• 69% of Brits don't think their bank could detect an AI deepfake impersonating them
• Four in five do at least expect ID verification to get slicker in the next five years
• Most want convenience and are only prepared to wait two extra minutes

More than two in three (69%) Brits worry their bank couldn't distinguish them from an attack designed to impersonate them, according to new Digidentity research.

With rising AI threats targeting businesses and consumers alike, four in five would now expect their bank to have AI deepfake detection tools in place when they're running selfie-based identity checks.

This comes as more and more citizens end up preferring increased security measures like vide-based identity checks – with half (53%) believing that one-time security codes sent by text are no longer sufficient.

Banks aren't fully trusted to spot a deepfake

The fallout from a successful attack can have far-reaching consequences, so it's no surprise UK citizens are scrutinizing their banks. Vicims could face months or years of disputing fraudulent debts or challenging mystery accounts opened in their names, for example.

The report comes after Bank of England governor Andrew Bailey called for social media companies to take greater responsibility for tackling scams.

But there is hope, and citizens feel fairly confident in banks' longer-term strategies. Four in five (79%) expect ID verification processes to become smoother over the next five years despite these systems ultimately having to do more heavy lifting, and with good reason. Three in five (62%) respondents said they'd only be willing to tolerate up to two additional minutes of checks.

However, there are challenges. Banks don't typically run their own verification systems, instead choosing to opt for a third party.

"Banks put their own reputation behind the identity provider they choose," Digidentity UK Country Manager Jonathan Evans wrote. "If that provider lets an imposter through, it is the bank’s relationship with its customer that suffers."

Looking ahead, banks are under the most pressure they ever have been to verify legitimate activity and block malicious attacks, but they're also under equal pressure to deliver experiences that are convenient and quick for the end user despite rising threats.

• SpaceX wants to add 10GW of data center capacity during 2027
• Analysts believe Musk's aggressive construction schedule could become reality
• Colossus 2 is already demonstrating the scale of SpaceX's ambitions

SpaceX claims it could add an extraordinary amount of artificial intelligence computing capacity within the next 18 months, potentially matching Google's cumulative development over decades.

Elon Musk believes his company can install 10 gigawatts of additional data center capacity in 2027, an ambition that has attracted skepticism.

However, analysts believe the proposed expansion is achievable, citing construction progress, internal manufacturing capabilities, and access to dedicated power generation equipment.

SpaceX's construction pace could change the data center race

According to SemiAnalysis, SpaceX could increase its computing infrastructure from approximately 3GW to more than 10GW through developments planned across several locations.

If achieved, that expansion would represent approximately 10GW of additional capacity within 18 months, matching what Google accumulated throughout its entire existence.

The analysts point to the rapid development of Colossus 1 and Colossus 2 as evidence that Musk's ambitious construction schedule remains achievable.

Colossus 2 offers a particularly important example, with its electricity generation capacity reportedly increasing from 0.4GW in January toward 2.4GW by December.

The facility reportedly includes 89 gas turbines capable of generating approximately 2.5GW, alongside 1,377 Tesla Megapack battery systems installed across the property.

However, environmental restrictions will eventually limit the permitted output from those turbines to approximately 1.24GW, creating additional requirements for electricity supply.

The facility is also expected to receive approximately 1.2GW from the electricity grid, although analysts estimate another 300MW remains necessary.

Battery storage could help address that difference by supplying additional electricity when demand fluctuates, alongside possible contributions from external power sources.

SpaceX reportedly has temporary permission from Mississippi authorities to operate additional turbines, provided the company phases them out by July 2027.

Tesla's manufacturing capabilities could give SpaceX an advantage

One factor potentially accelerating construction is SpaceX's relationship with Tesla, which could provide access to manufacturing capabilities and electrical infrastructure components.

SemiAnalysis suggests Tesla may be producing modular mechanical and electrical equipment, allowing installation teams to deploy completed systems without waiting for conventional suppliers.

Such an arrangement could reduce procurement delays affecting competing developers, particularly when specialized power equipment and other essential infrastructure remain difficult to obtain.

The analysts also identified approximately 9GW of potential computing capacity across the facilities included in their assessment of SpaceX's development pipeline.

Musk previously stated that Colossus 2 already operated 550,000 GPUs as of September 25, although the exact distribution across individual buildings remains uncertain.

Other developments, including Gun Range and Project SPAM, also appear to be advancing, with aerial photographs showing substantial construction activity.

Nevertheless, completed buildings do not automatically translate into operational computing capacity, particularly when electrical connections, cooling systems and server installations remain outstanding.

Microsoft CEO Satya Nadella previously explained this problem, saying, “You may actually have a bunch of chips sitting in inventory that I can’t plug in. In fact, that is my problem today.”

SpaceX must also secure sufficient GPUs, memory components and power generation equipment before its ambitious deployment schedule can become reality.

Even with substantial manufacturing resources, regulatory restrictions, construction materials and local opposition could complicate expansion across multiple locations.

Via Tom's Hardware

• AMD guided Q4 2026 revenue to about $17 billion, above Intel's $15 to $16 billion forecast, which would make it the larger company in terms of raw revenue for the first time
• Data center sales nearly doubled to $8.5 billion in Q3 as Instinct MI455X GPUs and Helios racks started shipping
• The lead, interestingly, is in revenue, not volume: Intel still ships about two-thirds of x86 CPUs, and AI accelerators are set to outsell Epyc inside AMD this quarter

AMD has told investors to expect fourth-quarter 2026 revenues of around $17 billion, plus or minus 3%.

If that lofty forecast holds, AMD would bring in more money than Intel in a single quarter for the first time in the two companies' decades-long rivalry.

For context, AMD's outlook arrived with a record third quarter in which revenue rose 45% year over year to $13.4 billion, even as Intel, which reported a week earlier, has guided to between $15 billion and $16 billion for the same three months.

A numbers game in revenue, not volume as data centers do the heavy lifting

Using the midpoint of both companies' estimates and assuming both predictions come true, AMD is essentially set to generate $1.5 billion more in revenue than its traditional chipmaking rival.

Much of this comes from AMD's data center gains: Data center revenue climbed 97% to a record $8.5 billion, or 64% of total sales, up from 47% a year earlier, with the segment's operating margin widening to 34%.

AMD CEO Lisa Su said Q3 was the first quarter AMD shipped Instinct MI455X accelerators and Helios racks for revenue, with Anthropic, Meta, Microsoft, OpenAI and Oracle bringing their first racks online.

Revenue shipments of 2nm Epyc "Venice" processors also began, although, as TechRadar Pro reported at launch, the full Venice stack will not ship until 2027, so there is potentially more uplift in the pipeline assuming demand remains on track.

The picture elsewhere was mixed at best: Client revenue rose 7% to $2.9 billion, gaming fell 40% to $773 million on weaker console chip sales, and embedded grew an impressive 29% to $1.1 billion.

For Q4, CFO Jean Hu guided data center revenue up about 50% sequentially, which Tom's Hardware puts at roughly $12.8 billion, with client and gaming down more than 20% and embedded up by double digits. She added that Q4 will be the first quarter in which AMD's data center AI business, meaning Instinct GPUs and racks, outsells its server CPUs, and by a wide margin.

Intel's numbers, on the other hand, are not essentially a clean baseline, even if AMD comes out ahead regardless: Its outlook accounts for lost output from the October 14 explosion at Fab 28 in Kiryat Gat, Israel, which killed two workers and shut the plant for around two weeks. "Our guidance includes an impact of roughly $300 million to revenue," CFO Dave Zinsner told analysts, according to Calcalist, almost all of it in 10nm server products. Add that back to Intel's high end, and you get $16.3 billion, still below AMD's estimated floor.

The crossover does not seem to have been on the cards so soon, however: Wall Street was looking for about $17.05 billion from Intel, according to Futurum, and around $16.2 billion from AMD, per Opening Bell Daily. As recently as April, consensus estimates had AMD surpassing Intel in revenue in 2027, and AMD's moves seem to have been considerably ahead of analyst estimates, resulting in the stock surging more than 100% YTD despite a steep AI selloff lately.

The move is a revenue milestone, not a volume one; Intel still ships most x86 processors, holding 65.9% of the market in Q2 2026, according to Mercury Research figures reported by HotHardware. AMD's revenue also includes businesses Intel does not run at comparable scale, from Xilinx-derived embedded chips and console silicon to, above all, Instinct accelerators, making for a skewed direct comparison between the two, especially as the latter is reaping AI bonuses at a scale the former cannot currently hope to match.

In fairness, AMD's data center segment has outearned Intel's equivalent group before: TechRadar Pro reported the first such quarter in Q3 2024, when AMD's $3.5 billion edged past Intel's $3.3 billion, but its recent revenue explosion has carried AMD into what seems to be an insurmountable short-term lead. In Q3 2026, it was $8.5 billion, compared with $7.3 billion for Intel's Data Center and AI group. But with GPUs set to overtake Epyc inside AMD this quarter, the overall crossover owes more to a market Intel has barely entered than to the CPU fight that defined the rivalry. Intel's own data center accelerator, the GPU D-series formerly codenamed Crescent Island, is still upcoming.

Other factors

It would also be a mistake to read Intel's forecast as an exodus of customers: Intel's Q3 revenue rose 21%, and its data center group grew 78%, and CEO Lip-Bu Tan said the immediate priority is to "unlock additional supply to meet extraordinary demand for our Xeon family."

Part of Intel's problem is capacity, layered on top of a PC slump that is hitting both firms: AMD is planning for PC unit shipments to fall by a double-digit percentage in 2027.

AMD CEO Lisa Su was candid about the fact that AMD cannot build everything it could sell. "It is true that demand is higher than supply," she told analysts.

The Q4 forecast rests on a roughly 50% sequential jump in data center sales, much of it from a rack platform that shipped for revenue for the first time last quarter.

That ramp has costs: Inventory rose about $3 billion to $11.5 billion, and free cash flow was a relatively measly $554 million. AMD's biggest AI deals also come with customer warrants, which produced a $1.1 billion non-cash charge in Q3 that the company expects to grow. Likewise, revenue is concentrated in a handful of frontier labs and hyperscalers even as circular financing concerns continue to grow.

AMD now expects data center sales to make up more than 80% of revenue in 2027 and server CPU revenue to double. Delivering that depends on partners such as HPE, which became Helios' first major OEM backer and has since said Vultr will deploy the first Helios systems it builds. It also depends on the extra capacity Su has promised for 2027, which TechRadar noted is aimed at AI rather than consumer chips.

If both companies land inside their ranges, AMD will have outearned Intel for the first time. The more telling story is where that money comes from: not a knockout in the old CPU war, but a shift in the industry's revenue toward accelerators and racks that now fill data centers.

• A Hong Kong-led venture plans a 168m, 20,000-seat LED sphere in Hangzhou for ¥8 billion (~$1.2 billion)
• Its 72,000m² exterior screen would be a third larger than the Las Vegas Exosphere, the biggest of its kind, at least on paper
• The project is still at the planning stage, with no construction dates, screen vendor or display specs announced

A developer in eastern China has unveiled plans for a spherical entertainment venue that looks to beat the Las Vegas Sphere on every headline measurement, for less than half the price.

The catch is that, so far, the project amounts to a signing ceremony, a rendering and a set of very ambitious forecasts.

The project, dubbed the Eye of the World in English, was reported by AV Magazine after its public debut on October 30 2026 at a services-sector conference in Hangzhou's Qiantang District.

A bigger sphere on paper

According to City Express, the plan calls for a sphere roughly 168 meters across and 122 meters tall, skinned in about 72,000 square meters of exterior LED, with a 270-degree wraparound screen and 20,000 seats inside.

It would occupy a 56,000-square-meter plot on the Qiantang River in the Jinshazui New Town area, with around 160,000 square meters of neighboring land earmarked for supporting development.

On the spec sheet, the comparison is flattering when one compares it to the Las Vegas Sphere. AV Magazine puts the Las Vegas venue at 157 meters in diameter and 112 meters tall, with a 54,000-square-meter Exosphere and 17,600 seats, which makes the Hangzhou design about 7% wider and 9% taller, with a third, more exterior screen. The seating gap, however, is narrower than it looks, as Las Vegas reaches around 20,000 when its floor is set up for standing crowds.

The headline budget is ¥8 billion (~$1.2 billion), against the $2.3 billion Las Vegas eventually cost, and the project's forecast annual revenue is ¥2.6 billion, or $400 million. At the (relatively harsher) exchange rate implied elsewhere in Chinese coverage, the two figures come closer to $1.1 billion and $360 million.

The venture pairs Hong Kong-based Xinshi International Investment Co. with Vision Eye (Hangzhou) Digital Entertainment Technology Co., and Invidis reports that it is led by Huang Junhao, investment president of the former and general manager of the latter. He says the domestically designed project will show that Chinese applied technology and cultural innovation have reached "world-leading levels," and at the launch he described it as "a spectacle of unimaginable size and unimaginable form."

Interestingly, for an LED project, the most telling detail is what's missing: The invidis report also notes that no vendor or installer has been appointed for either the exterior screen or the interior wraparound display, that design, zoning and infrastructure arrangements are still in progress, and that no construction start or completion date has been given.

Nobody has published a resolution, pixel pitch or playback architecture either, and that matters, because the systems behind a screen like this are as demanding as the panels themselves.

In Las Vegas, TechRadar Pro reported that 150 Nvidia RTX A6000 GPUs drive nearly 750,000 square feet of LED with three layers of 16K video at 60 frames per second, fed by a Hitachi Vantara system of 27 nodes and 4PB of flash storage delivering 400GB/s.

The structure won't be simple either: One analysis notes that Las Vegas used parametric modeling to optimize its geodesic lattice shell and cast-steel nodes, cutting weight by around 40%, and warns that a larger sphere demands even tighter tolerances, where millimeter-level errors could show up as visible seams or patches of uneven brightness. Any claim to be the world's largest, it concludes, depends on the engineering meeting the paper specifications.

An analysis on a Chinese publication, The Paper's government affairs channel, argues the cost gap with Las Vegas rests on China's domestic LED sphere supply chain, which now has a track record of delivery, and that the real difficulty is keeping brightness, color, and seams consistent across tens of thousands of square meters of curved outdoor LED through direct sunlight and temperature swings.

The launch coverage on Sohu flags acoustics as another headache: a concave sphere focuses sound unevenly, and any absorptive treatment has to sit inside the lattice shell without breaking up the screen.

A bloated forecast with questionable demand in tow?

The commercial forecasts look even bolder than the project at hand: The developer expects 9 million to 13 million visitors a year. Spread evenly, that is roughly 24,700 to 35,600 people every day, so even the low end exceeds a sellout of the 20,000-seat bowl 365 days a year.

The Paper's analysis supplies useful context. In 2025, Hangzhou's entire performing arts sector drew about 10.25 million attendees across 19,194 shows and took ¥3.12 billion at the box office.

One venue is forecasting roughly the whole city's 2025 audience and revenue, totaling about 83% of that box office total, though the projected figure presumably also includes advertising, sponsorship, and food and drink. The same analysis also describes The Eye of the World as "still just a drawing, with construction yet to begin."

The owner of the original isn't standing still. Sphere Abu Dhabi, on Yas Island, carries a $1.7 billion construction budget and will match Las Vegas with a capacity of up to 20,000, depending on configuration, a 16K-by-16K interior screen, and 167,000 speakers, with completion due by the end of 2029. A 6,000-seat venue at National Harbor in Maryland is targeting 2030.

Sphere is also lining up the technology behind its screens. Announcing a multi-year partnership with media server specialist Disguise in September, Sphere executive vice president Ryan Dolan tied the deal to "offering these experiences to audiences in new Sphere venues around the world."

TechRadar Pro's June 2026 coverage of China's first Sphere-style domes analyzed a completely different story about going smaller and cheaper: Qianjiang's Tiangong Nova, with 7,500 square meters of exterior LED, and Shanghai's West Bund Starry Sky, budgeted at ¥1.2 billion for around 3,000 visitors at a time.

Hangzhou keeps prices low and scales up, but not everyone thinks the world needs it. Discussing China's smaller domes on an AVNation panel, AVIXA's Jennifer Weaver asked: "Massive replicas of the Vegas sphere all over the place? Probably not." Hangzhou is betting they are wrong; for now, though, it has a rendering, a riverside plot, and a set of forecasts, while the original, by all accounts, remains busy every day of the year as a revenue-generating juggernaut, bolstered, at least in part by a much more affluent audience to boot.

• The FAA will eliminate floppy disks across more than 400 facilities
• America's air traffic modernization program has received $12.5 billion from Congress
• Outdated technology has repeatedly disrupted American air traffic operations

The US Federal Aviation Administration (FAA) is preparing to retire technology that has remained part of American air traffic operations for decades, including floppy disks.

The agency plans to complete the transition by the end of 2028, replacing outdated information systems still operating at more than 400 facilities nationwide.

The FAA had to modernize its technology after repeated technical failures and aging infrastructure raised concerns about the reliability of America's air traffic management network.

FAA finally moves to replace decades-old technology

Reports from Reuters claimed the FAA will introduce the Enterprise Information Display System, consolidating four separate platforms introduced during the 1980s.

The replacement will give controllers access to essential operational information through one interface, reducing their reliance on older equipment and disconnected systems.

These changes will cover weather information, flight alerts and other operational updates that controllers need when coordinating aircraft movements across American airspace.

FAA Administrator Bryan Bedford said the agency wanted to eliminate manual procedures and improve how important information reaches personnel working inside control facilities.

“We’re moving away from outdated, manual processes and replacing technology that once relied on floppy disks with a modern system that allows controllers to instantly share critical flight alerts, restricted airspace schedules and pilot reports,” Bedford said.

Although floppy disks largely disappeared from everyday computing more than 25 years ago, American aviation facilities have continued relying on them.

Their production had already declined substantially by 2010, making the continued dependence on these storage devices particularly unusual for critical aviation infrastructure.

Congress approved $12.5 billion in 2025 for air traffic technology improvements, while Transportation Secretary Sean Duffy wants another $30 billion.

The additional funding would support further modernization as the agency attempts to address several weaknesses affecting its existing operational technology

Aging infrastructure continues to expose weaknesses in American air traffic management

The floppy disk replacement represents only one part of the FAA's modernization challenge, with communication failures also disrupting flight operations.

In an October 23 2026 memo reviewed by Reuters, the agency warned airports about outages associated with damaged fiber optic connections.

One incident occurred on October 13, when a severed cable outside Houston's George Bush Intercontinental Airport interrupted information exchanges between FAA facilities.

The disruption forced authorities to temporarily stop departures, demonstrating how vulnerable essential aviation services remain when individual communication connections fail.

Another incident occurred in September 2026, when separate failures involving Philadelphia and New Jersey disrupted thousands of flights across the Northeast.

The FAA also continues relying on aging Compaq computers and replacement components purchased through eBay to maintain certain operations.

Furthermore, its facilities largely operate independently, using separate analog computing equipment rather than a fully integrated communication environment.

Explaining the situation, Bedford said "every facility is a silo," describing the difficulties created by disconnected operational systems.

The agency is also replacing paper flight strips, which controllers have traditionally used to monitor aircraft movements and maintain operational records.

That transition began in 2022, with 27 of the 49 major airports already abandoning paper records, although completion remains scheduled for 2028.

• Rmatamini's P99S packs a 16GB Tesla V100, Core i7-10850H, 16GB RAM and 512GB SSD into a 5.6L box for $699 on Newegg
• The same 16GB V100 PCIe card costs a whopping $584 new on Amazon, and renewed units can be had for $410
• Nvidia's R580 driver branch is Volta's last; CUDA 13 no longer targets the card, and potential buyers need to contend with a relatively new seller in tow

A new Newegg listing squeezes an Nvidia Tesla V100 with 16GB of HBM2 into a 5.6-liter box - and asks just $699 for the lot.

The Rmatamini P99S "Mini AI Workstation" pairs Nvidia's 2017 data center accelerator with a mobile-centric Intel Core i7-10850H, 16GB of DDR4, a 512GB NVMe SSD, a 700W power supply, and Windows 11 Pro, and ships free from China.

The number that makes this interesting sits on another site: HP's version of the same 16GB V100 PCIe card is selling new on Amazon for $584 through a third-party seller with only two units left in stock; Take that at face value, and the rest of the computer costs as little as $115.

A $584 card in a $699 box?

Amazon's pricing is where one would currently find a brand-new V100 GPU, which has seen a surge in demand due to both the relevance of older silicon and the fact that it uses HBM2, which holds its own in bandwidth compared to newer consumer-grade GPUs.

One could argue that a fairer comparison would involve a potentially refurbished unit that can be found for as low as $410 on Amazon, visible from the same listing as the new SKU. This means the mini PC itself costs either ~$115 or ~$290, depending on how one values the Tesla V100 GPU in a 16GB trim.

The V100, which once commanded a staggering $10,000 or higher price tag, is arguably dated silicon before fading away, but it is drawing attention again, and as much to do with today's market as with the card. Fast memory has become the most expensive part of any AI build.

TechRadar's analysis tied runaway graphics card pricing to AI demand for RAM in January, and a TechSpot analysis, published in February, found 16GB RTX 5060 Ti cards up 22% globally in roughly three months, a trend that has yet to break. Nvidia itself raised the price of the DGX Spark from $3,999 to $4,699 in February, citing memory supply constraints, for a machine whose early reviews had already flagged memory bandwidth as a limitation. Compact Ryzen AI Max+ 395 systems such as the MINIX ER939-AI start at $3,150.

Against that backdrop, the V100's spec sheet looks oddly current. Nvidia's datasheet rates it at 900GB/s of memory bandwidth, about twice the 448GB/s of an RTX 5060 Ti 16GB. Token generation in large language models leans heavily on bandwidth, which is why hobbyists such as Oscar Molnar spent 2026 bolting retired V100 GPUs into desktops.

The best public numbers come from YouTube channel Hardware Haven, whose tests Wccftech reported in May 2026. A modded V100 16GB produced about 130 tokens per second on gpt-oss-20b, compared to roughly 90 for an RX 7800 XT, and 108 tokens per second on Gemma 4 e4b, compared to 76 for an RTX 3060. Capped at 100W, it still managed 95 tokens per second while the whole system drew 170W at the wall.

Plenty of caveats in tow

In practice, 16GB holds mixture-of-experts models such as Gemma 4 26B-A4B entirely on the card, according to Andrew Leech's V100 LLM kit, but denser 30B-class models will not fit, and 16GB of system RAM leaves little room to offload the overflow to the CPU. RmataMini PC also sells a 32GB SKU of the same, with a bump to 32GB of DDR4 RAM, for an eyewatering $1,500, marked down from a sticker price of $2,000.

The biggest long-term caveat is software: Nvidia's driver deprecation schedule names the R580 branch as the last to support Volta, alongside Maxwell and Pascal. ComputingForGeeks notes that CUDA 13 can no longer build code for the V100, and that vLLM now requires a newer compute capability than Volta's 7.0. Ollama y llama.cpp still works on CUDA 12 builds, and community forks such as the 1Cat vLLM build keep the card useful, but the window is closing rather than opening. As Leech's kit notes, Volta's tensor cores support only FP16, so BF16 and newer low-precision formats are off the table.

The host platform is modest but adequate. The Core i7-10850H is a 45W laptop chip from 2020 with six cores, 12 threads, and 16 PCIe 3.0 lanes, matching the V100's PCIe 3.0 interface. Because the V100 has no outputs, displays are driven by the CPU's iGPU.

The seller is also relatively new, and RmataMini PC has a single seller rating; the product has none; warranty questions are referred back to the seller; and returns are for refund only, within 30 days.

For someone who wants a small, ready-made CUDA box with 16GB of 900GB/s memory and has no interest in sourcing adapters, fans, and ducts, $699 is an excellent price in a market where memory keeps getting more expensive.

But a finite software runway from Nvidia and a considerably pricier, but available 32GB SKU mean that users have much to consider before they take the plunge here for anything that could require more memory or a more current CUDA version in the near future.

• Russian hosting provider Beget was subject to a cyberattack on November 1st, 2026
• While 1000s of websites were hit, cloud storage, VPS hosting, and Kubernetes servers were wiped
• The attack came amidst a spate of drone and online attacks targeting Russian tech companies, presumably by Ukraine-based attackers or anti-Putin elements within Russia

A “massive cyberattack” took Russian web hosting company Beget’s entire operation offline, with abnormal activity recorded immediately ahead of the incident. All levels of its infrastructure were affected, with websites, cloud storage facilities, VPS hosting, and even hosting for Kubernetes containers knocked offline.

The impact of the attack took several days for the company’s experts to assess, with estimates that over 1,000 servers need to be physically checked. This indicates either lasting damage from the attack or infrastructure that was already weakened, perhaps by unpatched exploits.

Beget has been the subject of EU sanctions due to it providing hosting services to authorities in the Russian-occupied Ukraine regions of Crimea, Donetsk, and Luhansk.

Recovering data

Responding to news of the attack, Beget issued a defiant statement, boldly stating “We are not giving up and are continuing to look for ways to recover this data.”

"Restoring all services will take a considerable amount of time, as more than 1,000 servers must be manually inspected.”

The organization told Kommersant (translated from Russian) that most data should be recoverable from backups for the majority of its virtual hosting and cloud infrastructure, adding “We continue to work on data recovery, but due to significant volumes, this process will require a long time.”

However, Beget’s response to the attack is hampered by how it organizes and prioritises data, with many customers not using disaster recovery plans.

Anatoly Sazonov, Head of the Department of Compliance with the requirements of the Softline Group of Companies, told Kommersant, “Beget’s mass failure is an example of a systemic risk that arises from a high business dependency on a single provider. If a website, corporate mail, databases and other services are posted on one site, the provider’s problems can paralyze several areas of business work at once.”

EU sanctions

Beget was founded in 2007 and has become the second largest domain registrar in Russia for .ru domains, overseeing 14.7% of the market alongside providing web hosting, cloud computing, and other services.

As is standard, Beget has encouraged customers to change their passwords and remove and update and access keys, credentials, or other sensitive information that may have been uploaded to its cloud servers.

Beget has already had a difficult year, having been targeted by EU sanctions for providing hosting services in Russian-occupied territories in Ukraine. Whether the domain registrar and web hosting company was targeted because of this, or as a more general piece of cybercrime, remains to be seen.

Hosting.com's Starter plan offers a straightforward way to host a website, with enough resources to support small businesses, blogs, and personal projects. It includes hosting for five websites, 15GB of NVMe storage, an AI website builder for WordPress, and daily backups. You also get free SSL certificates, malware protection, and managed migration.

Best For: First-time website owners, bloggers, freelancers, and small businesses looking for affordable hosting with security features and WordPress tools included.

The Deal: Hosting.com has reduced its Starter web hosting plan from $13.99 to $1.49 per month for Black Friday, saving you 89%. The offer requires a 12-month subscription, costing $17.88 upfront instead of $167.88. Coupon BF26WHAL is applied at checkout, with the price returning to $13.99 per month on renewal.

Use code TECHRADAR for an additional 5% off on the already discount price across all 1/2/3/4 year hosting plans,

Best Hosting.com deal

Why we recommend it

Hosting.com's Starter plan provides hosting support for five sites and three staging sites. That gives you room to manage multiple projects and test changes before making them live.

The inclusion of an AI website builder for WordPress is handy if you're starting from scratch, and 15GB of NVMe storage will be sufficient for smaller sites without large media libraries. You also get email and security features, free SSL certificates, and a domain registration offer.

Daily backups, malware detection and removal, and free managed migration are all included, and Hosting.com also includes OpenAPI and MCP server capabilities. With a 99.9% uptime guarantee and support available around the clock, the Starter plan covers all the essentials for running a small website.

Price context & historical value

Hosting.com's Starter plan normally costs $13.99 per month, or $167.88 a year. The Black Friday promotion reduces the annual cost to just $17.88, saving you $150. You'll need to make sure code BF26WHAL is used for the discount, and also use code TECHRADAR for an additional reduction.

This offer locks in the introductory rate for just 12 months. After the first year, the plan renews at full price.

Checkout our round ups of the best Black Friday website deals and the best Black Friday web hosting.

Should you buy it?

✅ Buy Hosting.com's Starter plan if...

If you're launching a blog, portfolio, or small business website and want a hosting package that includes WordPress tools, security features, and daily backups. The ability to host five websites and create three staging sites provides flexibility as projects grow.

❌ Skip Hosting.com's Starter plan if...

If you're planning a resource-intensive website or need substantial storage for large media files. The 15GB allocation may be restrictive, and the considerably higher renewal price makes this less attractive if you're looking for consistently low hosting costs.

• Shop all Hosting web hosting deals

• Google commits $17 billion despite mounting resistance to American data center expansion
• Voters rejected 40 of 41 data center-related ballot measures during the midterms
• Wyoming is getting Google's first data center despite growing political uncertainty

Google is moving forward with a $17 billion investment in Wyoming, despite growing political opposition to large-scale data center developments across America.

There has been intense resistance from local communities, which seem to be concerned about electricity use, water consumption, noise, and other effects.

Recent midterm election results have intensified those concerns, with several outcomes suggesting that opposition to data center expansion is gaining political traction.

Political resistance has not slowed Google's Wyoming investment

Alphabet's Google announced plans on Thursday to establish its first Wyoming data center, working alongside four energy and infrastructure partners.

The development involves Tallgrass Energy, Black Hills Energy, American Electric Power and Related Digital, with construction expected to bring substantial investment into Laramie County.

Wyoming offers a politically favorable environment for the project, having given Donald Trump more than 70% of its presidential votes during the 2024 election.

However, political support at the state level does not necessarily eliminate concerns emerging from communities experiencing the consequences of expanding computing infrastructure.

During this week's midterm elections, candidates opposing data center developments secured victories in important contests across Texas and Georgia.

Only one of 41 data center-related ballot measures decided during the week produced an outcome favorable to the industry's interests.

The developments come as investors reconsider the financial and political risks associated with increasingly expensive AI infrastructure projects across the U.S.

According to recent reports, Microsoft has declined to extend a multibillion-dollar arrangement involving AI infrastructure provider Nscale in Texas.

Energy strategy and community commitments underpin the development

Google's proposed facility will incorporate a planned 702MW fuel-cell installation, supplied by American manufacturer Bloom Energy and contracted through AEP.

The system will generate electricity from natural gas through electrochemical processes, with carbon capture and storage technology intended to limit associated emissions.

Captured carbon dioxide will be transported for permanent underground storage at Tallgrass Energy's nearby Eastern Wyoming Sequestration Hub.

Google said the development would increase its contracted carbon capture capacity to almost 1 gigawatt, building upon an earlier American project announced last fall.

The company also plans to obtain electricity from gas-fired generators operated by Tallgrass Energy and Black Hills Energy.

“We’re committed to partnering with communities to build data centers the right way. We will pay for 100% of the power we use, and when we build data centers, we’ll cover the infrastructure needs directly driven by our growth,” Google said in a blog post.

Water consumption is another important consideration, particularly because large computing facilities can place additional pressure on supplies available to surrounding communities.

Google intends to operate the Wyoming installation using air-cooling systems, avoiding the substantial water requirements associated with conventional evaporative cooling arrangements.

The company has also promised investments in local water replenishment programs, although community engagement has just started and Google has received feedback from over 200 residents.

Local participants want clear information about employment opportunities, economic contributions, educational support, and the cumulative consequences of the data center.

Google expects the development to generate hundreds of millions of dollars in property tax payments for Laramie County and School District 1.

Its community programs will support science and technology education, professional training, entrepreneurship and skilled trades development across Wyoming.

CUBOT KingKong Dura: 30-second review

Most rugged phones chase bigger and bigger batteries, often to the point where they become a viable replacement for a sea anchor.

The oddly named KingKong Dura takes a different route: it comes with a modest 5000mAh battery you can remove, and CUBOT includes a second one in the box.

That is a rare feature in rugged phones, and a useful one. Run one battery flat, swap in the other, and carry on. No power bank is needed, and in this context charging the battery doesn’t involve putting it in the phone.

The replaceable battery enables a design slimmer than the usual rugged brick. It measures 13mm thick and weighs 299.5g. The screen is 6.88 inches, with a 120Hz refresh rate. The phone carries IP68, IP69K and MIL STD 810H protection.

Phones that have replaceable batteries are often expensive, and this one isn’t, so how does that work?

In short, CUBOT pared the specs down to the bare minimum to run Android 16. The Unisoc T620 is an old 12nm chip, with a screen resolution of only 720 x 1640. There is no 5G, and WiFi is only WiFi 5, to mention just a couple of shortcomings.

You get what you pay for, but thankfully you don't pay much.

So this is a working phone for people who need power on tap, or only occasionally need a rugged phone. It isn't for people who want speed, and it isn’t the best rugged phone I’ve seen. This pitch by CUBOT is decidedly niche.

However, it’s easy to be snobbish about the lack of performance in devices like this, and CUBOT clearly see a market for these devices, and a customer who won’t be phased by its lack of grunt.

CUBOT KingKong Dura: price and availability

• How much does it cost? From $190/€210/£143
• When is it out? Available now
• Where can you get it? CUBOT's European store and online retailers

Direct from CUBOT, two SKUs of the KingKong Dura are available. One has 8GB of RAM and 128GB of storage, and another, my review hardware, has 12GB of RAM and 256GB of storage. Both options are available in black or green.

For EU customers, the lower-spec model is €209.99, and the higher one is €259.99, making 50% more RAM and 100% more storage only a 50 Euro bump.

For whatever reason, I couldn’t get the CUBOT website to map these prices into US Dollars or UK Pounds, but American customers might want to consider buying it via an online retailer like Amazon.com. In the USA via Amazon, I found only $189.99 for the 8GB+128GB model.

It’s not available on Amazon UK yet, so you can buy it directly from CUBOT or use AliExpress. On that site, I found it for £142.01 and £206.15 with free shipping to the UK, but delivery from China was about 10 days.

That is a low price for a rugged phone running Android 16, and it undercuts every replaceable-battery design I’ve seen by some margin.

With both batteries included, the Dura is a bargain for its niche.

• Value score: 4/5

CUBOT KingKong Dura: Specs

CUBOT KingKong Dura: Design

• Slim by rugged standards
• Removable battery
• By-the-numbers layout

The Dura looks like a rugged phone that has been given slimming injections. It is plainer than most. There are no excessive sci-fi flourishes, and the lines are utilitarian.

At 13mm and 299.5g, it is slimmer and lighter than most rivals. It is thicker than the 11.9mm WP68 Air, but only just. It should sit in an average pocket without an issue.

The removable battery is the feature that shapes everything. Rugged phones with swappable batteries were common years ago, and they have mostly vanished.

The battery uses a simple Lock/Unlock slider on the rear of the phone, though it's worth noting that there is no capacitor in this phone, so before the battery is swapped, you need to shut down the Dura.

Somewhat confusingly, the battery has a circular depression and the CUBOT logo, which looks like it hints at wireless charging. But it turns out this is misplaced styling, and neither the battery nor the phone has this feature.

As I’ve already mentioned, you get a second battery, and it comes with a holder that has a USB-C port, so you can charge it without the phone. I don’t think it charges as quickly as in the phone, but that’s an incredibly useful feature that doesn’t cost extra.

The only downside to that charging shoe is that, other than a red power light, there is no feedback on the battery's charging status. Even four LEDs showing the state of charge would have been fine.

What’s clever about the battery design is that it has an integrated seal to keep out water and dust. That’s been achieved without making the battery difficult to remove. Once unlocked, two slots at the bottom let you hook a fingernail and pop it out.

Above the battery area is the camera cluster with the sensors on the left and a small-ish camping light on the right. What’s mildly confusing is that this area is the highest point on the back, but CUBOT chose glass as the surface, which seems like a poor choice.

The rest of this phone is rather boring, as it goes with the predictable side-mounted fingerprint scanner/power button and volume rocker on the right, and PTT button and SIM tray on the left.

The only other external feature of note is that there is a hole on the bottom right for a lanyard/strap, but one isn’t included.

Overall, there is little here to get excited about, but also not much to dislike either.

Design score: 3.5/5

CUBOT KingKong Dura: Hardware

• Unisoc T620
• 720 x 1640 display
• 2x 5000mAh battery

The Unisoc T620 is an entry-level chip that was first released in mid 2024. It uses an older 12nm process, not one of the newer 6nm, 4nm, or even 2nm processes flagship SoCs use today. It pairs two Cortex-A75 cores with six A55 cores, which has been a popular configuration for Octo-core SoCs since it first appeared with the MediaTek Helio P65 back in 2019.

To put that in perspective, this is older technology than the Dimensity 7025 in the Oukitel WP68 Air, which is itself a 6nm design from 2022. The Dura is a step behind that in the chronology of its technology. Even if it looks only2.5 years old to the casual observer.

The RAM is 8GB or 12GB, depending on the model. CUBOT adds virtual memory on top, quoting totals of 20GB and 32GB. As ever, that is storage borrowing the name of memory. Ignore it, since most people will run fine without the need to activate that feature.

Storage is 128GB or 256GB. A TF card slot is available, but it shares space with the second SIM. You can have dual SIMs or a card, not both, and there is no eSIM to fix that limitation.

The screen is the same 6.88-inch, 720 x 1640 panel as the WP68 Air. It runs at 120Hz. At this resolution, text looks soft in some situations, and it's difficult to read a document without zooming in. It’s reasonably bright, but the glossy coating makes outdoor viewing a challenge.

What makes this design interesting is the two batteries, each rated at 5000mAh, giving a total potential battery of 10000mAh. To be totally accurate, the rated capacity is 4850mAh or 17.945Wh, but the typical capacity is 50000mAh and 18.50Wh for each.

With 33W charging, these can be brought back to full capacity quickly, and both can be recharged at the same time. The only caveat is that the phone only comes with a single 18W charger, but I think this was a mistake at the factory. Why? Because it also came with an old-style MicroUSB cable, which would destroy the USB-C port if you forced it in there.

Therefore, if you buy this phone, I’d check that you have the right power supply and cable before hooking things up.

From a hardware perspective, you should also know there is no 5G comms available. For a phone aimed at workers and outdoor use, 4G is usually enough. But it limits future value.

• Hardware score: 3/5

CUBOT KingKong Dura: Cameras

• 64MP and 2MP on the rear
• 16MP on the front
• Three cameras in total

CUBOT KingKong Dura has three cameras:

Rear cameras: 64.2MP main camera (OmniVision OV64B40 or Samsung S5KGW1SP03), f/1.8, with phase-detection autofocus +2MP macro camera.
Front cameras: 16MP.

This is a basic arrangement that matches the price. The 64.2MP sensor is either an OmniVision OV64B40 or a Samsung S5KGW1SP03, but CUBOT doesn’t say which it is.

Based on the slightly flattened colour profile, I'd bet on the OV64B40, but I could be wrong.

Either captures decent daylight pictures, but isn’t especially remarkable when the light level drops.

The 2MP macro is, once again, the weakest link. Fixed-focus lenses of this type rarely deliver. Cropping the main camera works better, making the 2MP sensor, probably by GalaxyCore, superfluous.

The 16MP selfie camera is a fixed-focus unit and should be adequate for video calls, but again the actual sensor isn’t declared. It might be Samsung, Sony or OmniVision, since these all make 16MP sensors for phones.

The primary camera becomes exceptionally disappointing in video mode, since a 64MP sensor still only offers 1080p. What is completely inexplicable about this is that there are only three resolutions available in the camera app: 1080p, 720p and 480p, but when you choose the FPS, there is only 30fps available. I assumed that as the resolution lowered, I’d get 60fps, and maybe 90 and 120fps. So having an FPS selection feature where the only option is 30fps seems incredibly dumb.

Interestingly, the forward-facing sensor has the same resolution and FPS limitations, showing how underutilised the rear sensor is for video. The only thing that the rear sensor has that’s better is stabilisation.

As with all phones, I tested what video encryption it supports, specifically Widevine, and was surprised to find that this phone supports Widevine L1. With the limited resolution of the screen, that only bumps streamed video capability up from 480p to 720p, but it's better than L3.

• Camera score: 3/5

CUBOT KingKong Dura: Performance

The table compares the Dura with the WP68 Air, which I tested in an earlier review. It is only 14.5g heavier, but it has enough internal space for an 8000 mAh battery instead of the Dura's 5000 mAh battery.

And that should be the focus of this section, because the performance of both SoCs is dire compared with more modern designs, and the one in the Dura is especially lacklustre.

If you combine the two 5000 mAh batteries on the Dura, you would get over 21 hours of operation, an hour more than the WP68 Air. However, the WP68 had more unused capacity at the end of the test, and that phone had only 80% of the combined Dura battery.

This all comes down to power efficiency, and the MediaTek Dimensity 7025 appears to use its power much more effectively than the Unisoc T620.

Given that the Unisoc T620 is 12nm, and the MediaTek Dimensity 7025 is 6nm, this shouldn’t be shocking. And the GPU on the T620 has 32 shaders, a quarter of that in the IMG BXM-8-256.

Neither is a performance platform, but the Dura has one advantage: you could carry more than two batteries if CUBOT starts selling them separately.

But even with three or four batteries, there are rugged designs out there that could run for longer with their internal capacities.

• Performance score: 2.5/5

CUBOT KingKong Dura: Final verdict

The KingKong Dura has a self-evident personality. It is a cheap, tough phone with a battery you can swap, and its pitch is not any more complicated than that.

That could be a real advantage on a long job, a camping trip or a remote site, if you bring the extra battery along and you’ve remembered to charge it.

Its weaknesses are equally clear. The chip is slow, the screen is soft, and there is no 5G comms. These are budget compromises, and they are honest ones at this price, if you can accept them.

If you only need a phone that lasts, and you don't want to carry a power bank, it makes some sense. However, if you want a fast daily driver, the WP68 Air or something newer is a better bet. As the power equation is tied to efficiency and performance, not having either of these things can only save you so much battery capacity.

Should I buy a CUBOT KingKong Dura?

Buy it if...

Don't buy it if...

For more ruggedized devices, we've reviewed the best rugged tablets, the best rugged laptops, and the best rugged hard drives

• The average tech worker earns £60,600, but half of them also have a side hustle
• 49% have a hybrid working routine, enabling flexibility for side hustles
• The workers who need the money the most generally have the least flexibility

IT workers might be among the best paid, with UK employees earning an average of £60,600 - but one in two actually take on extra work alongside their full-time roles, new research has claimed.

This compares with barely a third (30%) of all full-time workers, making IT workers among the most likely to have side hustles – 23% freelancing and 7% having a more traditional second job, according to a Careerminds study.

But the reasons aren't so clear, because three in five (58%) earn at least £50,000 annually and IT is the least likely sector to experience payday-to-payday financial pressure – so if it's not money struggles that's causing them to get side hustles, what is it?

Tech workers earn good salaries but are still more likely to have side hustles

The report implies workers may just want to increase their disposable income or save more, but tech workers are also in an unfairly unique position because virtually half (49%) of them follow hybrid working routines, compared with just 4% in hospitality. This extra flexibility could allow them to sell their sought-after skills on a freelance basis.

"Some jobs naturally lend themselves to freelance work," certified professional career coach Amanda Augustine wrote. "If you're a developer, for example, you might be able to take on a project in the evening or over the weekend."

But while IT workers typically earn healthy wages, nearly half (47%) of UK adults in full-time employment say they run out of money before payday, doing so on average 7.9 days early.

To that tune, one in five have cancelled subscriptions (21%), declined social plans (21%) or even skipped a meal (20%) just to help them keep that money for living expenses.

"The people who need additional income the most aren't necessarily the ones who have the flexibility, time or energy to take on more work," Augustine warned, noting that struggling workers from the least flexible sectors may be at an unfair disadvantage.

Careerminds implies we're at the beginning of a challenging period, where certain workers and individuals may not be in such a good position to earn an extra income.

• Half of workers use AI to appease their bosses even when it's not helping
• One in three say they've held back negative feedback about AI
• AI ROI figures are being skewed because of these false figures

Half (52%) of workers admit to having used AI despite believing they could complete the same task faster without it.

This comes as two in three (63%) workers say their employers now actively encourage or expect AI use for at least some tasks, with as many as 39% even highlighting their AI use to managers despite believing it didn't help their outcome.

The data from JobAssist ultimately points toward workers seeing AI use as a box-ticking exercise and something that's beneficial to their professional reputation and potentially career prospects – even though they don't see it as something that can always help.

Workers are boasting about their AI use to keep managers happy

Furthermore, nearly a third (31%) admitted to continuing using AI even after deciding it had not improved speed or quality, implying that workers might be worried that something could happen if they don't demonstrate a use of it.

A similar number (34%) also held their hands up to avoiding negative feedback around AI given that leadership encourages its adoption.

"If employees hesitate to report disappointing results, managers may receive an overly favorable account of how well their AI initiatives are working," the report summarizes, urging workers to be more open and honest about when they find AI to be most effective.

For businesses, false adoption figures could mean that AI ROI might be overstates. But with employers relying on these statistics to guide their AI strategies, it could mean that they end up wasting money investing in the wrong areas.

"Companies can measure whether employees are using AI, but those measurements cannot establish whether the technology is improving their work," JobAssist concludes.

8849 TANK 5 Pro: 30-second review

The Tank 5 Pro has but one purpose - it adds thermal imaging to the Tank 5 and changes almost nothing else from the prior design. 8849 admits as much on its own website - the only difference between the two phones, it states, is the thermal camera.

This design was already the most complete rugged projector phone on the market. It had a flagship chip, a gorgeous screen and a battery that recharges in about 90 minutes. The Pro keeps all those strengths and adds one more: thermal imaging.

You get a FLIR Lepton 3.1R sensor with a 95-degree field of view. 8849 claims that is a world first for a phone. Most thermal phones manage 50 or 57 degrees. A wider view means you see more of a room, a wall or a fuse board without having to move the device around.

The price of that view is resolution. The sensor delivers 160 x 120 native pixels. Software upscaling takes it to 480 x 360, and visible light edges are overlaid to help you read the image. It is useful, but it will not rival a dedicated high-end imager.

So who is it for? Electricians, builders, surveyors, solar installers and rescue teams spring to mind. Anyone who has ever wondered where the draught is coming from might also apply. If you will never point it at a hot pipe, save your money.

Would I consider this one of the best rugged smartphones? That depends on how much functionality you want in a single handset, but it deserves some recognition for its unique feature set.

8849 TANK 5 Pro: price and availability

• How much does it cost? From $1150/£885/€809
• When is it out? Available by November 20 2026
• Where can you get it? Direct from the maker or via an online retailer

8849 has announced that this phone will be available from November 15, 2026 in the USA, with other regions from November 20, 2026, directly from the maker.

There are three SKUs: 16GB+512GB, 18GB+512GB and 18GB+1TB - US pricing for the three is $1149/ $1,199/$1,299, respectively. In the UK, that’s £885/£924/£1001, and in the EU those prices are €1.045,95/€1.091,95/€1.182,95.

Where is it cheapest? America obviously. It’s more than $20 per SKU in the EU and $45 more in the UK, making that location the most expensive place to buy.

Not sure why the UK gets the thin end of that wedge, but that’s the price 8849 decided.

I’ve noticed that 8849 has ditched the 16GB model of the original Tank 5, and it now only comes in the 18GB+512GB option for $999.99, £770 or €910,95 depending on your region.

That makes the inclusion of a Thermal camera an extra $200, £154 or €181 for each area.

But if it were my money, I’d buy the 16GB+512GB Tank 5 Pro model, as most people would never notice the difference, and it’s only $150 more than the 18GB+512GB Tank 5, and it has the thermal camera.

Competition is thin. Rugged phones with thermal cameras exist, but none that I know of also pack a projector that I’m aware of. The Oukitel WP70 Ultra offers thermal imaging and a 16,000mAh battery, but it’s not released, and no price has been confirmed.

The real comparison is with a standalone thermal imager, like those by ThermalMaster, but these typically start at around $250 for a USB-connected camera.

What these add-on cameras lack is the ability to merge the thermal data with conventional light camera data for image fusion, which makes for much better readability.

Obviously, this isn’t a cheap phone, but in terms of specifications and features, it seems like good value, especially if you live in America.

Unfortunately, the flip side of being in the USA is that the service providers are notoriously anti-consumer, and you will need to be on AT&T to use the Tank 5 or Tank 5 Pro. Curiously, other 8849 phones and tablets support Verizon and T-Mobile only.

In Canada, you can use Bell, Rogers, Videotron, Freedom, and Telus. In the UK, most will work, including EE, H3G, Sky, Telefonica(O2), Virgin, Vodafone and Gamma. Australians can use Optus, Telstra and VHA. Major carriers in other countries are supported, so you should be good if you check the supported frequencies before buying.

• Value score: 4/5

8849 TANK 5 Pro: Specs

8849 TANK 5 Pro: Design

• Thick and heavy
• Vents for the projector on both sides
• Now with a thermal camera

Hold the Tank 5 Pro, and it feels exactly like the Tank 5, which is good if you want something that feels built to double as a sledgehammer. The chassis blends brushed metal plates, rubberised corner armour and polycarbonate panels. It’s built for abuse, and if you drop it on your foot, you'll be the one giving, not the Tank 5 Pro.

The dimensions are identical to the Tank 5 at 177.1 x 87 x 33.8mm. So is the weight at 715g. That is surprising. A thermal module with its own shutter and optics has to go somewhere, and it must weigh something. Well, when I weighed it, the scales said 733g, so the thermal electronics don’t snub the laws of physics.

The thermal lens sits top right in the centrally positioned camera cluster, with the IR emitter below it. The visible light sensors are on the left. What’s great about this arrangement is that the primary sensor is physically close to the thermal camera, so when the two images are overlaid, there is almost no difference in the viewing angle between them, producing a good alignment when the data is merged.

The camping light bar, night vision lens, and four infrared LEDs remain in a V-shape in the middle of the back, which prevents this phone from offering wireless charging. Given the battery size, that wasn’t a practical option, so it's not a real loss.

The projector vents still split the left frame in two, and the right frame still has its second grille and fingerprint power key.

My verdict about this phone design remains unchanged, because the form factor is almost identical, and it's slightly heavier.

To be precise, this isn't a pocket phone, but one better suited to a belt holster or vehicle mount. Those travelling on foot wouldn’t take long to conclude this wasn’t the best phone to carry any distance.

One new note on the Tank 5 I made related to the thermal camera. 8849 recommends an operating range of 0°C to 35°C for the phone. That is fine for a summer survey, but not being able to handle below-zero temperatures might be an issue. It is less helpful for a winter loft inspection, which is exactly when thermal imaging earns its keep. However, if you want to warm up the Tank 5 Pro, running the projector for a couple of minutes will noticeably raise the device's temperature. And the thermal sensor has a range of -10°C to 400°C.

Design score: 4/5

8849 TANK 5 Pro: Hardware

• MediaTek Dimensity 9400e
• AMOLED display
• 17600 mAh fast charge battery

The Dimensity 9400e remains the star of this show, as it was on the Tank 5. It is a 4nm flagship chip, and it sits far above the mid-range silicon found in most rugged phones. The Immortalis G720 MC12 GPU keeps games and sustained workloads moving.

As I’ve already mentioned, this phone comes in three flavours, matching 16GB or 18GB of RAM with 512GB or 1TB of storage. My review hardware has 18GB of RAM and 512GB of storage. And I noticed that the only Tank 5 SKU now comes with this much memory.

The 6.73-inch AMOLED display is also unchanged. It runs at 3200 x 1440 with a 120Hz refresh rate and 3000 nits peak. My only complaint about this screen is that it needs a reflection-reducing coating or screen protector, but it can still be seen outdoors, thankfully.

Thermal work often means bright roofs and sunlit walls, and this screen copes most of the time. Protection is still Panda Glass rather than Gorilla Glass Victus, which remains a disappointment, if you like the latter.

The battery is 17,600mAh, built as two 8800mAh cells. That is how 120W charging works. A full charge takes about 90 minutes, and reverse charging at 25W turns the Pro into a field power bank.

There is a small caveat, and I only noticed it with this phone, not its predecessor. After charging at 120W for 90 minutes, the charger gets hot to the touch. I suspect this also happened with the Tank 5, but for whatever reason I didn’t notice.

In my experience, electronics that get hot eventually fail, so keep your receipt.

Another area worth mentioning is connectivity, since this platform supports WiFi 7 and Bluetooth 6.0, dual-frequency GPS, and eSIMs.

When I covered the Tank 5, the specs said Bluetooth 5.4, and now, magically, it's become version 6.0. Not sure what, if anything, changed there, but now both phones say the same BT 6.0 capability.

The eSIM shares the second SIM interface, so you can pair it with a Nano SIM and a microSD card, making a microSD less of a deal.

Overall, this is a terrific spec for a rugged phone.

• Hardware score: 4.5/5

8849 TANK 5 Pro: Cameras

• 50MP main + 50MP telephoto + 50MP night vision + Thermal camera
• 32MP on the front
• Five cameras in total

The 8849 Tank 5 has four cameras:

Rear camera: 50MP Samsung ISOCELL S5KGN1SP, 50MP OmniVision OV50C40, 50MP OmniVision OV50D40, FLIR Lepton 3.1R Thermal camera
Front camera: 32MP OV32B40

Here I’m going to focus exclusively on the Thermal sensor, because you can read my opinion of the other sensors in my Tank 5 review.

The Pro uses a FLIR Lepton 3.1R, which captures 160 x 120 native pixels, 19,200 calibrated temperature points. VividIR fuses frames in real time and upscales the image to 480 x 360. MSX overlays edges from the visible camera, making pipes, wires, and structures easier to read.

Those numbers sound modest, because they are. This is a compact sensor, not a professional handheld imager. Treat it as a very capable tool for finding problems, not for documenting them to survey grade.

Where it diverges from many phones that include a thermal sensor is the field of view, as 95 degrees is exceptionally wide. Typical thermal phones offer a 50 to 57-degree view, according to 8849, and that wider field is super-useful when you are working in a confined space. In a small room or a loft, that could be a big gain. You see more per frame, so you can scan faster and miss less, hopefully.

There is a trade-off, and it is simple maths. Spread 160 pixels across 95 degrees, and each pixel covers about 0.6 degrees. At 10 metres, one pixel spans roughly 10cm. That is my calculation from the stated field of view, not an 8849 figure. A small target at distance will be a handful of pixels, so that’s less of an issue.

I estimated that the nearest that this sensor will focus is about 15cm, which probably isn’t near enough to easily track problematic components on a circuit board unless they’re big.

Thermal sensitivity is under 50mK, or 0.05°C. That lets the Tank 5 Pro pick out subtle differences, such as damp patches or a warm pipe behind plaster.

There are two gain modes. High gain covers minus 10°C to 140°C, with finer detail for building work. Low gain stretches to 400°C for engines, machinery and flames.

Despite the resolution limitations of this sensor, what makes it exceptionally easy to use is the MyFLIR Pro app, which comes pre-loaded onto the phone.

As you might expect, it offers spot and area temperature readings, alarms, nine colour palettes, photo, video and time-lapse capture and picture-in-picture. Images are compatible with FLIR devices and FLIR Thermal Studio Suite for desktop analysis. That is a proper workflow, and it sets the Pro apart from cheaper thermal add-ons.

For those who need thermal accuracy, it’s a mixed bag. In high-gain mode, accuracy is ±5 °C or 5%, whereas in low-gain it's double that. To put that in simpler terms, low-gain isn’t suitable for soufflé baking if your oven isn’t reliable.

In terms of Emissivity, there are four choices: matte (recommended, semi-matte, semi-gloss and glossy (not recommended. These settings allow you to match the texture of the object you are scanning, avoiding wildly inaccurate readings.

In video capture, the rate is only 7.9 fps, but with these devices, less than 9 fps is normal.

The team behind MyFLIR has focused on making it easy to use, and it certainly ticks those boxes. It still won’t compete with a dedicated camera for flexibility and accuracy, but for jobs like plumbing and working on engines, it's more than adequate.

8849 TANK 5 Pro Camera samples

• Camera score: 4/5

8849 Tank 5: Performance

• 4nm SoC
• Premium performance

Not sure if there is much value in comparing the Tank 5 Pro and the Tank 5, since under the hood it’s the same hardware. But if only to prove that, I cycled the benchmarks, and they’re within 1% of each other in most tests.

The Tank 5 Pro is a high-performance phone, as was its father before it, but what these numbers can’t show is the impact of running the thermal camera, as they don’t show the impact of using the conventional sensors or the projector.

Those tests are beyond the remit of this review, but undoubtedly the projector is a major drain on power, based purely on how warm the phone becomes when running it. The thermal sensor is much less demanding, and with 17600 mAh to play with, most users are unlikely to notice much impact.

The Tank 5 Pro is a powerful phone, so there is no change from its predecessor.

• Performance score: 4.5/5

8849 Tank 5: Final verdict

The Tank 5 Pro is the most capable rugged projector phone yet made, and it puts the cherry on that cake by being able to see heat.

Without question, the thermal camera is the reason to buy, as it makes the Tank 5 Pro the complete package. It isn’t the highest-resolution thermal solution around, but the 95-degree view makes it quick to use, and the FLIR software makes it credible.

Nothing else has changed, including the drawbacks. The phone is very heavy. The projector still needs darkness. And an update policy that 8849 now says covers five years needs to be proven. But given how competitive the rugged phone market has become, launching phones on a two-year life cycle probably isn’t enough for more savvy purchasers.

The choice is straightforward. Need thermal imaging? Buy the Pro. Do not need it? Buy the Tank 5 and pocket the extra cash. Well, not pocket it, since the phone definitely won't fit then, if it does now.

Should I buy a 8849 Tank 5?

Buy it if...

Don't buy it if...

Also Consider

For more ruggedized devices, we've reviewed the best rugged tablets, the best rugged laptops, and the best rugged hard drives

• Researchers found over 1,400 exposed ARTEX instances online, with many linked to offensive security tooling
• ARTEX combines AI agents with reconnaissance capabilities and has been increasingly adopted in real-world attacks
• Some operators accidentally exposed malware, exploits, credentials, and internal infrastructure through misconfigured servers

Around 1,500 malicious instances of ARTEX were found on the internet, available to anyone who knew where to look, experts have warned.

Security researchers Censys said they identified 1,465 hosts running exposed ARTEX consoles. Of those, 155 (roughly 10%) were also running at least one tool commonly used in cyberattacks, including Cobalt Strike, Vshell, or Adaptix C2. They also found 32 publicly accessible directories on ARTEX hosts, some of which exposed exploit binaries, post-exploitation tools, and other sensitive files.

Usually, the ARTEX console does not require public-facing access, the researchers said, stressing that overall global deployment is “likely higher”.

Hijacking ARTEX

ARTEX is a legitimate penetration testing tool hijacked for malicious purposes, meaning at least some of these instances were set up by threat actors.

The story of ARTEX is similar to that of the now infamous Cobalt Strike. It was released by a developer called Autumn-27, who also authored ScopeSentry - a tool Censys describes as a “widely used Chinese asset-discovery and vulnerability-scanning platform”.

The researchers described ARTEX as an “AI agent layer on top of that recon foundation”. In other words - it is an LLM-powered version of Scope Sentry that syncs assets found by the tool’s domains, subdomains, IPs, ports, and endpoints.

ARTEX was released in late July 2026 and quickly gained popularity. On October 7, security researchers CrowdStrike reported seeing it used in malicious attacks and a day later, the developer distanced themselves from the illicit activity and stressed the tool was built for testing within authorized asset scopes. They then turned it into a closed-source tool with no further releases. However, despite the repository being taken down, Censys claims archived copies, translations, and different rebrands are still circulating.

Since then, numerous researchers observed it being used in different attacks. The CrowdStrike report described an unidentified threat actor using ARTEX to target South Korean financial institutions. The attacker, believed to be a financially-motivated Chinese actor (as opposed to being a state-sponsored one), compromised a broker-facing loan inquiry service at one bank and an employee mobile system at another. In both instances, they walked away with sensitive data.

They used ARTEX with DeepSeek v4.1-flash as their primary AI model, while also running Claude Code and Grok. The researchers found the operation in part because the attackers left publicly accessible directories where they exposed prompts, session histories, and ARTEX configuration files.

Censys later said the directories had been left exposed in mid-August, roughly seven weeks before CrowdStrike published its findings. File names also suggested they were targeting payment and gaming services, as well.

At one point Microsoft linked it to Storm-0940, a China-aligned threat actor responsible for compromising ConnectWise Cloud Backup customer environments. Apparently, the threat actors used ARTEX to automate scanning, password spraying, and OAuth application abuse, peaking at 2,900 sign-in attempts per minute at one point. Some Microsoft’s customers were allegedly seeing privileged cloud access granted in less than three hours.

Exposing their own infrastructure

Now, Censys says ARTEX is being paired with other offensive tools. At least 70 hosts were seen running both ARTEX and CyberStrikeAI, and others were seen deploying PentAGI, HexStrike AI, and RedAmon. Ironically enough, the researchers also found some actors had inadvertently exposed sensitive information through poorly secured directories. One installation exposed the operator’s entire home directory, including SSH configuration files and shellac command history, while others contained Windows lateral movement tools, malware implants, reverse-shell scripts, and exploit code.

Censys also stressed that not all of the exposed instances are malicious. They also said that roughly 650 of the detected hosts belonged to two groups of identically configured hosting infrastructure, leaving approximately 815 unique installations.

Available in black, gray, and pink colors, the Hinomi X2 Pro Ergonomic Office Chair is designed for anyone spending long hours at a desk, with a four-panel backrest, adjustable lumbar support, and 6D armrests. The successor to the X1, it offers 15 adjustable elements to help accommodate different body shapes and sitting preferences.

Best For: Home office workers, professionals spending long hours at a desk, and anyone wanting extensive ergonomic adjustments.

The Deal: The Hinomi X2 Pro Ergonomic Office Chair is reduced from $899 to $699 for Black Friday, saving $200 (22%). The promotion runs from November 1 through November 30.

Why we recommend it

Finding an office chair that fits properly can be difficult, particularly if you're taller, shorter, or regularly change how you sit. The Hinomi X2 Pro addresses this with a height range of 5'1" to 6'6" and 15 adjustable elements.

Its four-panel backrest supports different areas of your back, adapting to changes in posture. The 3D lumbar system offers additional adjustment for your lower back, and the 6D armrests can be repositioned to suit different tasks, including typing, reading, and leaning back.

The wear-resistant mesh upholstery will help with ventilation, and the aluminum alloy frame and base promise a sturdy foundation. The chair supports up to 330 lbs and has been tested to BIFMA standards, with verification by TÜV. Hinomi also offers a warranty of up to 10 years.

In his four-star review of the Hinomi X1, our expert Collin said: "This chair has fantastic adjustability, allowing it to be customized to your liking and even your specific use case." He summed up his review saying: "All around, this is a fantastic chair. If you are looking for a great ergonomic chair, this one is a great one to consider."

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

Price context & historical value

The Hinomi X2 Pro normally sells for $899, but the Black Friday promotion brings it down to $699, a huge $200 saving. This is the lowest price of 2026, and the deal runs until November 30.

British buyers can snap up the chair for £699, down from £899, from here.

Should you buy it?

✅ Buy the Hinomi X2 Pro if...

If you spend a lot of your working day sitting and want a chair with extensive adjustments. The four-panel backrest, 3D lumbar support, and 6D armrests offer plenty of ways to tailor the fit, and the mesh upholstery and substantial warranty add to its appeal.

❌ Skip the Hinomi X2 Pro if...

You don't spend a lot of time sitting at your desk or prefer a traditional padded office chair. Even with $200 off, the X2 Pro is a considerable investment, and its extensive adjustment options may be unnecessary for some users.

• Shop all the office chairs on sale at Hinomi

During the winter months, working at a computer can be uncomfortable, especially if suffer from cold hands like I do. The Afenau heated mouse pad offers a simple solution, warming the surface beneath your keyboard and mouse while doubling as a full-size desk mat.

With three temperature settings, a waterproof surface, and a built-in digital display, it's designed to make long sessions at your desk far more comfortable.

Best For: Home workers, students, gamers, and anyone who spends hours at a computer in a chilly room.

The Deal: The Afenau 80 x 33cm heated mouse pad costs £17.59 at Amazon. It offers three heat settings, a waterproof PU leather surface, and automatic shut-off.

Best Heated Mouse Pad deal

Why we recommend it

With heating bills remaining a concern during the cost of living crisis, the Afenau heated mouse pad offers a way to keep your hands toasty warm without turning up the heating. Measuring 80 x 33cm, it provides enough space for a keyboard and mouse, with heating elements that cover 95% of its surface.

There are three selectable temperatures: 40°C, 45°C, and 50°C and the digital display in the upper-left corner shows the selected temperature, which the heating system can reach in 10 seconds.

The polyurethane surface is waterproof and resistant to everyday spills and scratches, with a non-slip underside to keep everything in place. At 550g, the mat can be rolled up for storage when you're not using it.

There's a visual reminder designed to encourage you to take a break after an hour of sitting and automatic shut-off offers another safeguard.

Price context & historical value

At £17.59, the Afenau heated mouse pad is an inexpensive way to add warmth to your desk without investing in a heater. Its combo of adjustable temperatures, waterproof construction, and digital controls provide several features beyond those of a conventional desk mat.

This is the cheapest price we've seen it at, so if you want to stay warm while working this winter, I'd definitely recommend it.

Should you buy it?

✅ Buy the Afenau heated mouse pad if...

If you work in a chilly room or suffer from cold hands, this is a great solution. The large dimensions, three heat settings, and wipe-clean finish make it a practical desk accessory.

❌ Skip the Afenau heated mouse pad if...

If your workspace is already warm or you need something that heats the entire room. No manufacturer warranty is specified either.

• Shop all desk pads at Amazon

• Attackers are actively exploiting the Plex2Shell vulnerability chain to compromise internet-exposed NAS devices
• QNAP urges immediate updates, as older Plex Media Server versions allow remote code execution and full device takeover
• Systems exposed before patching may already be compromised and require credential resets and full remediation

QNAP has issued a security advisory warning users to install the latest version of the Plex Media Server app on their Network Attached Storage (NAS) devices.

It says older versions are being actively abused in the wild using the dreaded Plex2Shell vulnerability chain.

“Plex has confirmed that these vulnerabilities are being actively and widely exploited,” the company said in the advisory. “QNAP urges all users running Plex Media Server on a QNAP NAS to update immediately and to review their devices as described below.

What is a QNAP NAS and what does it have to do with a flaw in Plex?

A Network Attached Storage device is a dedicated system allowing users to store, back up, and share files across a network. Essentially it’s a series of hard drives connected to a local network for easy access. Consumers and businesses alike often use it as an alternative to subscription-based cloud storage models, and sometimes even to stream audio and video files.

One popular way to do that is via a software application called Plex Media Server, which allows users to organize and stream their personal collection of movies, TV shows, music, and other multimedia content. By installing Plex on their NAS, and a compatible player app on a smartphone and similar device, users can access their media libraries over a local network or remotely via the internet.

This last part - via the internet - is where trouble was recently found.

Earlier in November 2026, Plex disclosed having fixed four vulnerabilities in Media Server versions 1.43.2 and older: CVE-2026-96652, CVE-2026-96654, CVE-2026-96655, and CVE-2026-96656. Separately, these four are relatively dangerous. Chained together, they allow unauthenticated threat actors to execute arbitrary code with the highest privileges, remotely. In other words, they can use Plex Media Server to take over the QNAP NAS and, if they choose to, access and steal sensitive content found there. These four have been given the nickname Plex2Shell.

Plex released a fix in September 2026, and a proof-of-concept (PoC) exploit was made available a few weeks later. Since then, multiple security companies reported seeing Plex2Shell abused in the wild, mostly to install proxy software and cryptocurrency miners. The latter might not be particularly useful, given that most miners require powerful GPU units, but the former could prove lucrative.

Patching Plex

A proxy network allows cybercriminals to route their traffic through legitimate residential infrastructure, helping them not just to hide their identities and whereabouts, but also to launch disruptive Distributed Denial of Service (DDoS) attacks. As such, residential proxy services are always in demand, and can earn their operators a hefty sum.

The US Cybersecurity and Infrastructure Security Agency (CISA) added all four bugs to its Known Exploited Vulnerabilities (KEV) catalog on October 20, and gave Federal Civilian Executive Branch (FCEB) agencies a mere three-day window to apply the fix or stop using Plex entirely.

That being said, Plex addressed the flaws in Media Server version 1.43.3 (build 1.43.3.10861), which can now be downloaded via QNAP’s App center. The company confirmed that the bugs exist only in Plex Media Server, QTS and QuTS operating systems were not affected. Still, Plex said if the NAS device was reachable from the internet while running older versions of the software, updating is not enough and the NAS should be treated as compromised.

Those users are advised to complete a clean reinstall, compromise check, and credential reset.

The exact number of internet-accessible QNAP NAS devices is difficult to determine, but according to figures from BitSight, there are tens of thousands of them, right now. That does not mean all of them are vulnerable (many might not be running Plex Media Server to begin with), but the theoretical attack surface is quite large.

Via BleepingComputer

Timekettle LiveTalk AI Translation Earbuds: 30 second review

The ideal translation device would translate live as you talk, so a conversation in two different languages can happen naturally. This was the concept of the fictional Babel fish in The Hitchhiker’s Guide to the Galaxy, although in that case the earbud was a fish that you stuck into your ear.

The Timekettle LiveTalk on the surface promises this live-talk potential with near-instant text translation, plus audio. The earbuds have several different translation modes that are selected through the TimeKettle App, and, of course, AI is built in to boost the humanisation of those translations.

The way translations work depends on which of the four modes you choose. Whether you’re talking to someone face to face, in a meeting, listening to discussions, tuning in to multimedia content in another language online, or on a phone or video call, you must select a pickup mode before you start. From there, you can select normal speech-to-text translation, which, with a strong Wi-Fi or mobile signal, is fast once the initial translation starts. The more interesting part is activating audible voice translation as well.

The actual audio voice is good, and you can select either male or female. If you have in mind that this will be just like BabbleFish translating live with the audio in your ear, think again. The delay is 7 seconds; this means you still have that start-stop approach to conversation at best. It’s more like having a translator standing with you rather than live translation.

The speech-to-text translation is faster, but in either case, alongside the slight wait, you still have that disconnect. One of the most interesting features is sharing the earbuds: each person in a two-way conversation has an earbud, with speech and translation happening. The effect is good once you learn to speak in short bursts of no more than 7 seconds, although you can speak for longer; the translation will start to kick in, and due to the delay, having someone talking to you and the translation at the same time, translating some 7 seconds later, can all be a lot to take in and a bit confusing.

At present, however, when you factor in the different translation approaches and the range of translation options, these small, easy-to-use earphones perform surprisingly well for other situations such as watching live presentations, some multimedia content and quick conversations.

Timekettle LiveTalk AI Translation Earbuds: Price and Availablity

At present, the TimeKettle Live Talk AI Translation Earbuds retail for £129 and are available directly from the TimeKettle website as well as other major online retailers such as Amazon.

• Score: 4.5/5

Timekettle LiveTalk AI Translation Earbuds: Design

The earphones arrive in a compact Star Silver charging case; that colour profile is essentially white with a slightly pearlescent finish. The supplied case is of a nice design and helps keep the earbuds clean and topped up with charge.

The case itself is simple, with the TimeKettle logo on top, but otherwise minimalistic, with the compact design measuring 65 x 74 x 22 mm and weighing 78g. This makes the design easily pocketable, and the closing mechanism is solid, meaning that when closed, it stays closed. Otherwise, on the exterior there’s just the single USB-C port along the rear edge for the power cable. The cable is also supplied in the box.

Opening up the case, the two over-the-ear buds are magnetically held in position in the case, with charging contacts that enable the two buds to be charged in the case in the usual way.

Alongside the hardware is the app, simply named ‘Timekettle’, and this is where the majority of the adjustments and settings for the translation mode, style and language can be made. This is all clearly laid out, and after the initial Bluetooth connection, you can get straight into using the buds and start translating.

Before you start a conversation, you need to select the type of translation you want, and there are four main types: Face to Face, Live Listening, Voice Call Translation, and Multimedia Translation.

Once you select the translation scenario you want, you can go to the main translation screen. At the bottom is a play button that needs to be pressed to start the live translation; this isn’t initially apparent. However, after the first use, the workflow becomes clear: the play button starts the translation process and doesn’t play back audio. For each of the modes there are also a series of options and these are all well laid out and clear to use.

• Design: 4.5/5

Timekettle LiveTalk AI Translation Earbuds: Features

The earbuds themselves are nicely designed, and along with a solid feel to the case, they all feel like a premium product.

The earbuds follow the open-ear hook design, which is comfortable and also lets you hear conversations around you. At 6.9g per bud, they’re also extremely lightweight.

The electronics inside have also been fine-tuned for translation and playback, with dual-microphone air-conduction voice pickup and a dedicated translation EQ that helps to enhance the audio captured.

Through the app, you can also access six EQ preset modes for music. While using these earbuds for more than just translation seems obvious, it isn’t always the case; and it’s great to know that you can use these for calls and as an all-day option rather than just being isolated to translation use.

My review sample arrived in the Star Silver colour profile, but if that doesn’t appeal, then they’re also available in two other colours; Polar Black or Deep Sea Blue.

For connectivity, the earbuds use Bluetooth and support dual connections so that you can connect to both your phone and computer at the same time. There’s also plenty of compatibility with applications such as Zoom, Microsoft Teams and many more.

Battery life, once fully charged, is surprisingly good, with around eight hours per charge and the case providing power for a top-up charge. The earbuds are rated for 4 hours of translation or 8 hours of music playback per charge, with the charging case extending total translation use to approximately 16 hours. Timekettle also advertises up to 35 hours of total battery life, while a quick 10-minute charge provides an additional 2.5 hours of use.

I’ve reviewed many of these translation devices. Timekettle has been one of the more consistent players in the market, with a full feature set that makes its devices useful in the real world and a viable, more advanced alternative to Google Translate.

Google Translate is, of course, free. In contrast with the Live Talk earbuds, there’s an outlay for the product and then often an additional cost for the extended use of the translation service, although with 200 minutes a month included for free, that does give you plenty of scope for use. The paid tiers just extend the translation amounts if you need the functionality for business, for instance.

The first major difference between Live Talk and Google Translate is real-time, two-way translation, with the text translating quickly as the conversation happens. This doesn’t differentiate it from Google much at first, other than that I found the transcription speed faster. The translation accuracy was essentially the same, but LiveTalk also offers 25 offline language pairs through downloadable language packs. This is useful when travelling without a reliable connection, although Google Translate also supports offline translation.

Although without the AI boost, the translations are a little more formulaic. The translation through the TimeKettle App is easy to read, and I found the user interface is far more user-friendly than Googles.

But then, with a quick tap of the voice button, that face-to-face conversation is not only translated into text but also spoken directly into your ear. The translation speed to text is fast, although it depends on your connection speed and ambient noise; it's impressive nonetheless, but the translated audio is notably slower.

In the app, you select the translation environment you need: Face to Face for one-on-one conversations, Live Listening for lectures and meetings, Voice Call, or Multimedia Translation.

Currently, it supports up to 54 languages and 108 regional accents. For example, there’s French and French Canadian. While the system works best with an online connection, it also offers 25 language pairs that can be used offline. The system also uses the Babel OS 3.0 AI translation engine, which you can switch on and off; testing showed it improved translation quality and speed.

One of the features that impressed me was the automatic language detection, although you can still manually select the languages you want, for example, English to French.

Some of the most useful transcription features include real-time speech-to-text transcription with bilingual transcription that’s displayed on the phone’s screen. You can also ask for AI-generated meeting summaries and other summary templates for different situations; this is a feature that I have liked on AI note-taking devices in the past.

As ever, transcription has paid tiers, but the free tier is surprisingly good: up to 200 minutes of translation a month, covering all four translation modes.

Other paid plans include the Unlimited Plan at £9.99 a month and a yearly option at £99.99 a year.

• Features: 4.5/5

Timekettle LiveTalk AI Translation Earbuds: Performance

The LiveTalk earbuds arrive in a compact charging case, and before starting the test, I left them overnight to ensure both the earbuds and the charging case were fully charged.

Once fully charged, I ran through the setup process, which involved downloading the app, opening the charging case with the earbuds inside, and tapping the Connect Device option. At that point, the LiveTalk hardware was detected. There was a little to-and-fro with the app and checking the Bluetooth connection, but overall it was well guided and connected quickly.

On the home screen, you have four translation modes: Face to Face, Live Listening, Voice Call, and Multimedia Translation. I started with the multimedia translation.

I found that getting started with this feature took some work. Initially, activating translation while you play back a video in French is simple. Just activate the Multimedia mode, click play, which sets the app up for translation as soon as it hears the vocals, then go to the media source, in this case, YouTube, and tap play. The video then starts to play, and about three seconds later the text translation appears, followed by the audio translation, which starts about seven seconds after the original word is spoken.

The written transcription and vocal translation are accurate, but the audio delay takes some getting used to, and it’s surprising how quickly things can move on in a YouTube video. The text translation, on the other hand, is almost instant.

There were a couple of issues, though: the original audio and the translated audio came through at the same time, albeit with a 7-second delay on the translation. With the default settings, the original video drowned out the translated audio. In the settings, it’s easy to find the translated audio volume. However, the original audio slider wouldn’t move on iOS, so I had to turn down the phone’s volume to balance the audio. That seemed a little convoluted, but it worked.

Once that was done, and slightly ignoring the delay, it did actually make it possible to watch videos with the translated audio coming through the earbuds. The delay is around seven seconds; while it’s fine for watching business presentations in another language, it wouldn’t be so useful for watching the latest French or Swedish drama. There’s also the issue that there is only one voice, so in one video, as it switched between the two presenters, they both had the same voice, making it difficult to determine which of the two was speaking. In this instance, that wasn’t an issue, but if you’re watching a drama, subtitles might still be the better option, even with the delay. To be honest watching a full drama with the earbuds isn’t really the intended use, but I was interested to see what would happen.

Next, moving on to Face to Face, and here you have two options. The first is to share the earbuds, with each person in the conversation wearing one. I tested this by being the English speaker while a friend who speaks French wore the other. In this situation, you assign the left to, say, French and the right to English.

Again, there is a delay, although this time it was five seconds. Another point I noted was that if you speak for more than five to six seconds, the translation starts for the person you’re speaking to. This obviously makes sense, but it can be hard to understand when the person speaking to you is in French, and the translation comes through at the same time in English, just five seconds behind.

The other option is to wear both earphones, with the other person using the phone. This is a bit of a mix-and-match approach, and in the end, just reverting to the text translation on the phone, which is extremely quick, worked well without the distraction of the delayed audio translation.

Another point is that when I first tried this, I was in a crowded café, and the translation worked well at first. But as the noise in the café built up, the translation became less accurate, often picking up audio from people around us, which became an issue.

Moving on from Face to Face, I used Live Listening, which worked well, although there was again a delay. Here, I had a colleague read a few paragraphs of French in the hall below the office. The translation of what came through was again delayed by about five seconds; however, that delay was far less distracting. Again, the text transcription appeared much faster, although in a lecture, the benefit of looking at the presenter or the person you have come to see is far more valuable than sitting in a live lecture looking at your phone. While it isn’t ideal, it does work well.

The final option was the LiveTalk Voice Call Translation. This option was surprising, as I think my mind is still attuned to slow video calls. To start the conversation, I had to switch from Speaker to Timekettle LiveTalk in FaceTime's speaker output options, and then the original and translated audio started coming through. Again, as with Multimedia mode, I had to adjust the phone’s audio.

The only other point is, again, remembering to assign the correct voice to the right person.

Alongside direct translation, these earbuds offer a few other options. First, every translation is transcribed so you can check it later, along with the original audio recording. You also get AI summaries, which could be really useful. However, during this test, when I tapped the Generate button, the key points for most translations appeared, giving a decent overview of what was being said. However, the AI Minutes seemed a little more temperamental, with ‘Generation failed’ appearing on quite a few of the discussions that I had.

Another feature I liked about these earbuds is that they can be used all day and for multiple tasks, since they’re not limited to AI translation. The open design hooks around your ears and sits just above the ear opening, leaving your ear canal unblocked so you can hear everything around you clearly. This means you can wear them during normal video meetings, while listening to music, and when making calls. If you’re used to inear or cans this does feel different, but is ultimately a more work friendly design.

I also found the battery life pretty decent, and the 4 hours Timekettle quotes for translation are pretty much on the mark.

• Performance: 3.5/5

Timekettle LiveTalk AI Translation Earbuds: Score card

TimeKettle LiveTalk: Final Verdict

At the start of the test, I was initially impressed with the translation speeds, especially for the speech-to-text, which is almost real-time as long as you have a decent network connection. As I progressed through the test and the four modes, I had mixed results.

The translation quality was good from English to French, and vice versa, and the audio translation was also better than I expected; you can tell it's a computerised voice that lacks some of the tone and emotion you would expect from a human, but it's still impressive.

The most exciting part of using these headphones was the conversation aspect, and the hope that, like the Babel fish, I could simply stick one in my ear and start conversing in other languages. Unfortunately, this wasn’t quite possible. The delay was just too great, and with the person speaking and the translation coming at you at the same time, it’s just a little too much to take in all at once.

I found that the speed of the speech-to-text translation worked far better, and with the AI enhancement, there was less direct translation, and what was translated became more natural. However, this isn’t too far removed from Google Translate, although the speed and accuracy do seem to be better, and of course the big advantage is that you can use it offline with the language packs.

The points that really stood out were the speed of translation and the live translation for events and seminars; while there is that seven-second delay, it just isn’t so noticeable.

For multimedia and video conferencing as well, once you work out how to reduce the speaker's volume, that becomes far more useful, and it’s surprising how much easier it can be for speakers in their native languages to express themselves through translation. However, you’re always aware of that five- to seven-second delay.

While it still has some way to go before it offers live speech-to-audio translation, it is getting close. If you travel abroad regularly and often get involved in short conversations, this will be incredibly useful. Otherwise, as a speech-to-text translator and for understanding live events in other languages, it does work well, and I found it more convenient than Google Translate for continuously translating large volumes of spoken words without having to stop and start.

• Crédit Mutuel Arkéa confirms cyberattack led to the theft of roughly $133 million from the bank itself
• Attackers reportedly converted stolen funds into cryptocurrency and laundered them across multiple services and blockchains
• Bank says customer assets were unaffected, while investigators continue probing the third-party compromise

Crédit Mutuel Arkéa, a major French banking and financial services group, has confirmed it recently suffered a cyberattack that resulted in the theft of around $133 million.

The stolen money apparently did not belong to customers, but rather to the bank itself, and was converted into cryptocurrency to be siphoned out more easily.

The attack was first spotted by ZachXBT, a cryptocurrency analyst and blockchain fraud investigator who issued an alert on November 2 2026. “An EU entity is suspected of losing $106M+ from multiple centralized exchange accounts yesterday (Nov 1),” he said at the time. “The suspected stolen funds were withdrawn as USDC & USDT by an illicit actor.”

Blockchain as the tech of choice for both sides

It appears the attackers converted fiat currency into different cryptos before sending them out. As ZachXBT explained, the threat actor was “aggressively swapping” USDC (a stablecoin currency, one whose value is pegged to that of the US dollar) for ETH (Ethereum) and transferring the tokens to instant exchanges such as WellWallet and Keyrex. Some of the funds were converted to XMR (Monero, a known privacy-first token whose transactions are notoriously difficult to track) via Wagyu, and later bridged “across multiple chains.”

The irony is pbvious - blockchain and cryptocurrencies are often perceived as the go-to platforms for financial fraud and while many defend it by saying the fraud conducted via fiat is magnitudes larger, the fact remains the crooks transferred dollars to cryptos during theft.

On the other hand, the attack was not even discovered (at least publicly) until it hit the public blockchain where analysts like ZachXBT were able to see it and report on it.

A day later, ZachXBT issued an update, saying he identified an additional $27M+ of withdrawals, bringing the total amount stolen to $133M+.

“My working theory is a large financial institution in Western Europe was exploited and it’s currently being covered up,” he said. “No proper incident disclosure has resulted in $30M+ being laundered away which could have been frozen by EU law enforcement,” he added, suggesting that Crédit Mutuel Arkéa handled the situation poorly.

In his reports, however, he never mentions the name of the organization targeted. Crédit Mutuel Arkéa released a brief statement on November 4, saying it suffered an attack which it later managed to contain.

“As soon as this attack was detected [...] the decision was made, as a precautionary measure, to temporarily suspend certain transactions in order to secure operations and prevent any further fraudulent movements,” the announcement reads.

“This incident is closed and is now under control. Operations are being gradually resumed over the course of the morning, after the group and its subsidiaries ensured that checks had been carried out and enhanced security measures had been put in place.”

Not Lazarus

In an FAQ published a day later, the company said the attackers “went through the platform of one of our service providers and thus reached Crédit Mutuel Arkéa directly.” It did not want to say which service provider that is, but ruled out Fortuneo since that name was apparently being mentioned in local press.

The attackers used Arkéa's own account to carry out transactions with other banking players, it said, confirming that customer funds are safe and unaffected.

“We would like to reassure all our customers that this incident has no consequences for their assets. It only led to a delayed execution of certain transfer operations,” the FAQ further states.

The bank did not want to confirm the amount of money that was stolen, but commenting on ZachXBT’s $133 million figure, said it can “simply validate the order of magnitude.” In other words - it’s a close estimate.

So far no threat actors claimed responsibility for this attack, and chances are none will. ZachXBT believes this was not the work of notorious North Korean actors Lazarus Group.

Via BleepingComputer

• Iris Ohyama discloses breach affecting 7.7 million individuals after attackers compromised its ecommerce platform
• Exposed data included personal details and some encrypted passwords, prompting password reset recommendations
• Company continues investigating the attack, with no ransom demand, attribution, or confirmed misuse reported

Major Japanese consumer goods manufacturer Iris Ohyama has confirmed suffering a cyberattack which saw it lose data on 7.7 million people.

The company first disclosed the breach on October 29, 2026 when it confirmed “unauthorized access on the e-commerce site” Iris Plaza. At the time, the company said its investigation was ongoing, and that it would temporarily suspend the site to ensure customer safety. While it was looking into possible data leaks, it stressed that credit card information was not taken.

A week later, Iris Ohyama came forward with a more detailed report, in which it said that on October 28, if found “suspicious files” on an external cloud service it used. The subsequently launched investigation confirmed that the attackers stole sensitive data on 7.7 million people, which included member data on approximately 5.8 million people, and non-member data on 1.9 million.

Member data included names, addresses, phone numbers, email addresses, genders, dates of birth, and encrypted passwords (for some users). The non-member data was limited to names, addresses, phone numbers, and email addresses.

Hashed passwords and stolen credentials

The fact that the passwords were encrypted is good news, since it means the accounts cannot be accessed that easily. However, the company did not say which encryption or hashing method it used, so it is possible that the attackers could break some of them offline. It is thus no wonder the company warned users to update their login credentials, especially for those who use the same password for other services.

Those that do use the same passwords across various platforms are also at risk of having all of them compromised, since threat actors are known for trying out the stolen combinations across a multitude of services.

We don’t know who targeted Iris Ohyama, and to what end. No threat actors have yet claimed responsibility for this attack, the data has not yet leaked to the dark web, and we’ve found no evidence it was used in phishing or social engineering attacks.

In a statement given to BleepingComputer, the company said it received no direct communication from the threat actors, no ransom demands, and did not know who they were.

We also don’t know which third-party cloud service Iris Ohyama used, which malware was used, or how it was planted there. In the meantime, the company notified the country’s data protection watchdog agency, as well as local law enforcement authorities.

Targeting Japanese organizations

Over the last few months, Japanese organizations have experienced a barrage of cyberattacks, targeting firms of different sizes and in different industries. Just last month, we reported on an attack on GMO, a data breach incident at Japanese Burger King, and a malware attack on Daiichi Kosho. The GMO attack was particularly worrisome, since the company urged its customers to “unplug any cords connected to your PC”.

Burger King Japan also recently lost sensitive data on 11.18 million customers, most of which were using the company’s official smartphone app, while a malware attack on Daiichi Kosho’s supply chain resulted in the compromise of 8.7 million of both customers and employees.

All of these are large organizations and major enterprises, and Iris Ohyama is no exception. The company is a large consumer goods manufacturer, known for household appliances and everyday products, and has been around for decades offering air conditioners, vacuum cleaners, kitchenware, lighting, pet food and accessories, as well as different food products like rice and water.

For the financial year 2025, the company reported an annual revenue of $1.53 billion, with a headcount of more than 6,300. It also has a substantial international presence, with manufacturing and distribution operations in the US, Europe, China, and South Korea.