataset has been discovered online
• The documents include Personally Identifiable Information (PII)
• It's not clear whether criminals have accessed the information

A dataset belonging to a clinical research firm has been discovered publicly exposed online without an encryption or password-protection.

Security researcher Jeremiah Fowler discovered the DM Clinical Research database containing 1,674,218 records, totaling 2TB, including names, medical information, phone numbers, email addresses, medications, and health conditions - along with other data which would put anyone exposed at risk of fraud, identity theft, or social engineering attacks.

Although the name of the dataset indicates the details belong to DM Clinical Research, it's not clear if this was owned and managed by them directly or by a third-party - but here’s what we know so far.

Valuable information

It’s unclear how long the database was exposed before the researcher sent a disclosure notice, but it was no longer accessible ‘within hours’ of the notice being sent. There’s a chance that threat actors may have accessed the information, but only an internal forensic audit could determine this.

“Our team is currently reviewing the details of your findings to ensure a swift and comprehensive resolution," DM Clinical Research replied to the disclosure. "Protecting sensitive data is a cornerstone of our organization’s operations, and we are committed to addressing any vulnerabilities in alignment with best practices and applicable laws & regulations”.

Healthcare information is extremely sensitive and highly valuable for threat actors. Because of this, healthcare organizations are being hit hard by cyberattacks - especially by ransomware and data breaches - which is why data protection is so important in industries that hold personal information.

In 2024, a cyberattack led to the compromise of 190 million American, forcing some applications offline and UnitedHealth also suffered a ransomware attack which resulted in customer information leaked onto the dark web - highlighting just how attractive the industry is for criminals.

Serious consequences

This could be really damaging for patients, especially those with serious medical conditions that may come with stigma, like psychiatric conditions, HIV, or cancer. If criminals access your medical information, they can construct social engineering attacks pretending to be a doctor, health insurance company, or medical professional.

“Any public exposure of health-related information could have potentially serious implications. While things like financial data and some PII can change over time, personal health histories do not,” Fowler points out.

For companies, there are steps you can take to protect your data so that your organization is protected. Security breaches can cost an organization millions, not just in direct costs, but in reputational damage for customers and business partners.

To ensure you’re storing customer data safely, encryption software is incredibly important. Businesses have a legal responsibility to protect their customer records, which means un-encrypted datasets could result in legal action and financial loss.

Using real-time threat and intrusion detection can be a vital tool too, like endpoint detection software, which works by scanning for intrusions and suspicious activity, and alerting security admins if anything is found.

After a breach, it’s important for firms to be transparent to mitigate the damage. This will ensure lasting consumer confidence and trust between your organization and its partners.

For individuals affected by a data breach, it's crucial to monitor financial accounts, bank statements, and transactions to look for anything out of place.

Especially important is being on the lookout for social engineering attacks like phishing - with medical information, criminals may pose as trusted professionals or, in the US where healthcare can compromise your financial situation, take advantage of patients who may desperately need money.

Be wary of unexpected communications, any unrecognised emails or phone calls, and don’t open any attachments that aren’t from 100% trusted sources. Make sure you create a strong and secure password, and don’t reuse it, especially for financial and health organizations.

You might also like

• Check out our list of the best firewall software around today
• Top IVF firm says hackers accessed private data during cyber incident
• We've also rounded up the best malware removal software on offer right now

• Shortly after HP acquired Humane AI, Rabbit is launching Android app
• Much like other AI agents the idea is that this Rabbit app is an agent to carry out tasks for you
• It's currently being tested for Android ahead of a teased release

Rabbit is showing off a new AI agent that can automatically carry out tasks for you on Android devices. This might be because Rabbit is is keen to show that it has another trick after Humane's sale to HP renewed focus and questions about the company's Rabbit R1 hardware. Perhaps aware that the hardware approach wasn't selling as well as anticipated, Rabbit has hopped away from hardware to smartphones, where far more people are likely to experiment with AI tools.

Rabbit's Android agent doesn't just remake the same features of so many AI chatbots, either. Instead of summarizing Wikipedia pages, Rabbit's agent can coordinate tasks across different apps based on your prompts. You could ask it to write and send dinner party invitations on WhatsApp, pick a meal from a recipe app, or assemble a grocery list based on that recipe. It can also find and save a YouTube music playlist for when you eat and even download a game from the Play Store for after the meal.

To really show off Rabbit's AI agent, the demonstration put together by the company showed how the agent could find a business idea you saved in Google Keep and write a revenue plan in Google Docs before sharing it with someone in your contact list.

Rabbit hole

This "doing" AI, as opposed to the "knowing" AI, is usually what people imagine when you describe an AI assistant. But, like any new tech, it’s not perfect yet. Rabbit’s demo shows that while the agent can pull off these tasks, sometimes it stumbles. Some actions take longer than expected, formatting can get weird, and every now and then, it makes a choice that raises an eyebrow.

Rabbit's agent isn't generally available yet. Based on the demonstration, once it is, you'll be able to get the app from the Play Store and link it to Android apps once you give permission. You'll then be able to prompt the AI to complete a task for you, and it will run in the background, which is a nice change from some AI tools that demand you leave the app open while it runs.

(Image credit: Rabbit)

Rabbit will have to contend with more competition with this move. A growing wave of AI tools offers to take control of your devices and automate your tasks.

OpenAI has been developing the app-controlling Operator, and Anthropic is testing a feature called computer use for its Claude AI assistant, which allows an AI to interact with desktop software and the internet. There's also third-party Browser Use, which specializes in navigating the web, filling out forms, and completing requests online. Rabbit isn't imitating any of them precisely, focusing on Android devices and apps. Instead, Rabbit’s agent seems geared chiefly toward everyday use compared to OpenAI and Anthropic’s emphasis on professional activities.

This reimagining of what Rabbit can offer is exactly the sort of reinvention many have suggested could keep Rabbit independent. The standalone R1 didn't have much momentum after it came out and often felt like a limited smartphone with an odd feature set. Now, instead of asking people to buy new hardware, Rabbit is meeting them where they already are: on their phones.

You might also like

• With the Humane AI Pin now dead, what does the Rabbit R1 need to do to survive?
• Rabbit denies that the Rabbit R1 is fundamentally just an Android app
• I used the OpenAI Operator rival Browser Use and it's impressive, but takes some technical skill to use

• Apple's new iPhone – the 16e – is the latest member of the iPhone 16 family
• Even with similar looks, there are some missing features
• A single camera sticks out the most, but it also lacks ultra-wideband, mmWave, and a GPU core

While the iPhone 16e is the latest member of the iPhone 16 family, Apple had to rebalance the feature list and hardware inside to reduce the price to $599 / £599 / AU$999. After all, that’s $200 less than the iPhone 16’s starting price in the United States, so some functionality simply had to go.

That’s not to say it isn’t feature-filled, though. The iPhone 16e still gets an A18 processor for swift performance and the ability to run Apple Intelligence. It also pairs with a 6.1-inch Super Retina XDR display, a single camera on the rear, Face ID, and a design that makes it compatible with the iPhone 16, 16 Plus, 16 Pro, and 16 Pro Max.

So what’s missing? Well, aside from losing some connectivity, it’s also a tale of cameras and how you charge the iPhone 16e – let’s dive into it.

One less GPU core than the iPhone 16

(Image credit: Apple / Future)

Considering it’s more affordable than the iPhone 16, this small tweak with the Apple-made A18 processor isn’t surprising. The iPhone 16e still comes with the A18 chip, but instead of a 5-core GPU like the iPhone 16, it only gets 4-cores. It still features a 5-core CPU and 16-core Neural Engine, so it’s likely this slightly less powerful graphical processing unit won’t make much of a difference in day-to-day usage.

Apple likely created this slightly less powerful A18 chip to keep it within the family. Alternatively, it could have used the A17 Pro chip, which has a 6-core CPU, 5-core GPU, and 16-core Neural Engine, as it did with the iPad mini.

Even so, expect the iPhone 16e to act and perform like an iPhone 16 – meaning that iOS 18 and various apps should run without a hitch.

Last-gen Photographic Styles and one lens

(Image credit: Apple / Future)

The first thing concerning the camera is fairly obvious – the iPhone 16e simply has one lens, just like the iPhone SE that came before it. Without the ultrawide you’ll be a little more limited in the types of photos you can capture, and it rules out the ability for the phone to take macro shots.

Still, you’re left with a significant upgrade over the iPhone SE third-generation as the main lens here is a 48-megapixel fusion camera that can shoot at 1x or 2x. You’ll have access to edit your captures in the Photos app with the ability to use AI to ‘clean up’ your photos. However, if you want to add your own flare or take to how the iPhone shoots, you won’t get the latest and greatest Photographic Styles from Apple.

This isn’t a deal-breaker, but you won’t necessarily be able to finesse and tweak these to your liking. The other missing piece of hardware is the Camera Control – it’s not here. Though, with the arrival of the Action Button, you could set this to open the Camera app.

MagSafe is missing

(Image credit: Apple / Future)

One of my favorite parts of the iPhone, since it was introduced with the iPhone 12, is MagSafe – it makes wireless charging much easier and powers an ever-growing accessory ecosystem. I had high hopes that Apple would have tossed the circular magnetic ring onto its next affordable iPhone, but that’s not the case.

The iPhone 16e simply offers wireless charging using the original Qi standard, just like the iPhone SE’s third-generation predecessor. It can also charge at a maximum of 7.5 watts, which is a bit less than MagSafe or Qi-2.

One potential advantage, though, is with no ring and a slimmer charging coil, Apple might have been able to use a bigger battery cell inside. Apple promises up to 26 hours of video playback, which tops the iPhone 16.

No mmWave or ultra-wideband connectivity

Lastly, regarding the significant things missing, the iPhone 16e only supports 5G Sub-6 Ghz. This means this iPhone can’t connect to the mmWave spectrum for potentially faster speeds. The iPhone 16 supports this, but it’s worth noting that Apple is keeping this commonality with the previous iPhone SE.

A larger missing form of connectivity here though is Apple’s second-generation ultra-wideband chip, which enables things like Precision Finding with AirTags or newer AirPods and is also being used by some forthcoming smart home gadgets like door locks. This a strange omission, though it’s likely done to reduce costs and separate the iPhone 16e from the iPhone 16.

While the iPhone 16e supports Bluetooth 5.3 like the iPhone 16, it only rocked Wi-Fi 6 versus Wi-Fi 7. Thread connectivity is also missing here.

Only two colors

(Image credit: Apple)

One of my, and many others', favorite parts of the iPhone 16 lineup was the standard and Plus colors. Simply put, they were delightful, fun, and vibrant – truly adding some flare to the iPhone and allowing you to pick the one that matched your personal style.

For those who wanted to see an Ultramarine or Pink iPhone 16e, though, that's not happening ... at least at launch. The iPhone 16e only comes in black or white, both matte options.

You might also like

• 5 things you need to know about the iPhone 16e
• The iPhone 15 Pro could get one of my favorite Apple Intelligence features soon – and it’s about time
• Apple's careful approach is killing my interest in Apple Intelligence – here's what I want Apple to do

• At $499, is nostalgia worth the price?
• Boots instantly, offers basic file management and supports SD cards
• King Jim finally brings the DM250 to US buyers

In an era dominated by sleek business laptops and multifunctional tablets, the idea of a portable digital typewriter might seem like a step back in time. Some companies, however, are counting on this as their USP.

Take Pomera's DM250 for instance, which, per Liliputing, aims to revive the retro charm of distraction-free writing in an old-school typewriter— albeit missing a mechanical keyboard.

The DM250, produced by Japanese company King Jim, features a 7-inch greyscale transflective TFT LCD display with a resolution of 1024 x 600 pixels. This screen is visible in ambient light but also includes a backlight for use in dim conditions.

A simple yet functional writing machine

Unlike the older Pomera DM30, which featured a foldable keyboard and an E Ink display, the DM250 has a scissor-switch keyboard with a 17mm key pitch and a clamshell design, similar to the Planet Computers Astro Slide 5G.

With dimensions of 263 x 120 x 18mm and a weight of about 620 grams, the DM250 is positioned as a portable word processor rather than a full-fledged business laptop.

It runs on a lightweight Linux distro which allows for near-instant booting when the lid is opened and users can manage files and folders with basic functionality.

This device also includes 1.3GB of internal storage, which is expandable via an SD card slot to up to 32GB.

The Pomera DM250 offers multiple connectivity options, including a USB 2.0 Type-C port for charging and data transfers. It also supports WiFi 4 (802.11b/g/n) and Bluetooth 4.2, enabling wireless file transfers via the Pomera Link app.

Originally launched in Japan in 2022, the DM250 was only available in its native language. However, King Jim has now announced that a US English version will be available for order starting February 18. This version will include an English keyboard layout and a built-in dictionary for spell-checking. The retail price is $499, though early buyers can secure one for $349.

That's an incredibly high price for what this is, even when Japanese models have been spotted running custom Debian installs.

You may also like

• Here are the best mobile workstations around today
• These are the best free writing apps you can get right now
• And these are the best AI writing tools available

• This tiny docking station will fit snug on any home office desk
• The MEMDock G2 is ideal for MacBook users
• Perfect for users with additional monitors

Docking stations are a dime a dozen in 2025, but finding one that ticks all the boxes can often be a challenge. Luckily, the MEMDock G2 does precisely that - and it looks fantastic to boot.

The MEMDock G2 builds on its previous iteration, which was designed primarily as a MagSafe dock for iPhones. Naturally, it’s bigger than its predecessor, but the G2 still boasts an impressively compact frame, making it the perfect companion to a MacBook Pro on a home or office desk.

It’s not limited to laptops, though, and is compatible with desktops and even gaming consoles, making it a versatile option for professionals and casual users alike.

Under the hood of the MEMDock G2

Complete with a customizable mechanical keypad - which includes six programmable keys that can be tuned to your heart’s content - the G2 also includes a single control knob.

Users also have the option to assign specific keys to particular functions or macros through its complementary app. The docking station features seven ports in total. These include a 100W power delivery (PD) charging port, allowing users to charge a range of devices simultaneously, alongside two USB 3.0 and one USB-C 3.1 port offering speeds up to 10Gbps to streamline file transfers.

From a productivity perspective, however, is where the MEMDock G2 is a gamechanger for users. The docking station includes an HDMI output, which supports a 4K monitor up to 60Hz, making it ideal for those with an external display or mixing and matching screens.

Given the dock has great power supply options, this also means fewer cables littering your home office workstation alongside any additional monitors.

In a hands-on review by Yanko Design, the publication noted that setting up the docking station is surprisingly easy and is very much a plug-in-and-play device.

There are certain downsides, though. While Windows users have complete access to key customization capabilities, macOS doesn’t currently support custom key programming features. MEMDock has hinted at future updates to expand these capabilities, however.

You might also like

• This docking station can drive up to 100 million pixels across three 8K monitors, and even M4 macs can benefit from it
• This mini PC has a rare expansion connector that no rival offers — and could be useful for some great DIY projects
• We've also rounded up the best business laptops around

• With up to 600W power draw, the ROG XG Mobile is fast, but power-hungry
• Creative professionals are catered for —unless they’re on Mac
• Asus trims 18.2% off the size of this desktop-class powerhouse

Asus has introduced its latest external GPU, the ROG XG Mobile, which comes equipped with either an NVIDIA GeForce RTX 5070 Ti or the high-end RTX 5090.

Per Videocardz, the RTX 5090 model features 10,492 CUDA cores and 24GB of GDDR7 memory, making it one of the most powerful eGPU solutions available.

This device also features an updated design with a new hinge mechanism and a built-in stand, reducing its size by 18.2% while maintaining a compact, portable appearance — somewhat resembling a high-performance heater — at under 1kg.

A creative powerhouse — unless you use a Mac

The ROG XG Mobile comes with Thunderbolt 5 connectivity supporting transfer speeds of up to 80Gbps. However, it remains backwards compatible with Thunderbolt 4 and USB4.

Additionally, it serves as an expansion laptop docking station, offering multiple connectivity options, including USB Type-A ports, an SD card reader, HDMI 2.1, DisplayPort 2.1, and a 5G Ethernet port capable of speeds up to 5000Mbps.

Cooling is a major focus of the new ROG XG Mobile; Asus has integrated a vapor chamber and dust filter, increasing the cooling surface by 54% compared to previous models.

Its redesigned rear-side vents contribute to quieter operation, reducing noise levels by 3dB. The device also includes an integrated 350W power supply, but with the RTX 5090’s high power demands, total system consumption can reach up to 600W under load.

Asus has priced the RTX 5070 Ti version at $1,999, with the device set to launch on February 25, coinciding with NVIDIA’s official preorders for the RTX 50-series laptop GPUs. While the price for the RTX 5090 model has not been confirmed, it's expected to be significantly higher.

This eGPU is a compelling option for professionals using business laptops who need high-end GPU power in a portable setup. However, for Mac users, the news isn’t as promising. Apple Silicon chips, from the M1 to the latest M4, do not support eGPUs, leaving many creatives without an external GPU option.

You may also like

• Seagate smashes largest HDD world record with 36TB hard drive and reveals a 60TB model is coming
• Nvidia unveils new GeForce RTX 5090, RTX 5080, RTX 5070 Ti, and RTX 5070 graphics cards at CES 2025
• Check out the best 3D modeling software for 3D printing and more

• Google researchers are warning of an ongoing phishing campaign
• It distributes QR codes which grant the attackers access to people's Signal accounts
• The targets are mostly military personnel, experts warn

Russian state-sponsored threat actors have been increasingly targeting Signal Messenger users, with QR code-powered phishing attacks, malware, and more, experts have warned

A report from Google’s Threat Intelligence Group (GTIG) notes Signal’s use among military personnel, politicians, journalists, activists, and other high-risk groups has lately become widespread, triggering Russian state-sponsored threat actors’ increasing interest, particularly since the beginning of the Russo-Ukrainian war.

As a result, different threat actors (most notably APT44 and UNC5792) have been trying to abuse the “linked devices” feature in the attack. Linked devices allows users to connect multiple devices, such as laptops, tablets, and mobile devices, to the same account. To simplify the login process, users can scan a QR code from a device that’s already logged in, instead of typing in a password, or registering a new service.

QR codes

That being said, cybercriminals have started sending out phishing emails with invites to fake groups, different security alerts, and similar, which also carry a QR code. If the victim scans it, the attacker’s device gets logged into their account, getting access to contacts, messages, and more.

Since the phishing email doesn’t carry a malicious link, or attachment, that can be scanned by email security solutions, these emails often make it past filters and into people’s inboxes.

Beyond phishing, Russian and Belarusian threat groups are also using malware and specialized tools to exfiltrate Signal messages directly from compromised Android and Windows devices.

These efforts include scripts like WAVESIGN, which periodically extracts messages from Signal’s database, and Infamous Chisel, a known Android malware variant. Other actors, such as Turla and UNC1151, have leveraged PowerShell and command-line utilities to steal stored Signal messages from compromised computers, as well.

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

• OnePlus is looking for FDA approval for its Sleep Apnea detection feature
• The OnePlus Watch 3 launched without FDA certification for its ECG feature
• Without FDA approval, these features are unusable in the US

OnePlus is hoping the OnePlus Watch 3 could get FDA approval for its new sleep apnea detection feature, bringing it one step closer to rivaling the best Apple Watches.

We’ve been testing the OnePlus Watch 3 all this week, and it’s a powerhouse of a watch. With five days (120 hours) of battery life, sophisticated health tracking, and all Wear OS 5’s smarts and third-party apps at its metaphorical fingertips, it’s looking likely to end up on our best Android watches list.

However, one area where it cannot compete with Apple’s watches is its features’ approval by the US Food and Drug Administration (FDA). With FDA certification seen as the gold standard for health-based features, the OnePlus Watch 3’s ECG feature failed to receive this certification in time for launch, leaving ithe feature unusable for potential US users.

Not to be deterred, OnePlus is seeking FDA regulatory approval for a different feature, sleep apnea detection, first identified by Wareable during an interview with OnePlus' Dr. Leo Zhang, Head of Research & Development. Sleep apnea is one of the features scanned for as part of the OnePlus Watch 3’s 60-second Health Check-In feature, alongside snoring detection, oxygen saturation, and other night-time metrics.

During the Apple Watch Series 10 launch in September 2024, Apple made much out of its new Breathing Disturbances feature, which notifies users that they may be experiencing signs of sleep apnea and recommends they check in with a medical professional. The feature was FDA-approved days after the device’s launch.

OnePlus has a significant advantage over Apple regarding sleep tracking, and that’s battery life. Compared to the Apple Watch’s meager 18 hours, the OnePlus’ 120-hour behemoth can go from day to night with no charge in between. Apple gets fast charging, but if you forget to charge it, you’re still likely left with a dead watch in the morning.

Adding an FDA-approved sleep apnea detection feature could be a major kudos for OnePlus – although given the gulf between Android and the best watches for iPhone, it’s unlikely this will be the deciding factor for users looking to switch operating systems.

You might also like...

• I gave up my Garmin for an old-school digital Casio watch for a week: here's what happened to my running
• Your smartwatch could help predict when you're about to get depressed, according to research
• Garmin's latest update offers a to-do list task manager app, in a swerve towards traditional smartwatch territory

• 1More's new ANC headphones cost just $50 / £60
• Hi-Res USB-C Audio, LDAC, and spatial audio
• A 90 hour battery life (60 hours with ANC on)

Audio brand 1More has now released a new entry in its budget ANC headphone line, and you’re getting a lot for the price of just $59 / £59 (about AU$119) – they're only $50 in the US for the launch (via DigitalTrends).

The new SonoFlow SE HQ31 over-ear headphones are a new iteration on last year’s SonoFlow SE model, keeping the latter’s low price tag while packing in plenty of more premium capabilities.

This time, 1More has borrowed some features from its higher-end SonoFlow Pro model, adding support for Hi-Res and spatial audio, making these headphones a great choice for those who want high tech on a low budget.

• See the 1More SonoFlow SE HQ31 at Amazon US
• See the 1More SonoFlow SE HQ31 at Amazon UK

The main improvement seems to be the addition of a wired connection over USB-C – USB-C to 3.5mm cable included – for higher-quality sound, compared to the wireless-only predecessor. But the new SonoFlow SE headphones also appear to almost double the former model’s battery life, from 50 hours to a claimed 90 hours without ANC. That’s four days of continuous use!

Even with ANC turned on, you're promised 60 hours, matching the highest number you get from premium options in our list of the best noise cancelling headphones.

Softer edges, but no more travel case for carrying these headphones around... (Image credit: 1More) Fir Green, or Wine Red?

Other specifications include fast charging, Bluetooth 5.4 support, 40mm drivers, and AI-boosted ANC for reducing 42 decibels’ worth of ambient noise.

The company has good form for ANC specifically, even in affordable headphones: in our 1More SonoFlow review last year, we said they would “should suit most office environments and commutes, cutting back most of the lows and mids you’ll hear throughout the day”, and we’ll report back from our own tests when we get our hands on the new model.

There are some changes to the design, opting for softer, more rounded edges; and while these headphones are still foldable for travel, there’s no longer a hard carry case included in the purchase, so you’ll either have to source your own travel solution or be content with these cans rattling around in your backpack.

The SonoFlow SE HQ31 headphones are available now in four different colorways including Black, Blue, Pink and White.

You might also like

• Stay aware of your surroundings with the best open ear headphones in 2025
• Grado's elite new open-back headphones feature huge 52mm drivers, and a beautiful walnut wood finish
• Beats Powerbeats Pro 2: everything you need to know about Beats' all-new athletic earbuds

• SambaNova runs DeepSeek-R1 at 198 tokens/sec using 16 custom chips
• The SN40L RDU chip is reportedly 3X faster, 5X more efficient than GPUs
• 5X speed boost is promised soon, with 100X capacity by year-end on cloud

Chinese AI upstart DeepSeek has very quickly made a name for itself in 2025, with its R1 large-scale open source language model, built for advanced reasoning tasks, showing performance on par with the industry’s top models, while being more cost-efficient.

SambaNova Systems, an AI startup founded in 2017 by experts from Sun/Oracle and Stanford University, has now announced what it claims is the world’s fastest deployment of the DeepSeek-R1 671B LLM to date.

The company says it has achieved 198 tokens per second, per user, using just 16 custom-built chips, replacing the 40 racks of 320 Nvidia GPUs that would typically be required.

Independently verified

“Powered by the SN40L RDU chip, SambaNova is the fastest platform running DeepSeek,” said Rodrigo Liang, CEO and co-founder of SambaNova. “This will increase to 5X faster than the latest GPU speed on a single rack - and by year-end, we will offer 100X capacity for DeepSeek-R1.”

While Nvidia’s GPUs have traditionally powered large AI workloads, SambaNova argues that its reconfigurable dataflow architecture offers a more efficient solution. The company claims its hardware delivers three times the speed and five times the efficiency of leading GPUs while maintaining the full reasoning power of DeepSeek-R1.

“DeepSeek-R1 is one of the most advanced frontier AI models available, but its full potential has been limited by the inefficiency of GPUs,” said Liang. “That changes today. We’re bringing the next major breakthrough - collapsing inference costs and reducing hardware requirements from 40 racks to just one - to offer DeepSeek-R1 at the fastest speeds, efficiently.”

George Cameron, co-founder of AI evaluating firm Artificial Analysis, said his company had “independently benchmarked SambaNova’s cloud deployment of the full 671 billion parameter DeepSeek-R1 Mixture of Experts model at over 195 output tokens/s, the fastest output speed we have ever measured for DeepSeek-R1. High output speeds are particularly important for reasoning models, as these models use reasoning output tokens to improve the quality of their responses. SambaNova’s high output speeds will support the use of reasoning models in latency-sensitive use cases.”

DeepSeek-R1 671B is now available on SambaNova Cloud, with API access offered to select users. The company is scaling capacity rapidly, and says it hopes to reach 20,000 tokens per second of total rack throughput "in the near future".

(Image credit: Artificial Analysis) You might also like

• Nvidia and AMD trade blows over who is faster on DeepSeek AI benchmarks
• A look at the Nvidia GPU that powers DeepSeek's AI global ambition
• AI phenomenon DeepSeek is officially growing faster than ChatGPT

• GoPro refreshes the Max 360-degree camera with modest hardware updates
• New 360 modes added to GoPro's Quik app, including object tracking
• A big price drop lowers the list price to $349.99 / £349.99 / AU$599.95

GoPro has announced a 360-degree camera, but it's not the Max 2 that we've been waiting years for. No, it's a refreshed Max with a couple of hardware tweaks – and a very welcome price drop.

The new camera has an upgraded mounting system for compatibility with all 1/4-20 tripod mounts, plus GoPro's latest Enduro battery for improved battery life – and that all that's new. However, the 'new' Max comes with a sizable price drop, with a new list price of $349.99 / £349.99 / AU$599.95.

It has the same image-making skills as the original Max, including 5.6K video from its twin lenses with complete 360-degree coverage. However, GoPro has improved its Quik app for the Max with new 360-degree features, including a neat-looking AI-powered object tracking that keeps your selected subject in the center of the action.

We haven't tried the new modes in the app yet, but 360-degree video edits can be a time sink, so any smart features that simplify the process are welcome.

Refreshing a five-year-old camera with minor hardware tweaks, amid the backdrop of a much delayed successor, feels like a puzzling decision on the part of GoPro. So what are we to make of the Max relaunch?

The original Max was launch in 2020 and is long overdue a successor if GoPro is to challenge new market leader Insta360. (Image credit: TechRadar) Will we ever see the GoPro Max 2?

The biggest news here is the price drop compared to the original Max – a drop of around a third. GoPro's hand was forced here, because its dated Max camera previously cost the same as the superior-in-every-way Insta360 X4 – the best 360-degree camera available with powerful features including 8K video.

We're keen to give the improved Quik app a spin – in addition to smart object tracking, it also offers keyframing to set precise views, CameraFX for cinematic camera movements, smooth auto transitions between clips, edits to single clips, and a frame-grab tool for exporting stills from your video clips. Combined, these new skills could provide an Insta360-beating editing experience.

However, refreshing the Max only serves to heighten the question – where is the Max 2? It's been delayed time and again, and we're wondering if it will ever materialize.

I'm still hoping that we'll see GoPro's all-new and improved 360-degree camera in 2025 – and at this point it will need to be a big upgrade, with 8K video and more in order to effectively rival Insta360's X4. The leading action camera maker typically saves its most exciting launches for September, so that could be a realistic scenario – we'll be sure to share any Max 2 updates as and when we get them.

You might also like

• Best 360 camera 2025: the finest choices for capturing every angle
• The leaked DJI Osmo 360 looks like very bad news for the delayed GoPro Max 2
• Forget 8K video, the GoPro Hero 13 Black’s swappable lenses are a masterstroke and the future of action camera design
• GoPro Max 2 hit by further delays – 2025 is the earliest we'll see the 360-degree action cam

• GTA 5 is getting a free PC upgrade in March that adds previously released PS5 and Xbox Series X features
• The update will add GTA Online features, improved graphics, ray-tracing features, and more
• New minimum and recommended specifications have been shared ahead of its release

Rockstar Games has announced that a free PC upgrade for Grand Theft Auto 5 will be released next month that will add PS5, Xbox Series X, and Xbox Series S features.

The studio shared the details in a new blog post confirming that the update will be available on March 4 and have features previously only available in the console versions of GTA Online.

These include the latest vehicles and performance upgrades at Hao's Special Works, animal encounters, GTA+ Membership access, as well as improved graphics, ray-tracing features, faster loading times, DualSense Wireless Controller support, and more.

The upgraded version of the game will also feature the most recent player improvements for GTA Online, which include kernel-based anti-cheat protection and proactive voice chat moderation.

Players who download the free upgrade will also be given the ability to migrate their Story Mode and Online progress.

With the addition of features like ambient occlusion, global illumination, ray-traced shadows and reflections, as well as Nvidia DLSS 3, AMD FSR1, and FSR3 support, the new and improved version of GTA 5 also means PC specifications are changing.

You can take a look at the minimum and recommended specs below to make sure your PC can handle the new upgrade.

Rockstar confirmed that support for the previous version of GTA 5 will continue for players whose hardware doesn't fulfill the minimum requirements.

However, for the original version of GTA Online, players who remain on that previous version will not be able to play in online sessions with those with the upgraded patch.

"PC Players who want to play with each other will need to ensure they’re playing the same version of the game," Rockstar said. "Both versions will be available to everyone who upgrades - and first-time purchasers will get access to both versions as well."

The upcoming patch comes as we eagerly await the release of Grand Theft Auto 6, which is scheduled to be released in Fall 2025 for PS5, Xbox Series X, and Xbox Series S. Unfortunately, much like the GTA 5, a PC version will arrive later; reportedly in early 2026.

You might also like...

• Upcoming games 2025 for console and PC
• Grand Theft Auto 6 will reportedly run at 30FPS on PS5 and Xbox Series X
• Grand Theft Auto 6's release could mean an uptick in PS5 and Xbox Series X sales, according to Take-Two boss

• Dead Boy Detectives star Jayden Revri will reprise his role as Charles Rowland in The Sandman season 2.
• His appearance was confirmed in two now deleted social media posts.
• Dead Boy Detectives was canceled last year, with The Sandman also coming to an end after two seasons.

The Sandman season 2 is set to have a big reveal as a star from the canceled Netflix show Dead Boy Detectives will appear in the final season of the dark fantasy epic.

The hit supernatural series Dead Boy Detectives was canceled last year, which prompted fans to campaign to bring the show back from the dead. It developed a cult fanbase when it first aired on the best streaming service and a big viewership was promised since the show is based on the characters from the DC Comics. Despite this though, it wasn't enough to resurrect it for another season.

Netflix also pulled the plug on The Sandman and won't be back after its upcoming second season. The series is finishing because the story of Dream/Morpheus (Tom Sturridge) is coming to a natural end, with showrunner Allan Heinberg revealing that they only had enough story for one more season. However, the ongoing sexual misconduct accusations against The Sandman creator Neil Gaiman will be another contributing factor for its cancelation.

It's not all bad news for Dead Boy Detectives fans though as What's On Netflix has reported that Jayden Revri, who played Charles Rowland in the best Netflix show will feature in The Sandman season 2.

A dead detective is back, sort of

This news came from a post from The Sandman fan page Sandman News, which noticed that Revri’s agency, Revolution Talent confirmed his appearance in The Sandman series on Instagram. According to What’s On Netflix, the now deleted social media posts reportedly said: "Season 2 sees the conclusion of the supernatural-fantasy series, with Jayden’s series-lead role in Dead Boy Detectives making a special appearance. Expect love, loss, destiny and redemption!”

The Sandman spin-off Dead Boy Detectives is one of the nine canceled Netflix shows in 2024 that you should still watch and follows two teenage ghosts, Charles and Edwin (George Rexstrew), who work alongside a clairvoyant to solve mysteries for their supernatural clients.

This is the closest we'll get to a Dead Boy Detectives season 2, so now that Revri is reprising his beloved role, it's a good opportunity to say a final farewell to the character and fill the Dead Boy Detectives void in your life now that the paranormal series is no more.

You might also like

• ‘The world of Peaky will continue’: Steven Knight confirms the hit Netflix drama Peaky Blinders won't end after its highly-anticipated movie
• Netflix's hit French thriller series Lupin might be coming back, but not as you know it
• Netflix is getting its first-ever English medical procedural series and a historical drama that looks like a Spanish Bridgerton

• Nvidia has supposedly run into performance issues with the chips for its RTX 5070 and 5060
• This has meant these GPUs have been theoretically delayed by a month
• As a result, initial stock levels of the RTX 5070 may be thin on the ground

We’re hearing more rumors about Nvidia’s RTX 5070 and 5060 GPUs being delayed and encountering hiccups with production, as well as suffering from possibly low stock levels at launch (once again).

VideoCardz noticed a post on X from analyst Dan Nystedt, who flagged up a report from the Commercial Times, a Taiwanese news outlet, making various claims about Nvidia running into trouble with the mentioned GPUs.

Nvidia’s latest RTX50 gaming graphics card series faces delays until mid-March at the earliest due to performance issues in the chips, RTX5070 and 5060, that needed to be debugged, media report, while wafer damage from the January 26 Taiwan earthquake is also a key factor in the…February 20, 2025

Apparently, the Blackwell chips used in the RTX 5060 and 5070 (known as GB205 in the latter case, and rumor has it, GB206 for the former) were suffering from performance issues which required last-minute addressing. And the need for that additional honing late in the day has “delayed the mass production schedule” (bear in mind that this is an article translated from Chinese).

Commercial Times further points out that the “Tainan earthquake disrupted TSMC’s wafer production” which is what Nvidia’s Blackwell chips are fashioned from, so this has compounded supply issues, we’re told.

The report asserts that the end result of all this is that the mass production of both the incoming RTX 5070 and 5060 (remember, the latter GPU isn’t officially confirmed by Nvidia yet) has been delayed by around four weeks.

The theory presented is that Nvidia previously had targets of mid-February for the RTX 5070 production lines to be running full tilt and mid-March for the RTX 5060, but that schedule has now been pushed back to mid-March and mid-April respectively.

(Image credit: ShutterStock) Analysis: A theoretical but depressingly believable scenario

What does all this mean in practice? Well, maybe nothing – it’s just a bunch of rumors after all – but the fact is that it does marry up with other speculation suggesting pretty much the same thing: a delay from February/March to March/April. (That came from one of the more reliable sources out there for hardware-related gossip, too).

On top of that, do know for sure that the RTX 5070 has been delayed – Nvidia originally announced that this GPU, and its Ti partner, would both go on sale in February. And while the RTX 5070 Ti is now out as planned, the RTX 5070 vanilla version has indeed been put back to an on-sale date of March 5. (It’s worth noting that the Ti uses a different Blackwell chip, GB203, as opposed to GB205 in the plain 5070 – and so the mentioned chip performance issues wouldn’t apply to the former, as that chip has already been put out there in the RTX 5080).

Furthermore, as you may have noticed, the RTX 5070 Ti has gone on sale today, but stock has vanished pretty much instantly. If you got a GPU, congrats, you were lucky (and if you’re still hunting, check out our guide on where to buy the 5070 Ti).

If production of the RTX 5070 is indeed running behind schedule as rumored – and Nvidia having pushed the GPU back does make things look shaky here – then we can fully expect a lean level of supply, and the same blink-and-it’s-gone stock scenario with the RTX 5070.

It’s all depressingly believable, too. Let me put it this way: do you think when the RTX 5070 becomes available to buy in early March, there’ll be plenty of stock kicking around? No, me neither, although I could, of course, be very wrong (and nothing would please me more, frankly, as I might want to buy one of these graphics cards).

As for the RTX 5060, we don’t know if this GPU is coming soon, as Nvidia hasn’t said anything officially, but the rumor mill reckons it’s due in the near future. If this speculation is correct, though, we’re now looking at mid-April, so the graphics card is still a couple of months away (in theory).

The more positive news brought forth with this report from the Commercial Times is that Nvidia is apparently ‘making every effort’ to speed up production for Blackwell GPUs and get everything back on track, and analysts believe that the lackluster supply will gradually improve going forward.

So hopefully, this won’t turn into a longer-term tale of skewed supply and demand, with seriously hiked pricing, as we’ve seen in the past.

You might also like...

• RX 9070 GPU could theoretically be an RTX 5070 killer, I’m just worried that AMD may not go for Nvidia’s throat with pricing
• Keen to buy a new gaming laptop with an RTX 5000 GPU? Save the date: Nvidia’s announced pre-orders open on February 25
• The end of the Nvidia RTX 4000 series is nigh – with the RTX 4070 expected to sell out in weeks

• Security researchers find high-severity flaw in popular WordPress plugin
• It allowed threat actors to run malicious code remotely
• A patch was released in late January 2025

Jupiter X Core, a popular WordPress plugin with more than 90,000 users worldwide, is vulnerable to a high-severity flaw that allows threat actors to run arbitrary files on the server, essentially giving them the ability to fully take over target websites, experts have warned.

WordPress security researchers Wordfence revealed it was found to be vulnerable to a “Local File Inclusion to Remote Code Execution” flaw, now tracked as CVE-2025-0366. It has a severity score of 8.8/10 (high) and affects all versions up to, and including 4.8.7.

Jupiter X Core is a companion plugin for the Jupiter X WordPress theme, developed by Artbees. It extends the functionality of the theme by adding advanced features, such as custom page-building elements, theme customizer options, and enhanced design controls. The plugin is primarily used by web designers, developers, and business owners.

SVG uploads as the problem

“This makes it possible for authenticated attackers, with Contributor-level access and above, to include and execute arbitrary files on the server, allowing the execution of any PHP code in those files,” Wordfence explained. “This can be used to bypass access controls, obtain sensitive data, or achieve code execution.”

Describing how a theoretical attack might look, Wordfence said that an attacker could create a form that allows SVG uploads, upload the file with malicious content, and then include the SVG file in a post, to run the code. The process makes RCE “relatively easy”, it added.

The bug was first spotted in early January 2025, with Artbees coming back with a patch before the end of the month. That being said, if you’re using Jupiter X Core, you should make sure you’re running at least version 4.8.8.

At press time, the WordPress website shows 46.8% of users running the latest version, meaning that more than 47,000 websites are still vulnerable.

Via Infosecurity Magazine

You might also like

• This top WordPress plugin had a security flaw that could let hackers hijack your site
• We've rounded up the best password managers
• Take a look at our guide to the best authenticator app