Technology

• FEVM FA80G mini PC packs a desktop Ryzen 9 9950X3D inside a 2.6-liter chassis
• The FA80G pairs AMD’s gaming-focused processor with Nvidia’s 24GB mobile flagship
• Nvidia gives the RTX 5090 Laptop GPU 150W inside this compact system

FEVM has unveiled a new compact desktop system pairing the Ryzen 9 9950X3D processor with a 24GB RTX 5090 Laptop GPU.

The pairing sits inside FEVM's FA80G chassis, a compact system built around AMD's AM5 desktop platform rather than mobile silicon components.

The flagship processor contains 16 cores, reaches 5.7GHz and carries 144MB of cache through AMD’s 3D V Cache technology.

Ryzen 9 9950X3D and RTX 5090 pack serious power

FEVM allows the processor to receive up to 90W, while the accompanying RTX 5090 Laptop GPU can receive up to 150W.

Nvidia’s mobile graphics processor has 10,496 CUDA cores, 24GB of GDDR7 memory and 896GB/s memory bandwidth.

It also supports hardware ray tracing, DLSS 4 and Multi Frame Generation, giving the FA80G several features associated with current high-end PC gaming.

The desktop processor is equally important, as the 9950X3D is designed for high-performance gaming workloads rather than conventional mobile computing.

This mini PC also supports up to 128GB of DDR5 memory via two slots, while two M.2 2280 connectors support drives reaching 4TB.

FEVM separates processor and graphics cooling, assigning four heat pipes to each component across the device’s internal thermal arrangement.

The company rates full load noise at 42dB, with CPU temperatures below 85°C and GPU temperatures below 80°C.

For connectivity, this device includes HDMI 2.0, HDMI 2.1, two DisplayPort 2.0 outputs, USB-C, USB-A, an SD reader and dual 2.5GbE.

FEVM has the hardware for a serious M5 Ultra gaming comparison

In comparison to Apple’s top M5 Ultra Mac Studio, the FA80G takes a more conventional high-end PC approach to gaming hardware.

The M5 Ultra offers up to 36 CPU cores, 80 GPU cores, 512GB unified memory and 1.2TB/s memory bandwidth.

Apple also includes hardware-accelerated ray tracing, Neural Accelerators and a 32-core Neural Engine within its highest M5 Ultra configuration.

Those specs give the Mac Studio substantial graphics and compute resources for professional rendering, video production and artificial intelligence workloads.

For gaming, however, the FA80G CPU and GPU combo fits the Windows PC gaming environment more directly.

It also supports DLSS 4, ray reconstruction, frame generation and Multi Frame Generation across its RTX 50 series gaming hardware.

Apple supports advanced gaming technologies through Metal, MetalFX and its Game Porting Toolkit, including tools for bringing Windows games onto Mac.

That means the M5 Ultra is capable of modern gaming, but developers must still account for Apple’s Metal-based gaming environment.

The FEVM instead combines established Windows gaming hardware with Nvidia’s dedicated graphics stack and AMD’s gaming-focused desktop processor.

For a device specifically judged by gaming hardware, that makes the FA80G’s specification unusually aggressive for a 2.6-liter system.

The M5 Ultra remains a much larger compute platform in memory capacity and integrated resources, but the FEVM is built differently for gaming

The flagship configuration of FEVM's FA80G costs ¥22,299 (about $3,300), before adding memory and storage to the bare-bone system.

Via Videocardz

Samsung Galaxy S26 FE: Two-minute review

Picking the Samsung Galaxy S26 FE to be your next phone would be a good decision. But whether it’s the best decision is the chewier question.

As has been the case for the past several years, Samsung has launched one final version of its Galaxy S flagship phones just before summer is out (in the US and UK, at least). This “Fan Edition” device aims to provide a more streamlined experience, focused on camera capabilities and AI, at a more approachable price.

However, Samsung has this all backwards. The Galaxy S26 FE’s cameras range from OK to weak, and the AI tools will soon be available on older models. Plus, the price hike this year makes the phone a little less of a bargain.

Arguably, the real reasons you’ll want a Galaxy S26 FE are its big screen, fast charging, strong battery life, and surprisingly good gaming performance. But then, if those are your priorities, you may still be better off looking elsewhere.

Confused yet? Hopefully the rest of my Galaxy S26 FE review will clear things up.

Samsung Galaxy S26 FE review: Price and availability

• Starts at $699 /£699 / AU$1,149

The Galaxy S26 FE first went on sale on September 4, 2026. It costs $699 / £699 / AU$1,149 for the standard version, which is $50 / £50 / AU$50 more than the Galaxy S25 FE cost when it launched.

As a result, it means there’s a $200 / £180 / AU$400 difference between this phone and the cheapest full-fat Samsung Galaxy S26, or a $400 / £400 / AU$700 gap if you compare it to the similarly-sized Galaxy S26 Plus. However, those Galaxy S26 models have twice as much storage as the S26 FE by default, which makes direct comparison a little complicated.

Also, keep in mind that you get six free months of Google AI Pro (worth about $120 / £114 / AU$200 at current subscription prices) with the purchase of the S26 FE. It’s an offer worth taking advantage of for the 5TB of cloud storage capacity alone, let alone the extra Gemini powers.

Samsung Galaxy S26 FE pricing

US price

UK price

AU price

128GB

$699

£699

AU$1,149

256GB

$799

£799

AU$1,299

512GB

Not available

£949

AU$1,599

• Value score: 3.5 / 5

Samsung Galaxy S26 FE review: Specs

Before we continue, here’s a rundown of the Galaxy S26 FE’s hardware.

Dimensions:

161.6 x 76.9 x 7.4mm

Weight:

193g

Display:

6.7-inch FHD+   Dynamic AMOLED 2X

Resolution:

1080 x 2340

Refresh rate:

60/120Hz

Peak brightness:

1,900 nits

CPU:

Exynos 2500

RAM:

8GB

Storage:

128GB / 256GB / 512GB

OS:

Android 17 with One UI 9, 7 years of updates

Cameras:

50MP wide (f/1.8), 12MP ultra-wide (f/2.2), 8MP 3x telephoto (f/2.4)

Selfie Camera:

12MP front camera (f/2.2)

Battery:

4,900mAh

Charging:

45W wired, 15W wireless

Colors:

Blueberry, Graphite, Pistachio

Samsung Galaxy S26 FE review: Design

(Image credit: Richard Priday / TechRadar)

• No major updates from previous model other than new colors
• Worthy of the Galaxy S name in terms of looks and durability

Looks-wise, there are no major body changes between the Galaxy S26 FE and the S25 FE — not that Samsung really needed to make any.

You still have three camera lenses embedded directly into the back of the phone, then flat rails with rounded corners around the sides, leading to a big display with a single centered selfie camera on the front.

Unless you’re able to tell the S26 FE apart from a regular S26 by the color of the handset, the big telltales that you’re holding an FE are the shiny back panel, instead of a matte texture, and the larger chin bezel under the display.

(Image credit: Richard Priday / TechRadar)

Speaking of colors, Samsung’s three offerings are Graphite, Blueberry and Pistachio (pictured). The gray option will satisfy anyone who isn’t fussed about the color of their phone, or who knows they’ll be using a case to add some style to their new handset, but I do like both the blue and pale green options for the balance they strike between boldness and subtlety.

Like the other Galaxy S26 models, the S26 FE is rated IP68 for water/dust-resistance, which we expect from any flagship phone launching today. However, it uses Gorilla Glass Victus Plus on both sides to protect against scratches and smashes, which is a generation behind what you get on the S26 or S26 Plus.

• Design score: 4 / 5

Samsung Galaxy S26 FE review: Display

(Image credit: Richard Priday / TechRadar)

• Big and colorful, but lower-spec resolution and brightness
• Larger bezels than other Galaxy S phones

The 6.7-inch AMOLED display of the Galaxy S26 FE matches the size of the Galaxy S26 Plus’ display, but that’s where the similarities end. The FE’s panel only reaches an FHD resolution, rather than the Plus’ more detailed QHD, and peaks at 1,900 nits of brightness according to Samsung’s testing, rather than the 2,600-nit rating of the other S26 models.

Even though the S26 FE can’t deliver as much brightness or as many pixels as Samsung’s top-tier displays, the detail and the colors of the FE’s panel are still strong, and the screen remains fairly legible even out in what remains of the summer sun.

The bezels around the Galaxy S26 FE’s display are thicker than they are on the Galaxy S26 Plus or other premium Samsung phones, which is an understandable limitation given the FE’s lower price. It doesn’t interfere with whatever you want to watch on the screen, but it’s a difference you’ll notice if you have used other recent Galaxy S phones.

• Display score: 3.5 / 5

Samsung Galaxy S26 FE review: Software

(Image credit: Richard Priday / TechRadar)

• Samsung’s latest OS comes pre-installed
• Seven more generations of updates have been promised

The Galaxy S26 FE comes pre-installed with OneUI 9, Samsung’s take on Android 17. That actually puts it ahead of the other Galaxy S26 models, which currently can only access OneUI 9 as a beta version.

OneUI 9 gives the FE a lot of tools to play with. My FanCam — the video subject tracking tool which just debuted on the much more expensive Galaxy Z Fold 8/ Galaxy Flip 8 series — is here and works just as capably. That’s joined by other AI-powered editing features like Photo Assist and Audio Eraser.

(Image credit: Richard Priday / TechRadar)

Meanwhile, for productivity, you have Now Brief and its new custom cards option to give you even more personalized updates all in one place, more supported apps for Now Nudge, plus expanded cross-app functions and Gemini.

Perhaps most importantly for the Galaxy S26 FE, it comes with Samsung’s full seven generations of Android and seven years of security updates promise. That’s the same as other Galaxy S models, and one additional year over what the cheaper Galaxy A57 gets, which is crucial for the value proposition of the S26 FE.

• Software score: 4 / 5

Samsung Galaxy S26 FE review: Cameras

(Image credit: Richard Priday / TechRadar)

• Minor upgrades but identical sensors to previous model
• The main and selfie cameras are good, but the ultra-wide and telephoto cameras lag behind

The Galaxy S26 FE’s camera loadout has a 50MP main rear lens as its centerpiece. You also have 12MP ultra-wide and 8MP 3x telephoto cameras on the back, as well as a 12MP selfie camera embedded in the display. This is the same basic setup as the Galaxy S25 FE, and also what you'll find on the Galaxy S26 and Galaxy S26 Plus, except that the more expensive phones get a higher-resolution 10MP zoom camera instead.

Richard Priday / TechRadarRichard Priday / TechRadarRichard Priday / TechRadarRichard Priday / TechRadarRichard Priday / TechRadarRichard Priday / TechRadar

Our first example is a main camera shot of some brightly colored buildings in London’s Soho neighborhood. It’s a tricky one for any camera to balance due to the shaded shopfronts on the left side and the sky and brightly lit building in the center and right side. The Galaxy S26 FE still has some detail to offer in the shadowed brickwork, but Samsung’s strength in bright colors is on full display on the other half of this particular image.

Keeping with the main camera but looking out at the London Eye at night, we see another strong attempt from the Galaxy S26 FE. There’s some softness in the details, but the balance of colors between the lights and the night sky still makes for an appealing shot.

Looking across a sun-scorched park with the ultra-wide camera, we see a shot with plenty of scenery in it, with little to no distortion at the edges. The brightness and colors could use work, though, as even accounting for the patchy grass and overcast sky, things look overly drab.

The two shots of the church bell tower, one at 2x magnification using the S26 FE’s main camera and the other using the 3x telephoto camera, reveal how best to zoom in with this phone’s particular cameras. The in-sensor 2x zoom still keeps the subtleties of the colors and markings within the brickwork and foliage of the shot, thanks to the 50MP main sensor having plenty of pixels to spare even when zoomed in. However, while the telephoto camera has greater reach, the resulting image is darker and flatter, as you might expect from the lower-resolution sensor. It’s more of a trade-off than I’d like to see on a flagship phone’s zoom camera.

I’m normally a fan of Samsung selfies, but I’m not sure I like how the highlights are exposed in this particular shot. Samsung’s excellent portrait algorithm has applied the perfect level of bokeh to this image, however, cutting around my head, and especially my glasses, accurately and applying a gentle blur to the leaves behind.

Having gone through these samples, you’ll see that the Galaxy S26 FE’s photography potential is reasonable, at least for what are arguably the two most important cameras. For future generations of S FE, though, Samsung could do with upping its game if it wants to continue claiming this is a flagship product.

• Camera score: 3 / 5

Samsung Galaxy S26 FE review: Performance

(Image credit: Richard Priday / TechRadar)

• The Exynos 2500 chip is a good performer, especially for graphics
• 8GB of RAM is limiting for a premium phone in 2026

Here’s an area where I was pleasantly surprised by the Galaxy S26 FE. Samsung’s own Exynos 2500 chipset sits inside the S26 FE, rather than the Snapdragon 8 Elite Gen 5 chip that basically all the other top Samsungs of 2026 are using. But that hasn’t stopped the S26 FE from turning in some decent scores.

As the table below reveals, the Exynos 2500 lags behind slightly in CPU single-core results, but is on par with or ahead of the pack on the rest of the tests. It’s a long way out in front on the Solar Bay benchmark, no doubt thanks to the Exynos’ GPU, which is well-suited to the ray-tracing graphics that this test focuses on.

Samsung Galaxy S26 FE benchmark scores

Samsung Galaxy S26 FE

Geekbench 6 single-core

2,361

Geekbench 6 multi-core

7,831

3DMark Wild Life Extreme Stress Test best

5,116

3DMark Wild Life Extreme Stress Test low

3,456

3DMark Solar Bay

10,526

Bearing only 8GB of RAM compared to the 12GB in other premium Samsung phones, the Galaxy S26 FE does struggle with multitasking or intensive apps. While you can still game on this phone happily, it’s not a handset you’ll want to do anything more practical on than standard browsing and messaging.

• Performance score: 4 / 5

Samsung Galaxy S26 FE review: Battery

(Image credit: Richard Priday / TechRadar)

• The same battery/charging hardware as the more expensive S26 Plus
• Battery life day-to-day is good, while charging is plenty fast enough

Samsung offers the same 4,900mAh battery capacity in the Galaxy S26 FE as it does in the Galaxy S26 Plus. However, according to Samsung’s official battery estimates, the S26 FE doesn’t offer the same battery life, managing 29 hours of video playback compared to the S26 Plus’ 31 hours or the S26’s 30 hours.

For me, I used about 40% of the phone’s battery during a typical day’s use. This included some work, some photography, and some social media scrolling and video watching. The Galaxy S26 FE could definitely survive two days of regular use before needing to charge, as you would hope from a phone with this size of battery.

The Galaxy S26 FE is still equipped for 45W wired charging and 15W wireless charging. That’s the same wired speed as the S26 Plus and Ultra, but the wireless charging is slower, since the FE is not Qi2 compatible.

Using a 45W charger, Samsung estimates a 69% refill for the S26 FE in half an hour. In my own test, I found the S26 FE managed to fill to 43% in 15 minutes and to 100% in 58 minutes. Considering the battery life this phone can manage, a complete refuel in under an hour and getting over two-fifths full in a quarter-hour isn’t bad at all, even if faster options are out there.

• Battery score: 4 / 5

Should you buy the Samsung Galaxy S26 FE?Samsung Galaxy S26 FE scorecard

Attributes

Notes

Rating

Value

The cheapest Galaxy S26 model, although not by much post-price hike.

3.5/5

Design

Looks like any other Galaxy S26, except with some fun, unique colors and slightly less durable materials.

4/5

Display

Big and colorful, but lacking detail.

3.5/5

Software

Samsung’s best features come pre-installed, but will soon be available on older phones too.

4/5

Camera

The main and selfie cameras produce strong photos, but the phone’s let down by its ultra-wide and telephoto lenses.

3/5

Performance

A below average RAM capacity is understandable for the price, and the phone still plays games really well.

4/5

Battery

Less efficient than other S26 models, but still long-lasting and fairly fast-charging.

4/5

Buy it if…

You want the big premium Samsung experience for the smallest possible price
The Galaxy S26 FE is the cheapest Galaxy S26, and joint second largest. If you want a big display and long battery life from Samsung for less, then this is the model for you.

Exclusive software and AI features don’t interest you
You do get some excellent tools and toys to play with within One UI 9, but since this update will come to older Samsungs soon enough, this isn’t a phone to buy for this reason alone.

Your performance needs are about average
If you like to play games on your phone, or know you’ll spend long days away from a charger, the S26 FE will suit you well. It doesn’t offer the very best in either power or battery life, but arguably it’s a case of diminishing returns if you spend any more than this.

Don’t buy it if…

You care about photography
Taking snaps with the main and selfie cameras on the S26 FE is easy and results in great photos. But if you want more detail or know you like taking ultra-wide or zoom shots, there are many other phones that will suit you better.

You want the best long-term value
While the S26 FE is cheaper to buy right now, we’d recommend looking at one of Samsung’s more expensive S26 models, or those of another Android brand, if you’re hoping to get as many happy years out of your new phone as possible.

If you’re willing to spend even a few dollars/pounds more
The Galaxy S26 FE sits just below the starting price of typical flagship phones. That means you could save yourself 100 bucks or so, or you can spend the extra and unlock the next tier of hardware and features that could provide far more value to you.

Samsung Galaxy S26 FE review: Also consider

Samsung Galaxy S26 Plus

It’s quite a bit more expensive, but as an investment, it may be worth spending the extra money over the Galaxy S26 FE’s price to secure a phone that outperforms it on every important metric.

Read our full Samsung Galaxy S26 Plus review

Samsung Galaxy S26

Costing only two hundred bucks more than the S26 FE, the standard S26 gives you the premium Samsung experience at the best possible price, even if the display is quite a bit smaller.

Read our full Samsung Galaxy S26 review

Samsung Galaxy S26 FE

Samsung Galaxy S26 Plus

Samsung Galaxy S26

Price (at launch):

$699 / £699 / AU$1,149

$1,099 / £1,099 / AU$1,849

$899 / £879 / AU$1,349

Dimensions:

161.6 x 76.9 x 7.4mm

158.4 x 75.8 x 7.3mm

149.6 x 71.7 x 7.2mm

Weight:

193g

190g

167g

Display:

6.7-inch  Dynamic AMOLED 2X

6.7-inch Dyanmic AMOLED 2X

6.3-inch Dynamic AMOLED 2x

Resolution:

1080 x 2340

3,120 x 1,440

2340 x 1080

Refresh rate:

60/120Hz

120 Hz adaptive

120 Hz adaptive

Peak brightness:

1,900 nits

2,600 nits

2,600 nits

Processor:

Exynos 2500

Qualcomm Snapdragon 8 Elite Gen 5 for Galaxy

Qualcomm Snapdragon 8 Elite Gen 5 for Galaxy

RAM:

8GB

12GB

12GB

Storage:

128GB / 256GB / 512GB

256GB, 512GB

256GB, 512GB

OS:

Android 17 with One UI 9, 7 years of updates

Android 16 with One UI 8.5, 7 years of updates

Android 16 with One UI 8.5, 7 years of updates

Cameras:

50MP wide (f/1.8), 12MP ultra-wide (f/2.2), 8MP 3x telephoto (f/2.4)

50MP main (f/1.8), 12MP ultra-wide (f/2.2), 10MP telephoto with 3x optical zoom (f/2.4)

50MP main (f/1.8), 12MP ultra-wide (f/2.2), 10MP telephoto with 3x optical zoom (f/2.4)

Selfie cameras:

12MP front camera (f/2.2)

12MP (f/2.2)

12MP (f/2.2)

Battery:

4,900 mAh

4,900 mAh

4,300mAh

Charging:

45W wired, 15W wireless

45W wired, 15W wireless

25W wired, 15W wireless

Colors:

Blueberry, Graphite, Pistachio

Cobalt Violet, Sky Blue, White, Black, Silver Shadow, Pink Gold

Cobalt Violet, Sky Blue, White, Black, Silver Shadow, Pink Gold

How I tested the Samsung Galaxy S26 FEWhy you can trust TechRadar

☑️ 100s of smartphones reviewed
☑️ 15 years of product testing
☑️ Over 16,000 products reviewed in total
☑️ Nearly 200,000 hours testing tech

I reviewed the Samsung Galaxy S26 FE over the course of two weeks as a work phone, using it for writing stories, answering emails and messages, watching video and playing games. I did this both at home and on the move with a SIM card installed.

Specific tests included running Geekbench 6 CPU benchmarks and 3DMark graphics benchmarks, a charging speed test using a compatible third-party charger, the taking of hundreds of photos across all of the S26 FE’s cameras to test image quality, and the careful monitoring of battery drain during use to figure out how reasonable the manufacturer’s battery life promises are.

Read more about how we test

First reviewed: September 2026

• The Bentley Torcan is a high-performance SUV that can reach 375 miles on a charge
• The most powerful models will boast 876bhp
• The debut EV will share parts with Porsche Cayenne Electric

Bentley has been teasing its debut all-electric model for months now, revealing tasty morsels about its orchestral synthesized engine notes, and futuristic styling inspired by the EXP 15 Concept Car, but we finally have the full details.

Based on Porsche’s PPE architecture (they all belong to the Volkswagen Group, after all), the upcoming Torcal will measure five meters long, making it slightly shorter than a Bentayga, and will borrow many of the chassis, suspension and powertrain components from its Porsche sibling.

Bentley doesn’t like too many mentions of the shared underpinnings, so has made a big deal about its “bold new design language, rooted in Bentley’s DNA”.

Expect an imposing illuminated front grille, sculpted surfaces and powerful rear haunches.

It looks as if Bentley’s idea of an electrified SUV just ate a Continental GT and hasn't digested it yet, but that’s no bad thing. It is the antithesis of Ferrari’s approach with the Luce — that’s to say Torcal looks and feels reassuringly Bentley.

BentleyBentleyBentleyBentley

As for the powertrain, the dual-motor setup makes the Torcal the most powerful Bentley ever, with a whopping 876bhp in the range-topping S model.

That’s not quite the 1,140hp of the monstrous Porsche Cayenne Turbo Electric, but will be sufficient for a 2.8-second sprint to 60mph from rest in the sportier S version — enough to embarrass many supercars.

Bentley says it wants the Torcal to be used “year-round” so offers a massive 113kWh battery pack that delivers a range of up to 375 miles.

Better still, it can be charged at speeds of up to 400kW, meaning a 10-80% top-up session takes less than 20 minutes, potentially.

Analysis: the sound of the future

(Image credit: Bentley)

The British marque has scrapped its plans to go all-electric by 2030, instead opting to offer a selection of powertrains for customers.

After all, it is synonymous with booming V8 and V12 engines, which have historically delivered restrained but potent soundtracks to many models.

Not wanting to lose this USP in its debut EV, Bentley has turned to real instruments — drums, viola and bass guitar — to create what it refers to as the “Bentley Dynamic Symphony”, which is supposed to represent the subtle variations and tiny imperfections of a big V8 engine.

This “human-generated” soundtrack will provide aural feedback to the driver as the revs rise, becoming louder in Sport mode.

Apparently, even the indicator note stems from a recording of hammers striking leather in the Bentley workshop, while the touchscreen feedback noises are taken from the sound of a crystal glass.

Naturally, the interior is festooned with physical knurled controls and dials, while a special Comfort Mode within the infotainment offering delivers fewer distractions, while the option of a digital detox sees all but the most essential information hidden to allow attention to move away from screens.

According to sources, prices start at £173,300 (around $230,00/AU$326,000) for the Torcal or £199,300 (around $263,000/AU$375,000 for the Torcal S, and first deliveries are expected in Summer 2027.

The pharmacy chain denies wrongdoing, but is settling anyway.

Motorola previewed the Signature 27 with a 200MP periscope camera, Snapdragon 8 Elite Extreme Gen 6, and seven OS updates. Price and launch date remain unknown.

The post Motorola Unveils Signature 27 Camera and Chip Details Ahead of Launch appeared first on TechRepublic.

• A chatbot invented entirely false cargo details about a Chinese vessel
• Armed troops prepared to board the ship before anyone checked the source
• Military aircraft were already active when officials caught the error

A false intelligence assessment nearly triggered a US military invasion of a Chinese vessel after AI-assisted analysis wrongly identified its cargo.

The report circulated during the Iran conflict and prompted preparations to intercept the vessel before officials discovered serious problems with its underlying information.

Sources told CNN that armed personnel were preparing to board while military aircraft supported preparations, creating a potentially dangerous confrontation between major powers.

AI-assisted analysis produced the mistaken assessment

The episode began when an analyst examined intelligence concerning the ship's manifest and used a chatbot to interpret information.

Sources said the system combined publicly available material with classified signals information before producing an incorrect assessment about the vessel's cargo.

The analyst subsequently used AI again to prepare the findings in a standard intelligence format that was distributed among military officials.

Officials only examined the underlying information shortly before the planned operation, revealing that the reported cargo identification was wrong.

The source described the assessment as “entirely false,” while CNN reported that another source said it “almost started a war.”

The exact material that the chatbot misidentified has not been disclosed publicly, leaving important details about the original error unavailable.

The Pentagon and the military command responsible for Pacific special operations did not provide comments on the incident.

The episode comes as the US military expands its use of AI tools across intelligence, operations, administration, and battlefield decision-making.

In January, Defense Secretary Pete Hegseth announced an AI strategy intended to accelerate deployment across military missions and organizational functions.

“We will unleash experimentation, eliminate bureaucratic barriers, focus our investments and demonstrate the execution approach needed to ensure we lead in military AI,” Hegseth said.

The strategy includes programs for intelligence work, operational planning, enterprise functions, and wider access to generative AI systems.

The department later said more than 1.3 million personnel had used GenAI.mil, generating tens of millions of prompts within five months.

The incident exposes a difficult verification problem

The reported mistake involved more than an inaccurate chatbot response because its output entered a process used for consequential military decisions.

Once incorrect information appeared in a formal assessment, officials receiving the document had to determine whether its underlying evidence supported the conclusion.

There is no information on which verification procedures were applied before the assessment reached personnel preparing the maritime operation.

It also remains unclear whether the analyst knew that the chatbot had introduced an incorrect interpretation before the report circulated.

According to CNN, different military organizations use different AI systems, while officials have described varying practices for checking information produced by them.

That fragmented environment makes it difficult to identify a single verification process governing every AI-assisted intelligence product across the military.

AI tools can make military analysis faster and easier, yet their potential for significant errors requires strict human verification before operational decisions.

That need becomes more important when intelligence concerns a rival country, where an incorrect assessment involving a military vessel could trigger retaliation.

A mistaken attack could then escalate tensions between the countries involved, potentially producing a wider armed confrontation or prolonged political and military conflict.

The Bambu Lab H2D sits right at the top of our 3D printer guide - a versatile machine that performed exceptionally well in our tests. Now, the flagship machine is on sale.

• Best for: Experienced makers and small businesses who want a large dual-nozzle machine for engineering-grade filaments.

• The deal: The Bambu Lab H2D AMS Combo is now $1749 (was $1999) at Bambu Lab, while in the UK, the unit is £1449 (was £1649). The standalone and Full Laser Combo units are also on sale.

• Why we like it: Calling this "one of the most impressive 3D printers on the market," we scored this 3D printer 4.5 stars in our review. It's effectively a compact multifunctional manufacturing hub. Precise and fast, it excels at multi-material printing.

Today's best 3D printer deal

H2D AMS Combo 3D Printer: was $1999 now $1749
The Bambu Lab H2D features dual nozzles, a 350 x 320 x 325mm build volume, 350°C all-metal hotend, 65°C heated chamber, 1,000mm/s toolhead speed, closed-loop extrusion and built-in air filtration. There's also a 5-inch 720p touchscreen, and multiple built-in cameras for monitoring.

In the UK: now £1449 (was £1649)View Deal

Why we recommend it

In his 4.5-star review of the H2D, our expert Alastair found that "3D print quality is as good as it gets." He added that, "when it comes to print quality, there's no doubt that the H2D matches the excellent X1 Carbon and holds its own against the current crop of high-end 3D printers."

The Full Laser Combo is also in the sale, and for that, he found the "laser engraving is excellent, and the cutting/plotting is right in line with other dedicated machines."

You'll find the Bambu Lab H2D at the top of our roundup of the best 3D printers we've tested.

Price context & historical value

The $250 reduction takes 12.5% off the H2D AMS Combo's $1,999 regular price, bringing it down to $1,749. That's a substantial cash saving on an expensive printer, but it's not the cheapest we've seen it. In Bambu Lab's 4th Anniversary Sale in June/July 2026, the H2D AMS Combo was available for $1,699.

Should you buy it?

✅ Buy the Bambu Lab H2D AMS Combo if...

You want a versatile 3D printer that can accommodate ambitious projects and different filament types, including PLA, PETG, TPU, ABS, ASA, PC and PA, plus a selection of carbon- and glass-fiber-reinforced types.

❌ Skip the Bambu Lab H2D AMS Combo if...

You only intend to make occasional small PLA prints. The H2D's size, capabilities, and price are difficult to swallow if you aren't likely to take advantage of its more advanced features.

Also consider

H2D Laser Full Combo: was $2549 now $2149
This version of the H2D 3D Printer is the Laser Full Combo (you can tell thanks to the green glass). The 10W model is down from $2,549 to $2,149, saving you $400, while the more powerful 40W model drops from $3,199 to $2,699, saving you $500. Both are roughly 16% off.

In the UK: now £1749View Deal

This version of the H2D 3D Printer comes without the AMS Combo.

In the UK: now £1289 (was £1449)View Deal

• Shop all 3D printers at Bambu Store

RAM prices have become increasingly painful during the 2026 memory crisis, making discounts on higher-speed DDR5 kits welcome. The 32GB TeamGroup T-Force Vulcan ECO offers DDR5-6400 speeds and support for both Intel and AMD performance profiles.

• Best for: Desktop builders who want 32GB of relatively fast DDR5 memory with support for both major performance-profile standards.

• The deal: The TeamGroup 32GB T-Force Vulcan ECO DDR5-6400 kit is now $429 (was $479) at Newegg when you use code TEF395.

• Why it matters: DDR5-6400 provides much more theoretical bandwidth than slower DDR5 standards. You'll need compatible hardware and the appropriate XMP or EXPO profile enabled to reach the 6,400MT/s speed though.

Today's TeamGroup T-Force Vulcan ECO DDR5 deal

T-Force Vulcan ECO DDR5-6400: was $479 now $429
The Team T-Force Vulcan ECO includes two 16GB DDR5-6400 desktop modules with CL38 latency, 38-46-46-84 timings, 1.35V operation, silver heat spreaders, Intel XMP 3.0 and AMD EXPO performance profile support. For the full discount, use code TEF395View Deal

Why we recommend it

The T-Force Vulcan ECO gives you two 16GB DDR5 modules running at up to 6,400MT/s, with CL38 latency and 38-46-46-84 timings at 1.35V. It supports for both Intel XMP 3.0 and AMD EXPO, giving you flexibility when choosing a platform.

The modules include heat spreaders to help manage temperatures, and their silver finish is a stylish alternative to the usual black and white designs that dominate the desktop memory market.

Price context & historical value

Applying promo code TEF395 drops the price from $479 to $429, saving you $50, or around 10%. That's not an enormous reduction, but any money off is welcome in today's expensive memory market. This kit has been slightly cheaper at Newegg in the past though. Not so long ago it was available to buy for $415.99.

Should you buy it?

✅ Buy the 32GB TeamGroup T-Force Vulcan ECO DDR5 kit if...

You're building or upgrading a DDR5 desktop and want a fast 32GB kit with support for both major performance-profile standards. At 6,400MT/s, the Vulcan ECO offers considerably more memory bandwidth than entry-level DDR5-4800 or DDR5-5200 kits, although the real-world benefit will depend on your setup.

❌ Skip the 32GB TeamGroup T-Force Vulcan ECO DDR5 kit if...

Your current PC already has enough memory for everything you do, or your motherboard and processor can't take advantage of DDR5-6400. Given today's RAM prices, there's little reason to spend $429 unless you need the upgrade.

Also consider

32GB Vengeance DDR5-5200 White: was $508.99 now $407.99
The white Corsair Vengeance kit includes two 16GB UDIMMs rated for 5,200MT/s, CL40 latency and 1.25V, featuring aluminum heat spreaders, Intel XMP support and iCUE-controlled onboard voltage regulation.View Deal

32GB Vengeance DDR5-4800: was $505.99 now $399.99
This 32GB Vengeance DDR5 kits combines two 16GB modules with speeds up to 4,800MT/s, CL40 latency and a compact design optimized for compatible Intel motherboards.View Deal

Qualcomm’s PC chips were developed with the company’s phone DNA. Two generations in, we’re seeing payoffs.

CyberLink has long been one of our favorite alternatives to Adobe's creative suite, and currently there's a big sale on standalone alone apps as well as CyberLink Director Suite, which bundles the photo, video, and audio editor software alongside an audio editor and large royalty-free stock library.

• Best for: Content creators who regularly edit video, photographs and audio and want their main creative applications bundled into one subscription.

• The deal: CyberLink Director Suite is now $110 (was $150) for an annual subscription, and £95 (was £130) in the UK. Standalone apps are also discounted in the US here (click here) and in the UK here.

• Why we like it: CyberLink's creative tools have consistently reviewed well during our tests, offering an excellent all-rounder entry-point into video and photo editing. They're surprisingly advanced but beginner-friendly, and this is a good way to get the full suite.

Director Suite: was $149.99 now $109.99
This creative suite combines PowerDirector 365, PhotoDirector 365 and AudioDirector 365, providing AI video, photo and audio editing alongside integrated royalty-free Getty Images and Meta stock content.

In the UK: now £95 (was £130)View Deal

Why we recommend it

Director Suite combines three powerful creative tools. PowerDirector 365 handles video production, PhotoDirector 365 covers photo editing and graphic design, and AudioDirector 365 provides tools for audio restoration, recording and mixing.

All three core apps have reviewed well in our tests.

In his 4-star review of PowerDirector, our expert Steve said it "remains one of the best video editing software for beginners, as well as intermediate editors." He gave PhotoDirector 4.5-stars and called it a "powerful multi-faceted beast, which not only allows you to organize and improve the quality of your photos, but also offers Photoshop-like compositing tools, as well as numerous generative AI features."

In her 4-star review, our audio expert Angel said AudioDirector "goes from strength to strength using powerful AI tools to separate elements giving even total novices to recording control over their material."

The applications are designed to work together, so projects can be passed between them without repeatedly rendering files. There's plenty of AI features too.

Price context & historical value

At $109.99 a year instead of $149.99, you're saving $40 on Director Suite, equivalent to 27% off. The annual subscription works out at just $9.17 per month, compared with $29.99 if you pay monthly, which would add up to $359.88 over a year — $249.89 more.

UK buyers can save 27%, with the price of Director Suite reduced from £129.99 to £94.99, equivalent to just £7.92 per month.

For more choices, check out our roundups of the best photo editors, and best video editing software.

Should you buy it?

✅ Buy CyberLink Director Suite if...

You regularly work with video, photographs and audio and would prefer to have an integrated collection of creative applications than subscribe to different products. Director Suite also provides access to millions of royalty-free videos, images and music tracks from iStock by Getty Images and the Meta Sound Collection.

❌ Skip CyberLink Director Suite if...

You only need one type of editor. If almost everything you create is video, for example, paying for a suite containing dedicated photo and audio tools may not be necessary.

• US: Shop discounted creative software at CyberLink
• UK: See the full sale at CyberLink

When it comes to personal robots, Mondo's Beni camera robot might be the perfect mix of cute and utility, with an emphasis on cute.

As soon as I laid eyes on the pint-sized, two-wheeled bot, I was smitten. It darted quickly from place to place and, as Beni turned its square head in my direction and 'looked' at me with its two large and luminous eyes, I felt it peering into my lonely soul. Finally, a robot companion that wouldn't terrify me with grabby, slow-moving hands, or crowd me with uncanny valley legs and torsos.

Unveiled a few months ago by Mondo Robotics and set to arrive in consumer hands in October, the self-balancing Mondo Beni was initially intended as a child's toy, but it had features — in particular a 4K 60fps camera that can follow you around — that appealed to creators. So Mondo pivoted, and now the $799 rolling robot is marketed as "your first camera robot".

In addition to the camera, which can also shoot 1080p 120fps slow motion, Beni features a pair of sensor cameras that provide a 150-degree field of view for tracking and obstacle avoidance. I watched as Beni's handlers used the iOS app to select subjects for Beni to track, and then Beni followed them around like a very obedient puppy dog. One Mondo exec described Beni as "a ground drone".

Beni is surprisingly resilient. When someone kicked the robot over, it instantly righted itself back onto its two rubber wheels as it shook its blocky head and made a little noise.

(Image credit: Lance Ulanoff / Future)Programming cuteness

One of the things that makes Beni so darn cute is all the canned (Beni does not learn) animations and audio. It has about a dozen of each, and they give the little white and orange bot a distinct personality. There's a sort of programmed indomitable spirit; nothing can keep Beni down or stop it.

It can effortlessly jump (up to 10 inches), flip, and backflip. For now, those moves are triggered either via the app or a wearable remote control with a super-easy-to-use joystick. Mondo would eventually like Beni to jump over obstacles as it sees them.

I used the intuitive wearable remote to drive Beni and, with a press, make the little robot flip. There's also a big red button to stop Beni or, with long presses, have it squat down or stand back up.

Lance Ulanoff / FutureBeni with Mondo Marketing Director Cody SkeneLance Ulanoff / FutureLance Ulanoff / Future

Beni starts with 32GB of onboard storage for collecting video of you as it tracks you on a walk or run (up to 17.9mph), or as it follows your pet around (assuming the pet doesn't attack it or run away in horror). There's also a built-in microSD card slot for additional storage.

You can take good-length walks or runs with Beni, because it runs (or rolls) for an hour and a half on a charge. There's a removable battery pack on the back with light indicators to show you Beni's life… er… juice level.

Beni ships with a pair of relatively smooth rubber wheels, but you can buy the combo pack and get some accessories, including knobbier off-road wheels that are better equipped to handle dirt roads and hills.

Lance Ulanoff / FutureLance Ulanoff / FutureLance Ulanoff / FutureLance Ulanoff / FutureBeni arrives ready for you

(Image credit: Lance Ulanoff / Future)

One of the things I like about Beni is that it's not like a robot kit. The wheels, for instance, are easy to pop on and off, and nothing about Beni looks like it was made from an old Erector Set.

And, for better or worse, there is no dev kit. This is in contrast to the equally cute Pollen Robotics Microduck, another white and orange bipedal robot that can hang out with you (and use roller skates), but it's a bit more DIY than SMF (So Much Fun). Where Beni looks polished, the $399 Microduck looks a little more hand-built.

Beni can hear and follow some simple commands. The animations — a head turn here, a shake there — help it look like it's actually listening to you.

(Image credit: Lance Ulanoff / Future)

Beni's innate adorability might help overcome the obvious privacy concerns. After all, the robot is no taller than an average chihuahua, and its camera is at ground level. To film you as it goes, Beni will be looking up. This might make some people around you uncomfortable.

Maybe keep Beni to your solitary walks or runs, or make sure everyone around you knows that those cute Beni eyes are an indication that it might be filming.

I picked Beni up and found it surprisingly hefty, about the weight of an overfed puppy. It's cute to watch but not at all cuddly to hold. In fact, company execs told me that while there are lots of warnings about picking up Beni, the robot has no way of knowing if your fingers gets caught in, say, one of the joints. It might be best to keep Beni away from small children.

All this capability, power, and cuteness comes at a price. Beni starts at $799 in the US (it will ship to the UK, Canada, Australia, New Zealand, France, Germany, Italy, and Spain, but pricing for those regions has not been set) and starts shipping in October. It's not the ultimate robot, but Beni has the potential to captivate on looks and capability alone.

@techradar

♬ original sound - TechRadar

• The environment, energy costs, quality of life and taxes are all concerns
• Americans are largely split about whether data centers are even good news for jobs
• Negative sentiments are rising across the board regardless of political background

With the US midterm elections around the corner and AI data centers increasingly becoming part of political discourse, new Pew Research Center reporting has revealed Americans' attitudes toward these facilities have become even more negative as 2026 continues.

Surveying more than 10,500 US adults, the study found only 4% thought data centers are a positive thing for the environment, home energy costs and local residents' quality of life.

On the flip side, more than half (54%) thought they were 'mostly bad' for the environment, up from 39% in January 2026, around half a year prior, showing a marked increase just in a short space of time.

US citizens are increasingly turning against AI data centers

In fact, Americans believe US data centers are worse for home energy costs (50%), quality of life (49%) and local tax revenue (21%) than before.

And with hyperscalers all promising to add hundreds or, sometimes thousands, of operational jobs on top of thousands of temporary construction jobs locally, residents are even starting to disbelieve these promises. Interestingly, the proportion of nationals perceiving data centers as mostly bad for local jobs rose nine percentage points, from 15% at the beginning of 2026 to 24% halfway through the year.

Perceptions are now mostly split, with 24% seeing projects as mostly bad for jobs and 22% seeing them as mostly good.

This still leaves a considerable proportion of Americans – 18% to 32% depending on issue – uncertain about the impact of data centers.

Opinions are shifting

Today, three in five US citizens say they'd either be not too comfortable, or not comfortable at all, with a new data center operating in their area. Rural inhabitants are more likely than before to have negative perceptions, likely with more projects being approved for extra-urban locations where space is at less of a premium.

The study also delves into the political implications, confirming that Americans are increasingly unsettled by data centers right across the board, be it by age, gender, education levels or political party. While Democrats are more likely overall to have negative feelings, supporters of both major parties have actually become more unhappy throughout 2026.

Arguably one of the biggest findings of this report isn't necessarily where the split lies and whether Americans are worried about environmental, social or economic impacts – it's the sheer pace that negative sentiments have grown in little more than just half a year.

All of this is, of course, playing out during an important time for the US politically.

While President Trump continues to advocate for AI and data center buildouts in an effort to be a global leader, more and more states and regions are imposing temporary bans to reassess legislation and local impacts, including Texas itself – the home of OpenAI's massive Stargate project.

Google Vids lets personal accounts create AI video clips for free with Gemini Omni 1.1 Flash. Learn about the six-generation monthly limit and desktop access.

The post Google Vids Adds Free AI Video: What You Get and What’s Limited appeared first on TechRepublic.

The AI giant found out about the hack in August, but didn’t send a message until September.

Enjoy fast and reliable mobile internet for less when you travel to Japan and use our code.

One of the advantages of having written about tech for as long as I have is that you start to notice when the same ideas come around decade after decade, in different shapes and sizes. Especially the bad ones.

12 years ago Google invented something called Google Glass — a set of glasses complete with a camera so you could record what you were looking at — and it was so roundly hated by everyone that we all breathed a collective sigh of relief when they went away.

But now it's 2026 and smart glasses are back, powered by AI, and tied in with a world-famous fashion brand, but fundamentally they still have exactly the same problem, which still hasn't been fixed: they enable people to record other people without their consent.

The problem is that none of this should have come as a surprise. The privacy risks of putting a camera on somebody's face weren't merely predictable — they were predicted. Women were raising concerns about men using Google Glass to record them more than a decade ago.

And yet here we are again.

The curious case of Google Glass

(Image credit: Shutterstock / Hattanas )

Let’s remind ourselves what happened the first time that smart glasses arrived.

Google Glass arrived in 2013. It puts a small heads-up display and camera on the wearer's face, allowing them to take photos and record video from their point of view. Almost immediately, privacy became one of the defining controversies around it.

By 2014, women were explicitly raising issues of abuse. Wired reported that women were particularly interested in technology that could detect and block Glass because of concerns about men covertly recording them in nightclubs. "Even if they didn’t know if the device was recording, they felt threatened by its presence", said Oliver, the creator of the detection device.

Google Glass wearers were being called “Glassholes,” businesses were banning the device, and Google itself was publishing etiquette guidance because of complaints about people creepily filming others without permission.

By 2015, consumer Glass effectively died. Privacy wasn't its only problem — its price, awkward appearance, and lack of compelling everyday uses all played a part — but by January 2015 Google had stopped selling the consumer Explorer Edition. Glass survived as an enterprise product before Google finally discontinued the Enterprise Edition in 2023.

Smart glasses didn’t die with Google Glass. They had simply retreated to a lab to lick their wounds and wait for that vital next piece of technology that would make them relevant again. That technology was AI. In 2023, AI-powered smart glasses, partnered with leading fashion brands, started to emerge, with Meta leading the charge with its Ray-Ban Meta glasses.

This time they looked much more like ordinary glasses. Meta had also found the compelling use cases that Google Glass struggled to provide: its glasses could use AI to answer questions about what you were looking at, give you directions, and translate signs and conversations. But they could also still record what, or more importantly, who, you were looking at without needing that person's consent.

It’s like women don’t exist

(Image credit: Shutterstock / Alexander Fedosov)

Frankly, I don’t think Meta should have released them with that ability, but here we are. It’s like we’ve learned nothing from the Google Glass debacle. Almost as soon as the next generation of smart glasses emerged they were being used to harass women and record them without their consent or knowledge.

Researchers at the University of Sydney have been studying hundreds of social-media videos involving smart glasses and found men using them to approach and record women without their knowledge, including at gyms, workplaces and beaches. Australia's eSafety Commissioner says the technology can amplify gendered harms because recording can be almost undetectable, and specifically highlights pickup-artist content in which women often don't know they're being filmed.

The researchers analysed 350 publicly available Instagram videos from 2023–2026. In a subset of covert POV videos, around 60% of interactions were classified as potential harassment. And in about 43% of the videos, women were subjected to further abusive online commentary, sometimes exposing personal information.

These are real victims. CBS interviewed a woman who discovered she'd been secretly recorded only after the resulting video went viral, attracting more than 200,000 views. Her reaction said it all: “I had no say.”

And France has now opened an investigation involving alleged sexual harassment and discreet smart-glasses recording.

I don't think these products were designed to harm women. That's almost the point. Too often, the technology industry seems to design products around what the person using them can do, without giving enough consideration to the person standing on the other side of the technology. And when the potential for abuse disproportionately affects women, that's a particularly serious blind spot. Women have historically been underrepresented in the data and testing used for product design, with designers still describing the male body as the implicit default.

Meta clearly did think about privacy. The glasses have a recording LED specifically designed to tell people nearby when the camera is being used. But if your safeguard against covert recording is a tiny light on the front of the glasses — one that people have found ways to conceal or disable — is that really enough?

Did it consider what happens when a man wearing an almost invisible camera approaches a woman who's alone?

Meta's audio-only smart glasses

Meta Ray-Ban Audio smart glasses. (Image credit: Meta)

Yesterday at Meta Connect, Meta announced a new version of its smart glasses called the Ray-Ban Meta Audio glasses, which don’t have a camera. They got barely a mention in Zuckerberg’s keynote, but I wonder if they were released because of the existential threat to its business that the now commonly used “perv glasses” moniker poses.

Ray-Ban Meta Audio has no camera. You can still listen to music, make calls, translate conversations, and talk to Meta's AI, but the person sitting opposite you no longer needs to wonder whether they're being secretly recorded.

Meta hasn't said the Audio glasses exist because of the privacy backlash, and removing the camera inevitably removes some genuinely useful features too. For visually impaired people in particular, a camera combined with AI can be transformative.

But Ray-Ban Meta Audio proves something important: smart glasses don't have to contain a camera.

Twelve years ago, women were already warning that putting inconspicuous cameras on people's faces could enable men to record them without their knowledge, and today researchers are documenting precisely that behavior.

Google Glass users became known as “Glassholes” because people hated that the person they were talking to was recording them. Twelve years later, we're having the same argument about “perv glasses.”

That's what makes me angry. This wasn't an unforeseeable consequence of some revolutionary new technology. We'd already run this experiment and realized that these should never have got out of the lab in the first place.

• Adobe Premiere is now free on Android, following its earlier roll-out to iPhone
• Premiere offers everything you can get on Canva, and more, without the subscription fees
• Compatible with Android 13 and later and requiring 5GB of RAM, it is unlikely to run on most Chromebooks, and is without advanced desktop features

Adobe has launched its flagship Adobe Premiere software as a mobile app on Android, and has made it free to use. After its previous release on iPhone in 2025, the stripped down take on Premiere has proved to be a smart move from Adobe, attracting users away from upstart competitor Canva.

Like Canva, Adobe Premiere for mobile has AI features. While free to download and use the app, you’ll still need to pay for the AI credits to use those features, but this will likely prove more affordable than the monthly plan of the desktop version.

A standard Adobe Premiere Pro subscription on desktop will typically cost £21.98/$22.99 a month with an annual plan, which makes the new Adobe Premiere for Android an attractive option if you don’t need the full set of features. There’s a bonus, too: if you use a modern Chromebook with Google Play installed, you should be able to use this version of Adobe Premiere.

Is it really Adobe Premiere?

While Adobe Premiere Rush is already available on Android, this is a pretty basic take on the video editing tools now in demand by content creators. This new free Adobe Premiere – which doesn’t have the same features as the desktop original – is a more accomplished version than Adobe Premiere Rush, with multi-track timeline, audio enhancements, 4K video, and more.

Adobe is generally better value than Canva, and the company’s decision to release feature-packed mobile versions of Adobe Premiere on iPhone and now Android underlines how seriously it takes the competition. While Adobe has a massive market share in the creative software market, Canva’s small 4% is backed by a more active user base.

(Image credit: Future)

Both tools offer creative AI, but Adobe has a legacy of providing audio and video editing software. The free Adobe Premiere app offers access to the Adobe Stock and support for creator assets, too – and unlike Canva, doesn’t demand a login to start using it.

Free Adobe tools on Android

To install Adobe Premiere on Android, you’ll need Android 13 or later, with 5GB of RAM. The first release has a download of just under 60MB, and Firefly-powered generative AI features will require an active internet connection. Modern Chromebooks with over 4GB of RAM should be able to run the software, although there is every chance that Adobe removes that option.

Other Adobe tools are also available on Android. Aside from Adobe Premiere Rush (which would seem to be heading to retirement), Adobe Lightroom Mobile, Adobe Photoshop Express, and Adobe Express are among the most popular, offering mobile takes on the desktop Photoshop experience. These also provide an alternative to Canva’s monthly subscription, and again are free to use but have some paid features.

Meta says the keychain will be ready to ship by the holiday season, like some kind of bizarre AI stocking stuffer.

• OpenAI has filed court documents against Elon Musk’s xAI
• The papers reveal that ChatGPT’s integration with Siri was disappointing
• OpenAI alleges that barely anyone was using Apple Intelligence

If you’ve noticed an improvement in Siri’s abilities on your Apple device, you might have Google to thank for that, as its Gemini artificial intelligence (AI) now powers much of what Siri AI can do. Rival services like ChatGPT have also found their way into iOS — but new court documents reveal why that particular partnership has struggled to get off the ground.

The court filings were made in relation to OpenAI’s ongoing dispute with Elon Musk’s rival xAI firm (now known as SpaceXAI). In the documents, OpenAI alleges that, by the time xAI filed its complaint against OpenAI, “it was clear that Apple’s integration of ChatGPT was dramatically underperforming.”

The papers then go on to imply that this was not just some one-off failure. Instead, the Apple integration was “persistently underperforming,” OpenAI alleges.

Apple added ChatGPT integration to Siri in December 2024, but this required a multi-step opt-in process. That could be one reason for the substandard performance of the link-up, with a degree of friction slowing users down in their attempts to harness OpenAI’s tool.

Much of this section of the court filings is redacted, so it’s difficult to know exactly what was going on between Apple and OpenAI. But it seems clear that, at least from the latter’s perspective, the results were severely underwhelming.

What might have been

(Image credit: Getty Images)

When Apple Intelligence first launched, Apple provided users with an option to tap into ChatGPT if they needed something a little more powerful. For example, this was an option in Visual Intelligence, providing extra context and help with the images you shot with your iPhone’s camera.

Yet OpenAI’s legal filing makes it clear that barely anyone was using any of ChatGPT’s tie-ins with Apple Intelligence. OpenAI describes the effect the integration had on competitors (like xAI) as being “indisputably de minimis.” It added that its own expert, Dr. Catherine Tucker, calculated “the share of GenAI consumers who accessed ChatGPT through Apple Intelligence” and found it to be “consistent with OpenAI’s internal view that Apple Intelligence saw minimal usage.”

That hints that the problem might not have just been limited to ChatGPT’s integration with Apple devices, but with Apple Intelligence as a whole. Apple’s AI system was woefully underdone when it first arrived and clearly lagged behind its rivals. First impressions matter, and if most people were left unconvinced by Apple Intelligence, ChatGPT might have suffered the knock-on effects among Apple’s customers.

Did ChatGPT lose out because Apple fans had to dive into the Settings app on their device, find the relevant section and laboriously wade through several steps in order to enable ChatGPT? Because Siri’s limitations dragged ChatGPT down with it? Or did users simply prefer to launch ChatGPT’s standalone app instead?

We don’t know the answer to that, at least not right now. But with Siri getting a glow-up under the Siri AI banner, it feels clear to me that Apple wants a fresh start for its AI efforts. Maybe things would have been different for OpenAI had it integrated with the more powerful Siri AI rather than the underbaked Apple Intelligence. But with Siri AI righting many of the wrongs of Apple Intelligence’s fudged launch, OpenAI may be left ruing what might have been.

• Gambit researchers uncovered ongoing AI‑driven skimming campaign stealing 600,000+ payment records since July 2026
• Attackers used three autonomous harnesses (Strix, Cairn, Hermes) to compromise dozens of retail sites cheaply
• Victims include major US firms; campaign shows AI enables faster, persistent, low‑cost cyberattacks at scale

In July 2026, a hacker tasked autonomous AI agents to attack retail organizations around the world, deploy credit card skimmers, and steal payment data.

Since then, the bots launched hundreds of attack projects, compromised dozens of organizations, and stole at least 600,000 payment records - and to make matters worse, the campaign is still live, attacking and breaking into websites as we speak.

All of this was reported by security researchers Gambit, who said they managed to recover the operator’s staging server and through it - reconstruct the ongoing campaign. They also saw the skimmers live on victim websites, and sifted through logs and AI claims found on the attacker’s server. In just five days, between September 10 and 15, the agents made 105 attack waves and compromised 27 organizations “to varying degrees.”

Among the victims are a Fortune 500 hospitality company, a “major” US airline, a large private US industrial supplies distributor, and a US online fashion retailer. One of the AI tools would use a website ranking service to produce a list of potential targets, focusing primarily on those running custom-built software.

A fistful of dollars

But the victims are not the “interesting” part of this story - the attackers are. Gambit believes they are financially motivated Chinese threat actors. They are using three AI “harnesses” (frameworks, essentially), which can run almost the entire attack chain autonomously, striking around 10 companies a day, for a handful of dollars per company.

In four weeks, the attackers spent around $7,000, meaning that their entire cost for the operation so far was no more than $18,000. Breaking it down, it means that the attacker spent around $25 per target.

“Spread over the companies attacked, this is a marginal cost of a few US dollars to a few tens of US dollars for each targeted company,” Gambit’s researchers said. “The operator’s own cost review gives a similar figure, a mean of $25.46 over 101 completed scans, from $3.13 for the cheapest target to $79.31 for the most expensive.”

“Where access was achieved, it usually took less than a day, and in many cases just a few hours. We also detected instructions in the attacker’s playbook that could disrupt the operations of a company as a result of data deletion or cleanup procedures run by the agent - and this has indeed happened in some of the breaches,” Gambit said.

The three harnesses

The three harnesses are called Strix, Cairn, and Hermes.

Gambit describes Hermes as an open source autonomous AI agent with a persistent memory, skills that the agent wrote and edited itself, a searchable archive of past sessions, scheduled jobs, and a web console. On the staging server the researchers analyzed, it loaded a Chinese system persona called “SOUL - Red Team Operator”, which contained 121 skills (78 attack skills).

“Hermes is the operator’s console for orchestrating the activity and for direct hacking activities,” Gambit explained. “It used Anthropic’s opus-4.6 (after newer models refused its requests), with 1,951 prompts typed by the human across 260 sessions - only a few prompts per target. The human prompts are short instructions in Chinese, usually launching an attack, tasking the agent with a general next step, or what to do next after achieving access.”

Strix is an open-source AI pentest tool, while Cairn is an autonomous pentest engine. It receives target domains and an objective, such as to get a shell or admin access, then runs for hours until it achieves the objective, times out, or is stopped. Cairn used DeepSeek v4.1 Flash, it was said.

Gambit’s researchers seem to be rather impressed with the campaign. They described it as very low cost, with a level of patience, persistence, and creativity that most human attackers would be “unlikely to sustain”, managing to achieve “far greater results, far faster.”

They have also called to arms, urging organizations to “adapt to a reality where attacks are significantly faster and more comprehensive.” To do that, they must adopt a resilience-first mentality and deploy a security stack that can match the AI on speed.

Many of the affected organizations were notified, and the skimmers were removed, they said.