• The LOCUST X3 laser is developed by AeroVironment, Inc.
• It is designed as a “group 3 killer” – meaning it is intended to tackle particularly dangerous drones
• The Xbox controller makes the LOCUST X3 laser simpler to target

A new US Army contract worth around $465 million has been awarded to AeroVironment, Inc. for its LOCUST X3 anti-drone weapon, a 30-kilowatt laser that can be mounted on military vehicles.

Designed to take down mid-sized attack drones, the LOCUST X3 can specifically handle what the Department of Defense refers to as “Group 3” – unmanned aerial vehicles weighing 1,350 pounds and capable of lifting 125 pound payloads over a distance of 8 miles.

Interestingly, AeroVironment, Inc. has designed the LOCUST X3 to be relatively easy to control. Operators (which may be soldiers, sailors, airmen, or even Space Force guardians) can target the anti-drone laser using an Xbox controller.

“The apex predator”

Incredibly, this is the first high energy laser weapon to be awarded a production contract in US history, following several years in development, which included military testing.

Wahid Nawabi, Chairman, President, and Chief Executive Officer at AV said: “The transition of directed energy from experimentation to an enduring, fielded capability reflects years of collaboration, innovation, and operational success alongside the U.S. Army.”

Aaron Westman, senior director for business development at AeroVironment, Inc., outlined how the LOCUST X3’s role is viewed by the company. “It can really kill a lot of threats. Its job is to really, let’s say, thin the herd to be able to kill high volumes of threats.”

John Garrity, vice president of directed energy systems at AeroVironment, Inc, expanded on the use of an Xbox controller for targeting the anti-drone system.

“To a young soldier, sailor, airman, guardian that’s using these systems, it’s not unlike what they’re using at their own house for video games," he noted.

“Group 3 killer”

The purpose of the LOCUST X3 is to dispatch attacking drones relatively efficiently. But what does “Group 3” actually mean?

Garrity told reporters that “LOCUST X3 is intended to be the Group 3 killer. When we talk about Group 3, that’s just one of the more prevalent threats that you see in warfare today, seen in conflicts in Iran and in Ukraine with Shahed drones.” The LOCUST X3 is intended to be a one part of the US Army’s response to done attacks, with other weapons contributing to the multi-layered defense against Group 1-3 UAV threats.

The one-way Shahed-136 drones have been used against US troops in the Middle East. The LOCUST X3 is positioned as a solution to this, and not just due to its portability. It also has longer range than many similar countermeasures.

US interceptor missiles are believed to be largely depleted thanks to the conflict with Iran. Anti-drone weapons like the LOCUST X3 are viewed as a forward-thinking alternative.

• 69% of Americans oppose AI data centers near their communities, a national poll finds
• Public support for nearby AI data centers has fallen sharply since 2025
• Only 4% strongly support new data centers in their communities today

American resistance to AI data centers has intensified sharply over the past year, several national surveys released in 2026 have found.

A new Decision Desk poll conducted by SurveyMonkey found that over two-thirds (69%) of respondents oppose construction of these facilities near their homes.

Nearly half of those surveyed (45%) said they would strongly oppose a data center built anywhere close to their own neighborhood.

Data centers do not have widespread American public support

Separate research from Heatmap Pro, carried out by Embold Research in mid-August 2026, points to an even sharper decline in public backing, as its poll of 2,045 registered voters found 75% opposed building a facility near their community.

Only 4% of respondents said they would strongly support a proposed data center in their area, a drop from 13% recorded a year earlier.

Resistance now cuts across age, income, party affiliation, and the split between rural and urban communities nationwide.

Republicans remain 43 percentage points net negative on the issue, independents sit at 65 points negative, and Democrats register 75 points negative.

Rural voters, a group that has trended Republican in recent years, are 63 points net negative as well.

Officials across the political spectrum have taken notice. New York Governor Kathy Hochul ordered a one year pause on new facilities, and Texas Governor Greg Abbott froze several pending projects pending a required audit.

Heatmap Pro figures reveal that over 530 counties and municipalities have restricted or banned data center construction.

Political and economic fault lines

President Trump has pushed back publicly against the trend, framing local opposition as economically self-defeating.

“The only reason that communities throughout the U.S.A. should not want Data Centers is if they want to end up being backwards and poor,” Trump wrote on Truth Social.

“If they want to be successful and rich, with far lower taxes and jobs all over the place, let Data Reign.”

However, Republican supervisor Dawn Vogel backed a local moratorium earlier this year, putting her at odds with Trump's national agenda.

“I don’t think this is a Republican or Democratic issue…It’s an issue of what’s the best use for the land, what’s the best use for our natural resources,” Vogel said.

“You can have all the data you want to in the world, but if you don’t have water to drink, you don’t have humans to use the data.”

Lawmakers who once championed data centers as economic development tools are now revising their positions under mounting public pressure.

Ohio Senator Jon Husted, an early advocate, introduced legislation this year requiring data center operators to cover added electricity and water costs.

The gap between economic promises and local resource concerns appears to be driving much of the shift.

However, labor unions have generally taken a more favorable stance, citing steady construction jobs and strong wages tied to the facilities.

• The AVEN thrust-vectoring nozzle has been added to the X-BAT AI-piloted fighter with vertical take-off capabilities
• Thrust-vectoring allows a plane’s exhaust to be redirected, instead of relying purely on manipulating airflow across the wings for movement
• The engine was first trialed on an F-16 fighter in the 1990s

Shield AI has revealed the X-BAT AI-piloted fighter jet relies on a jet engine variation first developed over 35 years ago and tested on a General Dynamics F-16 in the 1990s.

The AVEN – Axisymmetric Vectoring Exhaust Nozzle – essentially replicates the actions of a wing, but instead of manipulating airflow across the airfoil, it directs the exhaust, in a process labeled “thrust-vectoring.”

Developed for the F-16 as part of a dedicated GE Aerospace program investigating Multi-Axis Thrust Vectoring (MATV), the nozzle was proved to work. The technology has recently been revived thanks to GE Aerospace and Shield AI, with the AVEN paired with an F110-GE-129E engine.

The VTOL X-BAT needs AVEN

Using the Axisymmetric Vectoring Exhaust Nozzle, the X-BAT is given flight capabilities that would prove demanding for humans to repeatedly endure. It enables hovering, but also superhuman reaction speeds, and removes the risk of G-force induced blackouts. Without pilot fatigue, the X-BAT can commit to extreme maneuvers in aerial combat.

Designed as a Vertical Take-off and Landing (VTOL) aircraft, the X-BAT can be deployed in areas with little or no runway. The Axisymmetric Vectoring Exhaust Nozzle is a key part of this, providing the necessary thrust for the X-BAT to take off. As the craft launches more like a rocket than a standard jet fighter, the AVEN and the F110-GE-129E engine appear to ensure a steady detachment from the launch dock as the X-BAT heads skyward.

It’s a setup that might resemble something out of a 1950s sci-fi comic, but in reality can only work successfully without a human pilot.

AVEN isn’t unique to the X-BAT. While the GE Aerospace project with the F-16 proved the technology worked, it was pursued elsewhere, most notably with the Russian Sukhoi Su-27SM revision of 2002. Earlier craft, like the Harrier variants, use “2D thrust vectoring” where thrust is directed horizontally or vertically via exhausts on either side of the aircraft. Indeed, the notion of thrust vectoring dates back to the V-2 rocket, the long-rang rocket used by Nazi Germany in the final months of World War Two.

Modernising the AVEN

Shield AI and GE Aerospace haven’t simply stuck to the original AVEN hardware and attached it to the F110-GE-129E engine. The nozzle has been re-engineered, with modern materials and production techniques. Inevitably, it has also been adapted to work with 21st century guidance systems, and the AI piloting and control systems are also accommodated.

First test flights of the X-BAT are expected to take place in Kansas by late 2026, ahead of a planned 2028 operational validation and full production.

The GL.iNet Beryl 7 is a travel router that is designed for those on the go who still need control over their networking.

Especially today, it feels like people are more aware of their networking than ever. There's fear over what people are able to see. There's fear of what other people can see of your own data, and there are people trying to take control of their own connections. That's one very real use case for the Beryl 7.

Alternatively, another use case that is possibly more common, or at least more widespread, would be that a lot of people travel with an abundance of gadgets these days. It used to be that people would have their phones, and that would be about the only thing they would connect to Wi-Fi, then maybe a laptop, then maybe an iPad.

Now the average family, to my understanding, travels with several tablets, laptops, phones, and oodles of other internet-connected devices, or perhaps that's just my family when we travel.

Regardless of which camp you're in, having a parallel router can actually prove to be incredibly helpful. Personally, I have seen some people use it as a way to connect to a VPN easily and with whatever device they have with them. I've seen some people use it as a way to be able to only have to connect to a guest Wi-Fi once and then extend that network to all of their devices.

I personally have used it in the sense of, when traveling, not having to disconnect and reconnect our devices at every stop that we make. Instead, I can connect just once and then keep all the kids' iPads and devices, as well as my own, on a known and already saved network, just simply for ease of setup.

I've also seen people use it as a way to turn Wi-Fi networks into a wired network and vice versa. All around, a little device like the Beryl 7 turns into a Swiss Army knife of networking capabilities.

GL.iNet Beryl 7: Price and availability

The Beryl 7 is available direct from GL.iNet store in the US here and in the UK here. The travel router is priced at $130 / £117.

It's also widely available from online retailers, including Amazon, where I've seen it down as low as almost $100.

The Beryl 7 specifically is not as nice as some of its bigger brothers, but for the price, if you're not changing a ton of settings but more setting and forgetting, this one might just be the perfect little addition to your tech bag.

It's lightweight, doesn't take a ton of space, and setup and go is super easy, especially with the ability to run this off of just a single low-powered USB-C.

GL.iNet Beryl 7: Unboxing and First Impressions

As a tech lover, I was excited to get my hands on this one. It seems like it's a solid update and even an upgrade from past models, and I wanted to see just how lightweight it really was. I got the mint green colorway, which is pretty striking, and for something like this, it is a little bit more fun than just a black model.

Normally, I'm an all-black or, these days, maybe even some silver technology kind of person, but I didn't mind the flash of color for something like a little networking device.

The weight is just as light as I expect it to be, or should I even say how much I hoped it would be. The ports are simple, the mode switch is sleek, and the ability to use USB-C really is just that convenient.

Even as a heavy tech reviewer, I've almost made it to where every single thing I carry is USB-C, which has become incredibly helpful for when traveling, working remotely, or at different offices and just going about my day-to-day. Needing to carry multiple cables and multiple different types of cables is a thing of the past… almost.

Having USB-C on even my travel router that I'm testing is super convenient. Further, it's nice to see two 2.5 GHz network ports, a USB 3.0 port (even though it is capped at a pretty slow speed), and two folding antennas that are satisfying enough to be near fidget-toy status.

GL.iNet Beryl 7: Design & Build Quality

Talking specifically about the build quality. While this is plasticky and the folding arms are fun to play with, it does feel like over a couple years, or even just some hard daily use, the arms could get pretty loose and won't snap open quite as nicely.

I'm also wary of dropping this because it feels like a single drop could crack the entire shell. Cases are available for the device, and if you're using it for business travel, I recommend picking one up.

With that being said, though, it is really nice to see little rubber feet on the bottom of the portable router, making it so that it's not just the sound or experience of plastic sliding along whatever I decide to put this down on.

Moving to the ports, they feel pretty well built. Cables snap in pretty easily and feel like they're a tight connection, and I appreciate the layout and spacing between all the ports.

It makes more sense to spread out even just these four ports than it does to have them all crammed together for a look of minimalism. Having them spread out a little bit on the back side of this unit makes more sense from just a usability standpoint.

GL.iNet Beryl 7: In use

I brought this with me on a few different planned trips, and then it's come in handy at other times as well.

The first trip I brought it on was a quick little family trip. Knowing that we were going to stop at a hotel on the drive and how insane it was to get everything set up for just a few hours in a hotel and then continue driving the next day, zero part of me wanted to spend time connecting my devices to the hotel Wi-Fi. I also needed to spend time setting up pack-and-plays and getting everything squared away.

I made sure to bring this travel router to try out what this would look like in real time. I already had all the devices programmed and ready to go ahead of time to the network I had set on the Beryl 7.

When we got to the hotel, I had to connect one single device to the hotel Wi-Fi, and then I was able to broadcast that signal to all of the devices. It was stable enough for 5 iPads, a laptop, and a couple of phones, and therefore earned its place in my bag for the next trip I went on.

The next trip I went on was a quick work trip where we stayed at an Airbnb. I knew I'd be working pretty heavily, and I did not want to have to deal with potentially sketchy Wi-Fi, connecting to an unknown source, and then again having to connect all the devices I travel with.

On this trip specifically, I had two laptops, two iPads, and my phone, just for my own devices, let alone my team that I helped support technology-wise. I brought the Barrel 7 so that I could connect one device and then broadcast my own network, or at least that was the plan. What I ended up actually doing was connecting directly to an Ethernet port and not even having to connect and then broadcast via Wi-Fi.

One thing that I wish I would have done on that trip is I wish I would have set up the VPN earlier on my Beryl 7. I was able to do it during the trip, realizing how handy it would have been to do ahead of time, but I ended up needing to remote into my network rack back at the office.

While I can do that through VPN connections on my laptop, having one on this portable router, for some reason, just feels so much easier because then I can connect quickly on my iPad as well without having to toggle through things. I can just know that anytime I connect to this router, or if anybody else needs to connect to it, they can all be on the same VPN.

Personally, for me, it was just to change some settings and prep some things for the upcoming week. But, I could see this being super helpful for those who have Home Lab servers that need troubleshooting or other things like that. Further, if you have a work VPN that you need to be connected to, check with your employer.

Potentially, this could be a great option for you to not have to connect several devices, but you can connect one. For some work, that may be a great option. But, one thing to keep in mind is the speed limitations if you are choosing to use VPN. Moving things around on the network that you're connected to should be fine, but pulling files through the VPN to your device or vice versa will feel very slow.

GL.iNet Beryl 7: Final verdict

Whether you're working out of a coffee shop and don't want to connect multiple devices, or you have a secure VPN you need quick access to on any device you're bringing. If you simply want to be able to connect to a Wi-Fi network and turn it into a wired network, or take a wired network and turn it into a wireless signal, a little travel router like this can make things a lot easier.

Or perhaps you're like me, and you're starting to learn more and more about how scary network traffic can be, and you want to make sure that you're never connecting directly to a sketchy Wi-Fi. That's a great first step, and utilizing something like this and a VPN to help scatter your traffic may be a spectacular solution for you.

If you're looking for more networking capabilities, more features, and more power, make sure to check out GL.iNet for some of their other spectacular options, or keep an eye out for more reviews coming.

• Alogic Clarity Touch 6K monitor packs 6016 by 3384 pixels into one massive display
• The webcam retracts fully into the monitor when not in use
• Thunderbolt 5 support is missing from this new flagship monitor

Alogic has upgraded its Clarity monitor range with a new 32-inch 6K model featuring an integrated 12MP webcam.

The company announced the display at IFA 2026 in Berlin, where it also showed several other new desktop products.

Despite its high resolution and new camera, the Clarity Touch 6K monitor does not list Thunderbolt 5 among its built-in connections.

Clarity 6K gets a sharper camera

The Clarity Touch 6K uses a 6016x3384 display, giving it substantially more pixels than standard 4K desktop monitors.

Its main new feature is an integrated 12MP 4K webcam that retracts into the display when not in use, removing the need for a separate camera while keeping the webcam concealed when video calls are not underway.

Alogic has also added the same retractable webcam concept to its new 27-inch Clarity 5K monitor.

Both models use USB-C for laptop charging, with support for power delivery of up to 90W through the display.

The device also supports HDMI 2.1, DisplayPort 1.4, USB-A and audio connections for peripheral devices, however, the absence of Thunderbolt 5 is notable because Alogic has introduced a separate Thunderbolt 5 dock alongside its latest monitor range.

That dock supports bandwidth of up to 80Gbps in symmetric mode and up to 120Gbps in asymmetric mode.

Price and availability remain premium

According to Alogic, the Clarity Touch 6K with webcam is scheduled to become available by the end of October 2026.

The price of this device varies across three major markets, including the United Kingdom, the United States of America and the European Union - as in the UK, this device is listed for £2,299.99 while it sells for €2,799.99 and $3,099.99 in the EU and United States, respectively.

Those in Canada and Australia can also purchase this monitor for CAD$3,999.99 and $3,699.99, respectively.

If you need a cheaper alternative, then you may go for the smaller Clarity 5K with webcam, which costs £1,499.99, €1,799.99 and $1,899.99 across the U.K., EU and U.S respectively

The Clarity Touch 6K arrives alongside an Edge 34-inch ultrawide monitor, built around a 5120 × 2160 panel with 100Hz refresh.

Per Alogic, its IPS display reaches 400 nits of brightness and covers 100% sRGB, 96% DCI-P3 and 91% Adobe RGB.

It also includes USB-C charging at up to 90W, plus HDMI, DisplayPort, USB-A and Gigabit Ethernet connectivity for accessories and external drives.

Alogic is also expanding the Aspekt range with 27-inch 5K, 32-inch 6K and 34-inch 5K touchscreen models featuring integrated docking and 10-point touch input.

The 32-inch version reaches 6016 × 3384 resolution with MPP 2.0 stylus support, and the 34-inch model offers a 21:9 layout with 100Hz refresh.

“With this expanded portfolio, we are giving professionals, creators and everyday users more choice in how they work and create,” said Ritesh Kumar, CEO and Director of Sales at Alogic.

“From high-resolution displays and expansive ultrawide screens to integrated webcams and powerful connectivity, each product is designed to make the workspace more capable, connected and easier to manage.”

The Lenovo ThinkPad T1g Gen 8 is a powerhouse and a semi-portable shell. It takes on the power of the beloved X1 Extreme, pairing an ultra 9 processor with quick memory, Thunderbolt 5, and decent storage. It's designed for engineers, creatives, and other power users who need a machine that can do it all while still having that hint of portability.

I think of this quite like the 16-inch MacBook Pro. It's the bigger choice in the line-up, but you would get it because you want the power and the ability, not necessarily the portability. The people who would get this laptop are the people who need the ports as well.

That's one of the reasons why I'm thrilled to see USB-A still, though it is just one, which is all that I would want. USB-C, Thunderbolt, an SD card reader, and an HDMI port. It honestly has every port I wish my MacBook Pro had, even as a guy who loves the MacBook Pro.

Lenovo ThinkPad T1g Gen 8: Price and availability

The Lenovo ThinkPad T1g Gen 8 is available from Lenovo in the US here, and in the UK here. Prices start at $3,632 / £5,0145.

My tested version of this laptop has an Intel Core i9-285H, an IPX5070 with 8 GB, 32 GB of RAM, and a 1 TB SSD. It's pretty hefty, and it does the job, that's for sure. The fact that it has a dedicated GPU shows that this thing is meant to work. The screen is also WQUXGA (3840x2400) with an 800-nit IPS panel.

You can fully inspect this thing out with a loaded Ultra 9 5070 with 64 GB and 4 TB of SSD. That will run you just over $4,000. My unit is not fully spec'd. It was pretty close and comes in at about $3,400 to $3,700, depending on how I look at it. If you get a base level with an entry point of an Intel Ultra Core 7 RTX 5060 and a terabyte of SSD, you can get it for just under $1,700 at $1,688.

So, much like most of Lenovo's line-up, you can expect this thing to go up and down the price chain however far you want. What's important to recognize is that they have laptops all across the spectrum for what you're wanting.

Lenovo ThinkPad T1g Gen 8: Unboxing and First Impressions

Right off the bat, this is a super sharp machine. I'm a big fan of Lenovo laptops for how they always look uniform yet still have a little taste of differences.

When it comes to being uniform, it still has the matte black that I love, the touchpoint, a generous trackpad, a very comfortable keyboard, a great screen, a physical shutter on the camera, spectacular ports, and all-around really standard and clean branding.

The things that are different, though, are that, unlike a lot of the other models that I've tested, this doesn't have a left and right-click button at the top of the trackpad that's meant to be used with touchpoint.

It does still have a little bump, though, so that if you know that it's there, it's still accessible. I tested it, and the right-click still works just fine. It's just not a physical button; it's built into the touchpad.

Outside of that, this still looks super sharp. While it is bigger than some of the other models, it's by no means too cumbersome to carry. I would happily carry this around every day in my backpack, and it fits even at 16 inches in most standard-size or slightly bigger backpacks. It's a pretty tight fit in something like a Bellroy Daily Pack, but in my beloved Nomadic or TomToc bags, this fits without any problem.

In the box comes the standard charger and always great packaging to make sure that the laptop stays safe while in transit or in the store. Admittedly, I rarely use the chargers that come with these devices personally, but that's because I usually opt for more efficient and more useful chargers from Anker, Ugreen, or even Nomad.

For something this size, though, I'll probably continue to use one of my bigger Anker chargers that have a few ports, phenomenal output, and even Bluetooth capabilities to give me readouts. I can see where my power is going for each port on the plug.

Lenovo ThinkPad T1 Gen 8: Design & Build Quality

As previously mentioned, this is an absolutely spectacular build device. If you're more familiar with Apple products, this does feel quite a bit like the MacBook Pro. I feel bad constantly comparing it to that, but the reality is that the MacBook Pro is a little more distinguishable and more known in the popular field.

What's unlike the MacBook Pro, though, is that this also features a trackpoint, a privacy shutter, human presence detection, upgradable memory, and that USB-A port that I wish my MacBook Pro had.

It's no secret that this is a big machine. It's a 16-inch powerhouse, so don't expect it to be as light as a feather or as light as some of Lenovo's carbon models. That doesn't mean that it's wildly uncomfortable either.

Yes, it's a big laptop, but for its size, it's so barely portable and relatively thin and light. Especially when you consider that you can get this spec'd out with a tandem OLED 120 Hz screen, massive amounts of storage, 64 GB of RAM, and a dedicated graphics card. That's pretty insane all around.

Lenovo ThinkPad T1g Gen 8: In use

I've had this laptop in rotation for my team and myself for the last 163 days as I write this paragraph. Throughout the time, I've used it for virtual meetings, projects in Notion, a lot of writing, and emails.

I'm taking over the world one AI at a time, trying to teach myself things through Claude and GPT. I've taught myself how to code. I've done a lot of website work. I've sent a lot of Slack messages and even watched a fair number of tutorials on YouTube.

While working on project management, this executive screen fits a good amount of data, especially with the higher resolution. I can fit quite a bit on this screen. When it comes to basic photo editing, the screen is still pretty sharp and fairly color accurate. It may not be the best photo editing laptop I've ever used, but for most this would be just fine.

This machine is powerful enough that I can jump in and run quite a bit of my workload on it without even feeling like it's slowing down at all. I've done a ton of work building out Notion workers utilizing this laptop specifically.

I've done a lot of work in VS Code, GPT, Claude, and Notion, running some pretty heavy workflows to get the sinks working the way I wanted them to. It all happened without any problems on this machine, thanks to its robust processor, RAM, and storage.

Beyond that, typing has been an absolute dream on this. I have always said, and probably will continue to say, that Lenovo laptops, specifically the ThinkPad variant, have some of the best keyboards on the market, perhaps ever. Even throughout all their changes over the years, they continue to have a spectacular keyboard.

The best way I can describe this laptop is that it is for people who need a professional machine for workloads to match. These people aren't the ones that are just working on documents or emails or quick office tasks like that.

It's the people who need to be able to jump into any kind of work at any time. They're using a ton of processing power, and they need a lot of ports. They're the most likely to be more tech-savvy individuals who bounce in and out of what they do.

Lenovo ThinkPad T1g Gen 8: Final verdict

All in all, this laptop has been wonderful to use. I have most appreciated the parts and the reliability of this machine. Being able to plug in any device I may need without having to carry a dongle is peace of mind that I do not take for granted.

Also, having enough headroom in your powertrain to take care of all the projects in one's own busy days is a superpower that many would appreciate.

If you're the generalist or the specialist of many, if you're into systems, IT, or you're just the kind of guy who always has a thousand projects he's juggling, then this is absolutely a laptop you should consider.

For most people, this would be overkill for them. In my technical world, people I work with would absolutely benefit from this kind of machine. It has the power to run bigger programs and the storage to manage data if that's part of what they're doing.

It has ports to jump in and out of whatever they are solving, and it's portable enough to carry around from place to place (even though they're not going want to walk around holding this in their hand anytime soon). It's possible, but it's just not comfortable.

It's sharp, it's sleek, it's powerful, and it has ports. It's expandable all around. Lenovo nailed it with this laptop.

For more top picks, see our guide to the best business laptops we've tested.

• Kinryū Labs found misconfigured APIS database in Vietnam exposing 220M passenger and crew records
• Data included PII, travel details, seat assignments, and baggage references from 2017–2026
• Archive locked and there is no evidence yet of dark web sale

Millions of travellers have had their sensitive information exposed on the internet, thanks to a “series of misconfigurations” discovered in a cloud database, experts have warned.

In early June 2026, security researchers from Kinryū Labs discovered an Elasticsearch cluster which, although inaccessible from the open internet, allowed access through an alternate route - a cloud-based path. Once inside that route, the researchers discovered that the cluster accepted default credentials, granting access to an archive with 29 indices, weighing roughly 107GB.

The archive was apparently generated by an Advance Passenger Information System (APIS), a system airlines use to collect and send passenger and crew information to country authorities before flight arrival or departure. Usually, the system collects people’s names, birth dates, nationalities, passport and travel document numbers, and flight details, which are then used for border control, immigration, and similar cases. In this case, however, it also contained people’s sex, document expiration dates, issuing countries, airlines, departure, destination and transit airports, seat assignments, baggage reference, and more.

It is not known who operated the APIS, or who owns and runs the database. All the researchers managed to find is that it was hosted in Viettel-assigned IP space in Hanoi, Vietnam’s capital.

Who was affected?

Of the 29 discovered indices, two were rather large: one contained 210,318, 069 passenger records, while another 10,465,631 crew records. In total, 220 million records, created between January 2017 and April 2026, from people who traveled to, from, or through Vietnam during that period.

The records don’t correspond to individuals, however. If a person travelled multiple times, they will show up in the archive multiple times. Canadians, Chinese, Korean, and New Zealandian nationals are among the ones whose information was exposed. The database is not limited to a specific airline, either - various airlines in Asia-Pacific, Europe, and Middle East regions were mentioned in the researchers’ report.

Since the researchers could not attribute the database to a specific entity, they reported it to the Vietnamese authorities, different airlines mentioned in the archives, and the country’s CERT (Computer Emergency Response Team), on June 3. The archive was locked down a week later, on June 8. According to BleepingComputer, it was the Singapore Airlines’ security team that took the lead on remediation efforts, telling the researchers they “engaged the relevant parties” and have “taken steps to contain the issue.”

Without a proper audit of the logs, and a full-blown forensic investigation, it is impossible to tell if any threat actors reached the database before the researchers, or if they exfiltrated and used the data found inside in identity theft, wire fraud, or other scams. What’s (somewhat) comforting is that there is no evidence of such activities on the dark web, and no hacking groups have made such claims. No one is selling the archive on the dark web, either.

Number one causes of data spills

Misconfigured databases remain one of the key causes of data leaks. Most businesses generate some type of data on their employees, partners, clients, or customers, and store it in the cloud for easy access and actionable insights. However, some businesses don’t understand the shared responsibility model of cloud security or are simply reckless and sloppy when it comes to securing their cloud-stored data. Others, according to Cassius Edison, COO of Closed Door Security, have a problem with visibility of their IT real estate:

“The range of technology now used and managed by firms globally has made misconfigurations an increasingly persistent problem,” Edison explained. “Many organizations fail to maintain full visibility of their IT real estate and fail to perform proper audits of their systems, which inevitably leads to oversights in security and monitoring.”

For Edison, tackling misconfigurations internally can be difficult, “especially at large companies where teams work independently across a range of systems,” and advises organizations to bring in independent pentesters and security auditors.

Some of the biggest data leaks in the world came not from hackers breaking into locked-down systems, but from businesses inadvertently exposing their customers.

In 2026 alone, we’ve seen more than 670 million identity records exposed by Infutor, a data-driven consumer identity management company, as well as more than three billion records exposed through a misconfigured MongoDB database managed by global identity verification services, IDMerit.

Via BleepingComputer

• South Korea could invest $22.3 billion in a massive Texas gas plant
• The proposed facility would generate 6.3 GW of electricity for AI
• Seoul says negotiations remain unfinished, leaving the reported investment unconfirmed

South Korea and the United States have reportedly settled on how much Seoul will contribute towards a new energy facility in southern Texas.

Korean outlet Edaily, citing unnamed officials and lawmakers, reported the contribution at close to $22.3 billion for the venture, which would be built in Encinal, Texas, and is expected to produce 6.3 gigawatts of electricity to power AI data centers once fully operational.

The contribution would mark Seoul's first American investment carried out under an economic arrangement the two nations signed in 2025 which saw South Korea commit $350 billion toward ventures inside the United States in exchange for friendlier duty terms on its exports.

A deal rooted in last year's pact

However South ​Korea's Industry Ministry said the recent media reports about the Texas deal were not accurate, revealing that official discussions with Washington are still going on, so no exact numbers can be confirmed yet.

Any formal announcement would come only after discussions conclude, alongside domestic steps such as approval from South Korea's legislature.

In a separate statement, the ministry noted that no decision has been reached on building an atomic energy facility on American soil.

South Korea is also weighing other ventures inside the United States, including a large-scale nuclear station and a separate gas export scheme tied to Alaska.

Scale, cost and open questions

Should the project proceed as reported, it would rank among the most expensive natural-gas-fueled generating stations ever constructed anywhere in the world.

Industry accounts also suggest the plant could become the second most powerful facility of its kind currently in operation globally.

At the moment, there is no clarification whether Seoul intends to finance the entire sum on its own or share the burden.

The Encinal facility, once running, would supply electricity to server farms supporting growing artificial intelligence computing needs across the region.

Washington has also not issued detailed confirmation of the numbers currently circulating in South Korean press coverage.

For now, financing structure, ownership stakes and construction timelines have not been settled publicly by either government.

Current details are very sketchy, but analysts say the eventual scale of investment could shift depending on how negotiations with Washington proceed this fall.

Whatever the details turn out to be, a facility of this size would place significant demand on the Texas electrical grid, water access and local infrastructure.

Given how US residents have reacted to projects with similar demands, that pressure may not be welcomed quietly by communities near Encinal.

Via Reuters

• US senators call on Army to answer questions over disbanding of futuristic drone battalion
• Letter to General LaNeve raises concerns over US commitment to NATO allies and Ukraine
• Drones are changing the face of the battlefield, and the US could miss out on important lessons and innovations

A group of bipartisan US senators is questioning why the US Army has decided to shut down a cutting edge drone battalion that was set up to understand the shifting battlefield dynamics in Ukraine.

The 3rd battalion of the 504th Parachute Infantry Regiment was founded in November 2025 by former chief of staff for the US Army Gen. Randy George. But less than a year later, acting chief of staff Gen. Christopher LaNeve announced plans to disband the battalion as part of a ‘back to basics’ approach.

But there are concerns the decision could harm the US military's warfighting capacity, and its ability to learn from a rapidly evolving battlefield technology.

Senators question decision to disband drone battalion

The letter, signed by Sen. Jeanne Shaheen, Sen. Thom Tillis, Sen. Angus S. King, Jr., and Rep. Michael R. Turner, requests that Secretary of the Army Daniel Driscoll and General LaNeve provide the data and analysis behind the decision to disband the drone battalion, and also questions “if this decision was based on Department of Defense leadership guidance or was an Army-internal decision.”

“We have deep concerns that eliminating this specialized drone unit will limit our ability to learn from allies, particularly the Ukrainian Armed Forces, and hinder our efforts to modernize drone warfare at the speed necessary to compete on the modern battlefield. This specialized unit was a prudent response in a moment when the character of warfare is changing faster than a conventional formation’s ability to adapt,” the letter states.

The letter also points out that Russia’s invasion of Ukraine and US coordination with the Armed Forces of Ukraine provided the basis for the new battalion, providing a “prudent response in a moment when the character of warfare is changing faster than a conventional formation’s ability to adapt.”

Additionally, disbanding the unit “hurts our ability to acquire the critical knowledge and skills that can be gained from working with and learning from Ukrainian soldiers.”

“We respectfully request the Army brief us on how it intends to continue learning lessons on drone warfare from Ukraine in the absence of the Unmanned Assault Battalion. We request this briefing to occur no later than September 21st, 2026,” the letter concludes.

Why are drones changing the battlefield?

The conditions on the frontlines of Ukraine, often compared to the trench warfare of the Great War, have spawned a boom of specializations in drone technology.

From reconnaissance and electronic warfare to logistics and CASEVAC, drones are now essential to Ukraine’s war effort.

Numerous innovations in drone technology are keeping Ukraine in the fight. Autonomous drones capable of hunting and destroying Russian Shahed drones have been used to great effect, and supply lines across open ground are being bolstered by ground robots capable of lugging 210 kilograms cross-country.

But not all drone innovations are in Ukraine’s favor. Ukraine’s military intelligence recently warned that Russia is developing new drones designed to cut electricity cables to cause power cuts through the winter months.

So there is a valid level of concern over the disbanding of the US drone battalion. If the US army has no way to deploy or test systems within a dedicated drone unit, it may miss out on vital lessons that could shape future conflicts.

Sometimes you need a standing desk; sometimes you need a small standing desk; sometimes you want a certain color; and sometimes you just need a phenomenal foundation to build on to create your dream desk.

That's how the Vernal Core 3 came to be for me. I have a team member who had a triple-monitor setup on a tiny little desk that wasn't stable, and overall it looked quite sad.

When I had the opportunity to check out the Vernal standing desk after seeing everything in their commercials and learning about the company, I decided it was absolutely worth it to see if it lives up to the hype so far. 157 days in, we've been using the desk for probably 150 of those, and so far there have been absolutely zero problems. It has worked like an absolute dream.

The solid wood desktop is beautiful. The walnut is crisp and gorgeous. The legs are smooth, and the control panel has worked great. We've been able to build out exactly what this team member has always wanted in their desk, all made possible by the desk's high quality.

One of the reasons why I knew I wanted Vernal for this was its newly promoted stability in standing desks. The company recently updated its internal framing to make it even more stable than ever.

What better way to test that than with a shelving system resting on the table and triple monitors that, if not for a stable desk, would rock and shake back and forth nonstop?

Vernal Core3: Price and availability

The Vernal Core3 standing desk starts at $510 from the US store, and £470 from the UK store.

But, as usual with high-quality models, if you choose a more solid options on the site, you can get up to around $1,000.

Still, for what this desk is, it's a pretty reasonable price. Some of the desks that I've had, made of high-quality materials like this, have cost twice that or even three times that. So, overall, good value for money.

Vernal Core3: Unboxing and First Impressions

The Core3 desk is simple, refined, and professional-looking, and this week the walnut desktop is sharp. The integrated style keypad is pretty good-looking. The legs are based on a widely used industry standard design. Most of what makes Vernal tick, though, is transparency and the drive to make something that works and keeps working well.

For this specific desk build, we needed a solid foundation because we knew we wanted a unique triple-monitor setup that would require high stability due to the monitor arm tension. Plus, I knew we wouldn't want extra space and stability for a shelf we're gonna add. Having a flat underside also allowed us to add features like our undermounted laptop dock.

Vernal Core3: In use

For the last 100+ days, this desk has been an absolute rock-solid companion to my team. It has not had any issues going up and down. It has not rocked or faltered in any way. We've been able to build out everything that we hoped for and more.

The stability of the dual-stacked monitor arms, with the left mount to the side and a vertical monitor, has worked great. The extra weight from the shelf has not caused any issues at all.

So far, we have absolutely loved this desk. It looks great, works great, and after using it for some time now, I would recommend it for future use as well. The desk is quality, the price is right, and the size is massive, so if you are looking for a large standing desk, this one is definitely worth considering.

The controller is easy to use, and the presets make it easy to save your favorite or most frequently used heights and keep them easily accessible. Having presets makes all the difference.

Yes, there are cheaper desks that are still technically standing that have a crank or have an up-and-down electric button, but having a preset means you can press it and know it goes to the exact height you need whilst sitting or your perfect standing height, so you can get situated while the desk then meets you where you want it.

Vernal Core3: Final verdict

In short, this is a desk you should consider if you have a knick-knack or 17 on your desk and you're still trying to find a way to organize your series workflow. This is the kind of desk you want if you are into gadgets. This is also the kind of desk you should consider if you have a few gadgets on your desk and want to project a professional image.

However, if you're the kind of person who just throws things on their desk, doesn't really care, or is going to stay seated at all times, then you can probably find another desk that fits your budget and desires a little better.

This is the desk meant for those who like to be organized, get things done, and do it in style. There are a handful of areas that always make sense to spend a little more, and in my view, I stay in this because I'm quickly becoming one of them.

For more top picks, see our guide to the best standing desks we've ever tested.

• AI data centers are creating a much needed boom for the American trucking industry
• But the boom might not last for long as the requirements for truckers fall significantly once construction finishes
• Moratoriums and poor opinions of AI data centers could also quell the demand

American truckers are being hired by AI data center developers to ship goods and construction materials to rural data center projects, with the trucking industry across the US seeing a boom in demand.

As a result of this, truckers have a significantly higher opinion of AI data centers than their compatriots. But this freight hauling boom is having some knock on effects on truckers themselves, the businesses they work for, and the rest of America.

A shortage of truckers is boosting pay, but the price of fuel - which has been steadily increasing since the outbreak of the war with Iran - and the increased maintenance that comes with an active fleet are costing companies a fortune.

Will the trucking boom last?

With many AI data centers being built in rural areas where land prices are cheap, much of the construction materials, components, and generators are being taken by road in regions where infrastructure isn’t built to sustain such traffic. This is having a knock-on effect on local trucking with fleets being hired to help upgrade local infrastructure to meet new demand.

In the short-term, the AI data center boom has been good for business, and has provided many trucking companies with some stability in work. The war in Iran has pushed fuel prices to new all-time highs, with diesel reaching $5.85 a gallon, making it increasingly expensive to fuel fleets of trucks especially at a time when demand for trucking has dropped substantially.

But given the mass opposition to AI data centers being seen across the US, and the accompanying moratoriums on construction multiple states are putting into place, the boom may be somewhat short-lived in the long term.

This is especially true once the current demand for capacity is met. Once a data center is built it only requires a small workforce, and the demand for freight drops off a cliff.

AI companies, take note

AI data centers are one of the key topics of the US midterm elections. Candidates opposed to data centers are seeing surges in support, while some of the biggest AI companies are funneling money through superPACs into candidates likely to support the construction of data centers.

This is hardly helping the reputation of AI and the companies behind it, with new attention being drawn to the money that often decides the direction and policies of American politics.

The trucking boom is therefore a lesson for AI companies to learn. If you provide actual, tangible benefits to people's lives and jobs, opinions improve.

Via CNBC

• Liquid Network hack exploited a bug in SideSwap, releasing 3,998 BTC (~$313M) to attackers
• Hackers claimed “white‑hat” intent, returning 3,400 BTC after fixes, leaving 598 BTC missing
• Network remains paused as Blockstream and Federation patch vulnerabilities and prepare safe restart

The latest twist in the Liquid Network Bitcoin hack is worthy of a short movie, if not a full-length feature film.

Apparently, the hackers are actually the good guys, who stole the money to “keep it safe” until a vulnerability in the protocol had been fully resolved. They promised to return the funds afterwards.

What is Liquid Network?

Liquid Network was designed to solve a specific problem on the Bitcoin blockchain - being rather slow. The transactions on the network are recorded in a “block”, which is added to the chain roughly once every 10 minutes. Also, each block can only hold a limited number of transactions, which means the network can handle a smaller number of transactions per second, compared to conventional payment systems. Transactions that don’t make it into a specific block then need to wait for the next one, thus extending the confirmation time.

To solve that problem, Liquid Network was built. It runs its own Bitcoin reserve and its own blockchain, also known as a “sidechain”. When a person wants to use Liquid Network to send money quickly, they first convert their Bitcoin into Liquid Bitcoin, or L-BTC. They can then send it to another person much faster than a regular Bitcoin transaction. The recipient can keep the money on Liquid or convert it back to regular Bitcoin and move it to the Bitcoin network.

This way, Liquid provides a faster network without requiring every transaction to happen directly on the Bitcoin blockchain.

Besides speeding up Bitcoin transactions, Liquid Network also allows users (companies and other entities) to create and trade other digital assets, including tokenized securities, or stablecoins.

The project was built by Blockstream, a Bitcoin-focused technology company founded in 2014. It is run by the Liquid Federation, a group of more than 80 member companies, including exchanges, infrastructure companies, and financial institutions. A smaller group of members (15 to be exact) operate the network’s “functionaries” (servers that keep the network running), while the wider group has a governance role. Members vote on three boards (Technology, Membership, Oversight), handling ideas such as technical direction, internal rules, memberships, and more.

What happened to it?

In early September 2026, still unidentified actors managed to generate around 4,000 L-BTC, without putting in the corresponding 4,000 BTC into Liquid. As soon as they did it, they sent the L-BTC through SideSwap, a legitimate service that is used to convert L-BTC back to Bitcoin and vice-versa. The system apparently regarded the withdrawal as legitimate because, as was later determined, a bug prevented SideSwap from distinguishing between “real” and “fake” L-BTC, and treated them all the same.

As a result, the Liquid Federation ended up releasing 3,998 real BTC (more than $313 million at press time) to the attackers. As soon as the operators realized what had happened, they halted new transactions and warned about possible disruptions until the service was restored.

Then came the movie twist: the hacker started communicating with network maintainers through on-chain Bitcoin transactions, promising to return the funds when the vulnerability is fully resolved:

“Please fix the bug first. The chain is under risk at latest commit right now. Make sure every node is patched. Then we will transfer the money back safely after confirming the fix,” one of the messages read.

Earlier today, Coindesk reported that the hackers partially kept their promise, returning 3,400 of the 4,000 BTC drained, which suggests that the flaw was remedied. The remaining 598 BTC, worth approximately $47 million, is currently unaccounted for.

“3,400 BTC of the roughly 4,000 BTC withdrawn on September 6 has been returned to the Liquid Federation wallet. The return followed confirmation from Blockstream that the affected bridge nodes have been patched,” wrote Samson Mow, former chief strategy officer at Blockstream, on X. “Approximately 598 BTC remains outstanding, and Blockstream continues to engage with the white-hat hackers.”

He added that the network remains paused while Blockstream and Federation members make additional fixes and security improvements, resolve the chain split, and prepare for a safe restart.

“Liquid wallets and services will continue to be affected during this time. No user action is needed, and please do not send Bitcoin to Liquid peg-in addresses until we confirm the network has restarted,” he added.

• Microsoft 365 users targeted by phishing campaigns using BigBear 2.0 and AiTM proxies
• Attackers impersonate IT staff via calls, Teams, and email to steal credentials and bypass MFA
• CloudSEK tracked 5,000+ stolen records; Arctic Wolf urges phishing‑resistant MFA and conditional access

Microsoft 365 users are facing a barrage of cyberattacks all aimed at a single goal - to try and expose credentials which can later be used against employers in data theft attacks.

Different groups conduct their raids and bypass multi-factor authentication (MFA) protection to access victim accounts - CloudSEK, for example, said that some groups are using BigBear 2.0, a new phishing-as-a-service (PhaaS) framework that allows crooks to intercept passwords and authenticated session cookies.

Arctic Wolf, on the other hand, focused on a single threat actor, which it dubbed PREY-0058. This group, despite significant overlaps with other collectives in terms of techniques, technologies, and procedures, is not a rebrand of older organizations. Instead, the researchers believe the lines between the groups are blurred and that there is a large group of affiliates, splinter crews, and other cohorts using the same phishing infrastructure and thus often confuse defenders and analysts.

Similar methods, similar results

The attack methodology is similar across the spectrum. Crooks would call their victims on the phone or approach them via Teams and email. They would introduce themselves as members of the IT help desk sent to sort out a specific problem or issue.

Then, they would either convince the victim to grant remote access, or to open a spoofed Microsoft 365 login page and enter their credentials there. In both cases, the goal is the same - to get the victim to type in their username, password, and 2FA code, on a fake site built by BigBear 2.0 or a similar phishing framework. This framework, using an attacker-in-the-middle (AiTM) proxy between the victim and legitimate Microsoft infrastructure, harvests credentials, MFA codes, and session cookies, and replays them through an API essentially hijacking a legitimate authentication session.

Once they gain access, the attackers can do all sorts of things, but they are mostly focused on exfiltrating sensitive data from Outlook, Teams, SharePoint, and OneDrive. Deploying ransomware is rarely seen.

The campaign CloudSEK has been tracking has been rather successful, the researchers argue, saying BigBear 2.0 was used to exfiltrate more than 5,000 credential records, “including 474 complete MFA-bypassed authentications, 1,032 plaintext passwords, and 4,148 session cookies - affecting 3,331 unique victim IPs across 40+ countries with the operation still active at the time of writing.”

“The multi-user PhaaS panel is leased to at least five affiliate operators identified through live Telegram exfiltration bots, each receiving stolen credentials in real time.”

Speaking to BleepingComputer, CloudSEK says the campaign targeted 461 organizations, out of which 258 have had at least one set of credentials compromised.

Defending with phishing-resistant MFA

Arctic Wolf’s researchers stressed that the attackers are focused primarily on US-based businesses: construction and engineering, healthcare and pharmaceuticals, real estate and property management, finance, and professional services, The Hacker News reported. The researchers advise organizations to implement Conditional Access policies, deploy phishing-resistant MFA, and restrict the scope of data users can access via SharePoint. Obviously, employee education on the dangers of phishing cannot be understated.

"Defenders can disrupt this activity by detecting anomalous residential-proxy token replay, SharePoint discovery and bulk access, mailbox harvesting, and newly registered authentication-themed lure infrastructure," Arctic Wolf said.

Phishing-resistant MFA is multi-factor authentication designed so that an attacker cannot trick a person into handing over authorization code, either via a message, or through a fake login page.

These include products such as passkeys, YubiKeys security keys, and authentication methods based on FIDO2/WebAuthn. Since phishing-resistant MFA cryptographically ties the authentication to the legitimate website, the authentication cannot simply be forwarded to an attacker.

• 3,000 Swiss government devices will use open source software instead of Microsoft 365
• 172 workers already tested Germany's openDesk software in an earlier proof-of-concept
• The next stage should be complete by the end of 2027, at the cost of $11.1 million

Switzerland's federal government has launched a pilot program to explore the feasibility of ditching Microsoft 365 in favour of open source alternatives, and it's starting with 3,000 machines of the government's estimated 54,000 total computers.

The government is looking to move around 7% of its machines away from M365 by the end of 2027, the chancellery confirmed, but it's not the first time the Swiss government has explored open-source alternatives.

An earlier proof-of-concept, the details of which have just been released, demonstrate 172 federal employees testing German-made openDesk for document processing and emailing.

Switzerland is progressing with its open-source plans

Crucially, the government doesn't look to be removing M365 outright from all 3,000 machines, but rather, workers will be given openDesk to run in parallel in order to avoid any disruptions.

Central to the migration isn't necessarily an anti-Microsoft initiative, but rather a pro-digital sovereignty push. Switzerland wants critical government operations to be less dependent on foreign tech companies, especially US tech giants given ongoing geopolitical tensions.

US cloud legislation potentially allows American authorities to gain access to Swiss government information, and with Microsoft being an American firm, its 365 platform is naturally under assessment.

More broadly, Switzerland isn't the only country to be walking away from Microsoft and other US tech vendors in favor of local, open source alternatives. France and Germany have also introduced similar schemes.

Microsoft is already working on a solution to pacify Swiss sovereignty requirements - in 2025, it announced a $400 million investment in Switzerland to expand its cloud and AI infrastructure, with a major focus on government, finance and healthcare customers.

The ongoing project covering 3,000 devices is expected to cost $11.1 million, per the chancellery, with funding coming from pro-sovereignty initiatives.

Building is no longer the constraint.

AI-assisted development has transformed how software gets built. Because tools can generate code, build prototypes and speed up testing, ideas that once took weeks or months to develop can now be explored in a fraction of the time.

Engineering capacity is no longer the constraint it once was. More organizations can experiment and bring new products to market faster than ever before. But building faster doesn't guarantee better products.

Human judgement is still necessary to decide which ideas deserve investment, which features solve real customer problems and which experiments aren't worth pursuing.

AI can dramatically shorten the journey from idea to release, but it can't make those types of business decisions for you (yet).

Research from MIT found that while AI has fueled a surge in new apps entering mobile marketplaces, usage has not increased at the same pace. Faster development is producing more software, but it isn't creating more customer attention. Customers haven't suddenly found more hours in the day simply because software is easier to build.

That raises the bar for decision-makers. As AI removes many of the barriers to creating software, success increasingly depends on learning quickly from customer behavior and understanding what customers genuinely value. It’s the only way to stand out in an increasingly crowded marketplace.

Customer insights are your biggest asset

After every release comes the same questions. What next? Do we invest further or move on? AI has made it possible to build and ship software more quickly, but it hasn't made those decisions any easier.

That's why product intelligence matters. Understanding how customers behave gives decision-makers the confidence to decide where to focus their effort. Behavioural data shows what customers keep coming back to, where they struggle and where they give up. It also helps distinguish between features that attract initial interest and those that become part of customers' everyday workflows. That distinction often says more about long-term value than launch-day engagement or anecdotal feedback.

It also reveals where customers complete key tasks, where they struggle and where they abandon journeys. Those signals often provide stronger evidence than customer opinion alone, because they reflect what people actually do rather than what they say they do.

Those insights make prioritization easier. They show which features deserve more investment, which ideas aren't landing and where the next opportunity probably sits. They can also help product teams decide when to refine an existing feature, simplify an experience or stop investing in something customers aren't using. Product decisions become grounded in how customers actually use the product, rather than internal assumptions and beliefs.

As AI lowers the barriers to building software, deeply understanding customers becomes even more valuable. The strongest organizations keep learning from the people using their products, allowing every release to build on real customer insight.

From AI-assisted coding to AI-assisted product improvement

AI-assisted development has changed how software is built, and product leaders are only beginning to explore what it can do beyond writing code. The next stage is using AI alongside product intelligence to strengthen the decisions that shape a product over time.

When working with large volumes of behavioral data, AI can help surface patterns more quickly, highlighting changes in customer behavior, unexpected user journeys and emerging trends that might otherwise be overlooked.

That gives decision-makers more confidence about where to focus their attention, while leaving more time to interpret what's happening and determine the best response. Rather than spending hours searching dashboards for answers, product teams can focus on understanding the behavior, testing possible improvements and deciding which ideas are worth pursuing.

Every release generates new behavioral data. Instead of relying on assumptions, product management teams can use that evidence to decide what deserves attention next. Over time, each release helps inform the one that follows.

Used in this way, AI becomes part of the product improvement process as well as the development process. The combination of AI and behavioral insight helps organizations learn faster, respond with greater confidence and keep building products around what customers actually need. Focusing on what the data uncovers about what customers actual need is the best way to endear your product to your users.

AI has changed the speed of product development. But every release still depends on good product judgment, which means knowing what to improve, what to leave behind and where to invest next. Behavioral insight is the best guide for decision-makers to ensure that every release reflects what customers actually do rather than what your internal teams assume they'll do.

We've featured the 10 best vibe coding tools.

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

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

• Logitech unveils $99 MX Keypad accessory
• Sitting alongside your keyboard, it offers nine customizable keys to give access to AI tools and agents
• Users can even create and share their own integrations

Logitech has launched a new $99 accessory it says will help programmers and coders work faster and more effectively than ever before.

The new MX Keypad offers nine customizable full-color LCD keys which can be assigned to launching commands, running macros, and accessing AI tools.

It's also very compact, around the same width as a Logitech mouse and weighing just 96 grams, meaning it will sit alongside a user's keyboard without taking up too much space on your desk.

Faster and more effective access

Logitech says the MX Keypad, developed alongside GitHub and available in two color schemes, will act as a control center for users, aiming to cut down on switching between apps - although the customization options can also be used for non-coding apps.

Any of the nine keys can be configured to control one or multiple apps and agents, from running complex prompt macros to launching everyday commands, with native GitHub, VS Code and Copilot integration at launch.

Using the Logi Actions SDK, developers can even create and share their own integrations, while community-built plugins for tools including Claude Code and OpenAI Codex are already available, and plugins created and tested by developers for GitHub Copilot and Claude Code.

The MX Keypad connects using USB-C and requires macOS 13 or later, or Windows 10 or later. As is typical with Logitech devices, sustainability is also a key feature, with the MX Keypad features certified post-consumer recycled plastic, 64% for graphite, and 49% for its pale gray colored models.

“Today, coding mastery depends on how easily developers can juggle between an ever-increasing range of AI tools,” said Anatoliy Polyanker, VP and GM at Logitech. “MX Keypad puts a physical control center at your fingertips, allowing you to orchestrate your AI workflows faster and more intuitively.”

The MX Keypad looks to offer a more streamlined version of Logitech's existing MX Creative Console, which provides an LCD keypad with a separate dialpad to adjust settings and move around your favorite creative apps, but costs $199.99.

• Half of UK CIOs say they're being held accountable for AI agent mistakes
• Bigger organizations, governments and public sector blame their CIOs more
• Tech leaders are also being asked to consider digital sovereignty while navigating vendor lock-in

New research has found half of UK CIOs say they're personally accountable when an AI agent makes a mistake - even though they may be little more than a figurehead for an entire department.

Interestingly, the bigger the company, the more personal accountability a CIO may have, despite likely having more workers and managers within the team, the report from 8x8 found.

Among companies with 50-99 workers, 45% of CIOs noted personal accountability for AI agent errors, but this grew to 62% among companies with 500+ employees.

CIOs say they're being held accountable for AI agent errors

Government and public administration organizations were found to put the biggest pressure on CIOs, with the figure sitting at 86%.

But more broadly than just facing the blame, CIOs are also struggling with data sovereignty. Two in five (39%) say the location of AI infrastructure has now become the primary factor behind which platforms they use, with questions about where data is stored and processed appearing both at the beginning of the procurement process but also at broad level.

"Sovereignty is the opening question in almost every conversation we're having with customers here," EMEA MD Jamie Snaddon wrote.

Nearly all (91%) UK technology leaders agree digital sovereignty has become a higher priority this year, with vendor lock-in, cost and complexity also influencing decisions.

With all of this added pressure, plus personal accountability, stacking up against the role of the CIO, the job is becoming riskier and less desirable overall. 8x8's research suggests that, with this current approach at least, CIOs may be more cautious adopting AI and could even be less willing to take on the job in the first place.

2025 was a transformative year for artificial intelligence, and 2026 is already proving to be equally significant.

While generative AI dominated conversations just a few years ago, the focus is now shifting towards agentic AI, where intelligent systems can take action, interact with business processes, and support employees in more meaningful ways.

As these capabilities continue to evolve, organizations are looking for better ways to connect AI systems to the applications, data and services that drive their operations.

This is where technologies such as Model Context Protocol (MCP) are becoming increasingly important.

Rather than creating bespoke integrations for every tool or system, MCP provides a standardized way for AI models and agents to access information and perform actions across an organization's technology estate.

While MCP is not a requirement for every AI implementation today, it represents a natural next step for organizations looking to move beyond isolated AI use cases and towards more integrated, scalable AI ecosystems.

However, greater integration also introduces greater responsibility. Effective governance remains essential for any AI deployment, but it becomes even more critical when autonomous agents are granted access to business systems, processes and sensitive information.

Organizations must establish clear guardrails that define what agents can access, what actions they can perform, and how their activities are monitored. Without appropriate oversight, businesses risk agents operating beyond their intended scope or creating unintended consequences across interconnected systems.

The impact on Software-as-a-Service

Few sectors will feel the effects of this shift more than Software-as-a-Service (SaaS).

For years, SaaS applications have been built around human interaction. Users access platforms through dashboards and interfaces, navigate predefined workflows, and manually complete tasks. The application itself serves as the primary workspace where work is performed.

Agentic AI introduces a different model.

Rather than navigating software in the same way a person would, agents can interact directly with APIs, services and data sources. This allows them to retrieve information, execute actions and orchestrate processes across multiple systems without relying on traditional user journeys.

That does not mean SaaS applications will disappear. In fact, they will continue to play a critical role in storing structured data, enforcing business rules and managing workflows. What is likely to change is how those applications are consumed.

Instead of being the primary destination where work happens, many SaaS platforms will increasingly act as sources of capability and information that AI agents can utilize on behalf of users. As a result, organizations may find themselves focusing less on which application employees need to access and more on how services and data can be brought together to achieve the desired business outcome.

Human interfaces will still matter. Users will continue to need visibility, exception handling and control mechanisms, particularly when business-critical processes are involved. The challenge for software providers will be balancing traditional user experiences with new AI-driven interaction models while maintaining compatibility, reliability and operational resilience.

Breaking down agent silos

The next stage in the evolution of agentic AI is not simply creating more agents. It is enabling agents to work together effectively.

Many organizations already struggle with fragmented systems, disconnected data and isolated processes. Without careful planning, agents risk creating a new generation of silos, each operating within its own limited context and producing inconsistent outcomes.

To avoid this, businesses must focus on shared context, connected data and interoperable services. The goal is not to have individual agents automating isolated tasks but to enable multiple agents to contribute towards broader business objectives across entire processes.

When agents can access consistent information and operate across organizational boundaries, the value shifts from discrete task automation to coordinated execution. Rather than supporting an individual stage of a workflow, agents can participate in end-to-end processes while remaining aligned to business policies, operational requirements and organizational goals.

This represents an important architectural shift. Software increasingly becomes something that agents consume programmatically, while integration, context and orchestration become central to delivering outcomes at scale.

Mitigating risk in AIaaS

Unlocking these new capabilities requires more than deploying AI tools. Organizations need governance frameworks, security controls and operational processes that allow autonomy to be introduced safely and responsibly.

As agents gain access to more systems and collaborate across workflows, operational complexity inevitably increases. Businesses must define clear policies around what agents can and cannot do, what data they can access, and what approvals are required before actions are taken.

These controls should be embedded into the orchestration layer itself, ensuring governance is applied consistently across all agent-led activities rather than being treated as an afterthought.

Traceability and accountability are equally important. Completing a task successfully is only part of the equation. Organizations must understand how decisions were made, what information was used, and which policies were applied throughout the process. This visibility will be essential for compliance, security and maintaining trust in autonomous systems.

The role of developers will also evolve. Rather than spending significant time building and maintaining point-to-point integrations, they will increasingly focus on designing agent behaviors, defining boundaries, managing orchestration and ensuring solutions operate within established governance frameworks.

The organizations that succeed will be those that balance innovation with control. Too little governance introduces risk, while excessive restrictions can prevent businesses from realizing the benefits of AI altogether.

Agentic AI should not be viewed as a replacement for software development or existing technology investments. Instead, it represents a powerful new interaction layer that changes how organizations access information, automate processes and deliver outcomes.

Businesses that invest now in integration foundations, governance models and workforce skills will be best placed to take advantage of the opportunities this next phase of AI creates.

We've reviewed the best business laptops.

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

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

For three years, businesses have bought AI the way they buy Microsoft 365.

Flat fee, line item, sat next to the Zoom invoice and asked no difficult questions. The budget holders were happy, the IT team ticked a box - and somewhere in a product division, engineers started doing things with it that nobody in finance fully understood.

That era is over.

On 14 May, Anthropic announced it would split agentic usage out of its Claude subscription and put it behind a metered credit pool, effective 15 June. GitHub Copilot will move to token-based AI Credits on 1 June.

Legacy Anthropic enterprise seats are being retired at renewal. And a tokenizer change in Claude Opus 4.7 has already pushed some API bills up by as much as 27% - with nothing moving on the published pricing page.

The thing being counted has changed. Not the stated price per thing. That is a meaningful distinction, and most enterprise procurement teams have never had to think about it.

Most CFOs don't know any of this yet.

The receipts

The receipts, though, are starting to arrive. Uber burned through its entire 2026 AI budget by April. Its CTO told The Information he's "back to the drawing board." KPMG's latest survey has US enterprises projecting average AI spend of $207 million over the next twelve months - nearly double a year ago.

Goldman Sachs data shows large companies already overrunning AI budgets by orders of magnitude. Salesforce CEO Marc Benioff has said his company's Anthropic bill will run to around $300 million this year - and that he wished there were a "smart router" that could work out which queries actually needed the most capable, most expensive models, and which could be handled by something cheaper.

Meta took down the internal tokenmaxxing leaderboard its engineers had built and even Microsoft has cancelled Claude Code access for staff in several key product divisions.

The angle most coverage has missed is that this is fundamentally a buyer-side story. What's happening at Anthropic and OpenAI is interesting. What's happening at the other end of the invoice is more interesting. What does the finance director do when the AI bill doubles mid-year with thirty days' notice and no contractual recourse?

What does the GC do when she looks at the supplier agreement and realizes it was drafted on a SaaS template that has no concept of a tokenizer, no audit rights, no price-change notice period, and no exit provision worth the paper it's printed on?

The right analogy

The right analogy here isn't software but electricity. When you buy electricity, you know the rate per unit, you can read the meter, your contract has notice periods and change-of-tariff protections, and there's a regulator with views about what suppliers can and can't do quietly.

None of that IT infrastructure exists yet for AI consumption. Businesses signed deals when usage was flat-rate and predictable. Now the meter is running, and in many cases the contract gives them no visibility over how fast, no right to challenge the reading, and no meaningful exit if the numbers stop making sense.

What "AI ownership cost" looks like in a business that has never had a FinOps function is, frankly, a mess. Usage is distributed across teams, often unsanctioned, running on departmental cards that never touch central procurement. The AI bill isn't a bill - it's fifteen bills, scattered across expense reports and shadow IT budgets, none of them talking to each other.

The first time many finance teams see the true picture is when someone pulls the data together and says the number out loud. That moment, for a lot of businesses, is coming in August.

The companies now pulling back - limiting which employees can access agentic tools, restricting use of the most advanced models, quietly cancelling licenses - aren't doing so because AI has stopped working. They're doing it because the cost model has broken the business case, and the contracts give them no leverage.

Specific things buyers should insist on

There are specific things buyers should now be insisting on. Price-change notice provisions - real ones, not buried in terms and conditions - that require meaningful advance warning before a pricing architecture changes. Tokenizer stability clauses: a commitment that the method of counting consumption won't shift materially without renegotiation.

Audit rights over consumption data. Exit and portability terms that don't require a legal battle to invoke. Whether existing contracts have any teeth when the metering changes underneath them is a live question. My suspicion is that most don't, because nobody drafting AI agreements in 2023 or 2024 anticipated that the unit of consumption would be a moving target.

Timing matters. OpenAI's two-month Codex trial and Anthropic's 50% capacity boost both expire mid-July. The window of subsidized, high-capacity usage is open right now. The bills will start arriving in August. Finance teams that haven't yet had a reckoning with their AI spend are about to have one, whether they're ready for it or not.

Reasons for spending

The businesses that come out of it ahead aren’t going to be those who spent the least but those that knew what they were spending, why, and what they got for it - and whose contracts gave them somewhere to stand when the rules changed underneath them.

That requires something most AI buyers have never had to build: a proper supplier relationship, a FinOps function, and contracts written as though they were buying a utility rather than a software subscription.

The flat-fee era is over. The question is whether the contracts, the finance functions, and the legal frameworks are ready for what replaces it. For most businesses, they are not. But the August bills will focus the mind.

We've featured the best finance software.

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

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

• Microsoft dev claims the age of typing code is coming to an end
• AI agents are increasingly replacing humans when it comes to writing code
• But humans still have a crucial role in verifying outputs, as AI model accuracy still isn't quite there yet

Microsoft engineer David Fowler has claimed "typing code is absolutely over," arguing that traditional developers are quickly becoming obsolete in an AI-first world.

Sparking debate in the comments section with views split, the implications of Fowler's X post ultimately imply that developers' roles are evolving, rather than disappearing altogether.

Roles are quickly covering supervising and reviewing, with developers prompting AI agents, verifying outputs and finetuning, rather than writing new code from the ground up.

Developers aren't obsolete, but their role is undergoing major changes

But AI isn't just writing lines of code anymore – Microsoft-owned GitHub Copilot agentic tools have proven that AI can actually handle much more of the workflow, including building the software, running tests and checking for failures, before the project even gets handed to a human for review.

Microsoft is also making its own Aspire development platform increasingly agent-friendly, with version 13.2 introducing support for MCP and an agent-friendly CLI.

We already knew in 2025 that as much as 30% of Microsoft code was AI-generated, per CEO Satya Nadella's remarks, but that figure is likely much higher today.

Separately, a study by Microsoft's own researchers found that accuracy falls even for frontier models, the longer-running a task becomes. CodeRabbit also found that AI-generated code contains an average of 1.4x more critical issues and 1.7x more major issues than the equivalent human-generated code, with Veracode warning that 45% of AI-generated code contains security flaws despite appearing production-ready.

While models will undoubtedly get better as generations progress, mistakes will likely remain a constant throughout, so while Fowler's claim that typing code is over, the role of the human developer certainly isn't.

81% of UK knowledge workers now use AI weekly. That's adoption. What it isn't, is transformation.

Most organizations have layered AI onto broken processes and fragmented systems and called it progress.

Meanwhile, 82% of UK IT leaders have absorbed unexpected AI cost increases, and 58% report high adoption with limited measurable productivity gains.

We have a usage problem dressed up as a strategy.

Here's what's actually going wrong and what needs to change.

Adoption without redesign is theatre

AI doesn't fix bad processes. It accelerates them. If your data is fragmented, your ownership is unclear, and your workflows are inefficient, deploying AI makes those problems faster, not smaller.

Real value requires asking harder questions: Where do decisions actually get made? Which processes should fundamentally change? Who owns the outcome? Until you answer those, you're generating AI activity, not business impact.

Own the outcome or don't deploy

Nearly two-thirds of UK IT leaders say they're fully accountable for AI-driven business outcomes, while AI deployment is spreading across departments, often outside governance structures. That's a recipe for accountability without visibility.

IT management sets the framework. That's necessary. But every business leader who owns a process needs to own how AI changes that process. What does success look like? Who monitors it? Who's responsible when it goes wrong?

If you can't answer those questions before you scale, don't scale.

Shadow AI is a signal, not just a risk

One in four UK workers use unapproved AI tools. The instinct is to lock it down. The smarter read: your people are telling you your current tools create friction, and they've moved on without you.

Restriction isn't a strategy. Channel that demand toward trusted tools with real governance, then use governance as an accelerant, not a brake. The organizations moving fastest are the ones that treat low-risk use cases as low-risk, and reserve serious scrutiny for high-stakes applications.

Context is the missing layer

Nearly half of UK IT leaders say AI initiatives stall because AI lacks organizational context. That's not a technology problem, it's a work infrastructure problem.

Think about how you'd onboard a new hire. You'd give them the org structure, the priorities, the decision rights, the rules. An AI agent needs the same. Without it, even capable models produce output that someone has to spend 30 minutes correcting, which is exactly what's happening.

The fix is connecting AI to where work already lives. Not asking employees to reconstruct context every time they open a prompt.

Measure outcomes, not usage

If your AI metrics are licenses purchased, prompts submitted, or hours theoretically saved, you're measuring the wrong thing. The question is whether the work is improving.

Are customer issues resolving faster? Are teams spending less time searching for information? Are the right decisions getting made with better speed? At Asana, we built an AI seller assistant and measured its impact on the sales process, response rates, net-new meetings booked. That's the bar.

The accountability question is only going to get harder

Agents are coming. Systems that act on behalf of people, not just assist them. When that happens, organizations will need to know: which agents exist, who created them, what they can access, what they're authorized to do, and how their performance is tracked.

This isn't a future problem. The organizations building that discipline now will be the ones who can scale agentic AI without the governance catching up after the fact.

The businesses pulling ahead won't be the ones using the most AI. They'll be the ones who've connected it to clear ownership, proportionate governance, and the workflows where execution actually happens.

Adoption is table stakes. Value is the real work.

We've listed the best productivity tools.

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

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

• Just 52% of Brits are familiar with the concept of a data center
• Positive sentiments are down, while negative perception is up
• This report calls for greater transparency and public engagement

New research has revealed only one in two (52%) Brits are aware of the term data center, but despite growing awareness (with the figure up from 40% in 2024), three in five (59%) are still unsure about what a data center is or what it does.

Despite slow growth in awareness and a UK designation for data centers as Critical National Infrastructure (CNI), fewer people believe data centers have a positive impact on the digital services they use both at home and in work.

Today, only 37% of British citizens see them as having a positive impact, down from 48% two years ago. On the flip side, negative perception has risen from 4% to 15% in the same period, the report from Telehouse found.

British citizens see data centers in a negative light

Among the biggest concerns is energy consumption, with more than half (56%) of the respondents worried about the amount of electricity these campuses use.

But with awareness of the concept in general rising year-over-year, Telehouse believes there's scope to turn this awareness into understanding and confidence. "As an industry, we have a role to play in closing that gap by working more closely together, being more transparent, engaging openly with the public and communities and doing a better job of explaining the value data centres enable," EVP and GM for Telehouse Europe, Mark Pestridge, explained.

The report indirectly alludes to the fact that worker and consumer AI use is disproportionate to education, which explains why public scrutiny may be accelerating more quickly.

When the UK government designated data centers as CNI, the then-Technology Secretary Peter Kyle described them as "the engines of modern life."

"The objective now is helping people understand why data centres matter and how it supports everyone’s lives," Pestridge added.

• 36.7% of UK workers say answering to an AI boss would improve their productivity
• However, almost 75% of those workers are happier answering to a flesh-and-blood superior
• Meanwhile, almost 40% of managers don’t believe that AI can deliver any productivity gains

Over a third (36.7%) of UK employees think an AI boss would mean an improvement to their productivity, with 64% of directors expecting similar gains, new research has found.

But despite these results looking good for AI, however, it seems many people prefer the idea of performance reviews being delivered face-to-face, rather than via chatbot.

However, there is one key opponent to AI managers. The human-based middle manager layer is unsurprisingly less swayed by the prospect of being replaced by AI, with almost 40% feeling productivity gains are unlikely.

Flattened structures?

The research from Careerminds UK contacted 600 full-time UK employees and asked them about their thoughts on AI in the workplace, also found that younger colleagues were more likely to be favourable to an AI management layer.

The survey comes at an interesting time, with predictions of flattened structures seemingly focused on the middle manager layer. If businesses feel that they can adopt AI to replace managers – particularly at a time when management training seems to be at a premium – then they need to act carefully.

Productivity boosts are one thing, but employees also feel a human superior is preferable to AI for things like delivering performance feedback (74.2%).

“AI may be able to ease some of that pressure by taking certain tasks off managers’ plates,” said Amanda Augustine, resident careers expert at Careerminds UK and a Certified Professional Career Coach (CPCC). “But the feedback findings make it clear that employees still value human judgement, context and connection.”

The artificial middle manager

While the productivity boost expected by 36.7% of those surveyed (versus 25.5% who feel they would be less productive) might seem encouraging, there is a challenging aspect to the report’s findings.

Some 64% of directors and above expect an AI manager to deliver productivity gains, with 28% of those expecting it to be “significant.” Among senior staff, the figure is 42.6%.

Among managers, that 38.5% figure is worrying, as it implies almost 40% of managers are expecting AI to replace them. Will the C-Suite have its way? Should it?

“We shouldn’t confuse employees seeing potential productivity gains from AI with wanting technology to replace their managers altogether. Middle managers are already feeling the squeeze, caught between growing expectations from senior leaders and the needs of their teams," added Augustine.

“The opportunity for employers is to use AI to help managers become more effective, not simply replace them”.

At a time when organizations are increasingly going hard on AI, some recognition of what employees need to succeed is important.

Rather than focusing on efficiency, Augustine observed leaders need to think about their employees, particularly those who have just started their careers who are “more likely to think about what they need from a manager day to day, whether that’s guidance, development, reassurance or context.”

• Report claims an incredible 80% of AI tools are running in organizations without IT oversight
• Browser agents and integration tools are escaping the attention of security and IT engineers
• Reco’s State of Agent Security 2026 report also tracked 637 vulnerabilities across agents and LLMs

Unmonitored deployment of AI tools is a risk to security, and puts data at risk, a new study has claimed.

The report from Reco, which draws its information from disclosed vulnerabilities, analysis of 500 Model Context Protocol servers, and Reco’s own platform telemetry, found four in five AI tools (80%) are running without oversight from IT departments.

Of particular concern is the scale of AI applications in use. Smaller companies use 414 AI tools per 1,000 employees without IT approval, apparently a combination of browser extensions and workflows beyond the usual review and approval process.

Data risks from unmonitored AI

Giving AI tools to employees might unlock productivity boosts, but their use has to be approved. That’s the key takeaway from the report, which highlights some concerning cybersecurity figures. For example, it assessed 500 agent tools and found “62% can both read local data and reach the internet.” This represents an opportunity for data exfiltration.

Elsewhere, 637 AI agent related vulnerabilities were identified in the report.

The adoption of AI is wider than specific use of a SaaS application or visiting ChatGPT. Reco found that AI agents are running within other tools, and inheriting user permissions. The implications of this are clear.

Operational risk

Analysis of the telemetry (gathered from 62 enterprise-scale businesses in financial services, healthcare, retail, and telecommunications between January 1 and August 1 2026) reveals a free-for-all attitude towards AI adoption. While businesses may have policies and procedures in place and processes to assess, evaluate, review, and finally approve new AI-based tools, these are being circumvented for low-level applications.

The rules work for SaaS procurement oversight, but not for browser extensions, and the result is “operational risk.”

“AI agents have moved from experimentation into daily business workflows, but our findings show only 20% of AI tools in enterprise ecosystems are currently governed by IT oversight," Reco CEO Ofer Klein noted.

“That leaves organizations exposed to a new class of operational risk. Agents embedded in applications can operate through existing permissions, OAuth grants and workflow access, creating toxic combinations that expose data and trigger actions beyond what any owner approved.”

Organizations will need to give IT teams the resources they need to manage and restrict unauthorized AI use, as the alternative means leaving the gates open to the possibility of data exfiltration.

• GMKtec NucBox M5 Ultra pairs an eight-core Ryzen 7 7730U with dual 2.5GbE, Wi-Fi 6E, and triple-display output while starting at just $330 a pop for a barebones configuration
• The underlying 2023-release CPU is capable, but a rebadge of the 2022-launched Zen 3 Ryzen 7 5825U, which is, in turn, a 2021-launched 5800U with a 100 MHz bump to base and boost clocks
• The price can be reduced to as little as $280 per unit for users buying 5 of the RAM- and SSD-less SKU

GMKtec has put its NucBox M5 Ultra back in the limelight, and the offering is familiar for the Shenzhen mini PC maker's mid-range lineup.

It comes with an eight-core AMD processor, two 2.5 Gigabit Ethernet ports, upgradeable memory and storage, and a price starting just over $300.

The catch for any prospective buyer is that you must bring your own DDR4 laptop RAM and SSD to get it going.

A capable NUC that is hiding its underlying CPU's age

The M5 Ultra runs the AMD Ryzen 7 7730U, an eight-core, sixteen-thread part clocked from 2.0 GHz to 4.5 GHz with 16 MB of L3 cache, built on TSMC's 7nm process with a default 15W TDP. Despite the 7000-series badge, it belongs to AMD's Cezanne lineup via the Barcelo-U refresh announced in January 2022.

The CPU it most closely matches in clock speeds, core configuration, and performance is the Ryzen 7 5825U, which is a slightly higher-clocked version of the January 2021-released Ryzen 7 5800U, which runs 100MHz lower on both base and boost clocks.

This is interesting because prospective buyers might feel they are buying a Zen 4-based CPU thanks to its 7000-series badge but would end up getting a CPU that was first available more than half a decade ago, even if it does hold its own in its price and performance segment.

A well-rounded offering with plenty of expansion under the hood

The M5 Ultra offers two DDR4-3200 SODIMM slots supporting up to 64 GB, and two M.2 2280 slots, allowing for a variety of configurations based on the user's needs.

It does have its limitations due to the underlying CPU, however: while GMKtec claims each M2 slot can support up to 8 TB, for a nominal 16 TB ceiling, speeds are capped at PCIe 3.0 because the 7730U provides only 16 usable lanes and no Gen 4 support out of the box.

Connectivity is where the unit particularly shines: it comes with two Realtek RTL8125 2.5GbE controllers, Wi-Fi 6E and Bluetooth 5.2, and a 10 Gbps, 100W USB port in addition to two USB 3.2 Gen 2 Type-A ports and two USB 2.0 ports, a very formidable offering for a Nuc-tier PC.

The system, which is powered by a 65W PSU over a Type-C connector, also offers HDMI 2.0, DisplayPort, and USB-C for up to three simultaneous connections for power users looking to run a multi-display setup.

One could argue that pricey RAM and SSD options make sourcing them considerably less appealing and a tall order, but even with that in mind, at today's prices, the NucBox has plenty of connectivity under the hood, decent power, and a raw customizability that is hard to argue with in 2026.

• Think tank Ceres found power generation for data centers in seven states withdraws 3.4 trillion gallons of freshwater a year, or 10.4 million acre-feet
• The figure might be a bit overstated, as it counts withdrawals, including hydropower water returned to the river immediately
• 78% of electricity in those states comes from water-using plants, and 66% of those plants sit in water-stressed areas

Non-profit think tank Ceres has published a new report analyzing the water used to generate electricity for data centers across Virginia, Texas, California, Illinois, Georgia, Ohio, and Arizona.

Ceres estimates data centers use 3.4 trillion gallons of freshwater a year, or 10.4 million acre-feet.

Bloomberg Law's Bobby Magill compared that unfavorably against the 8 million acre-feet California uses annually for urban purposes, a figure from a 2021 Legislative Analyst's Office report.

An important difference withdrawal versus consumption

Ceres's approach makes sense, but it paints only a partial picture at best. It measures water withdrawals, or the total volume of water a power company pulls from a water source like a river, lake, or aquifer, including water it essentially replaces shortly afterward.

For context, the LAO's 8 million acre-feet is a different quantity: water actually delivered and applied to residential, commercial, industrial, and landscape uses across California, and it is unlikely to be channeled back to its source.

Ceres does note this, stating that "withdrawal reflects the full volume an asset depends on to operate", which it presents as a strong indicator of risk exposure to multiple communities that are already living in drought-affected areas.

A large amount of the power being generated and measured is hydroelectric power. This makes it a bit ironic, given that using water for generation may not be considered 'consumed' effectively.

The number seems considerably higher than one would expect: Arizona State University researchers, using 2024 Lawrence Berkeley National Laboratory data, put the indirect consumption of US data centers, including power generation, at about 650,000 acre-feet a year nationwide.

Although the data are old, the numbers suggest data centers may be consuming a fraction of what Ceres implies, according to its own report, unless consumption rose drastically to 16 times its reported figure.

Ceres, however, for now projects that water for power generation could account for 72 percent of total US data center water consumption by 2030, on a consumption basis, even as cooling inside the buildings gets more efficient.

At a time when the two largest Colorado River reservoirs have hit record lows, Ceres noted that local opposition grounded in water and grid concerns disrupted $130 billion of data center projects in the first quarter of 2026 alone.

Shama Perveen, the report's co-author and Ceres' director of water research, framed this as a complex situation and one that required a more inclusive approach, saying "no single stakeholder can address these challenges alone".

Whether it pans out at a time when AI companies are spending money hand over fist to get their hyperscale operations (and subsequent cooling and power requirements) out of the door as the search for AGI continues unabated remains to be seen, however.

• Intel report claims 60% of business leaders expect to operate robot fleets within the next five years
• Conversely, only 40% have a human-robot workforce strategy
• The Robotics Readiness Gap report assesses the state of robotic adoption around the world

Intel’s 'Robotics Readiness Gap' report has revealed that many large businesses are planning to implement a human-robot workforce within the next five years, with a majority expecting a doubling of operational output with a full-scale deployment.

The survey of 800 senior leaders in organizations with over $500 million annual turnover also found while adoption plans are accelerating, the rollout of a human-robot workforce has some key challenges, not least infrastructure, skills, strategy, and safety.

Despite the positive outlook for robotics, the scenario has a heavy dose of reality. Only 40% of businesses surveyed have a suitable strategy in place, despite a belief that robots could prove more cost-effective than humans within three years.

Human-robot workforce

Within this context, the aim of achieving a human-robot workforce seems more achievable, although some of the findings suggest an aspirational element to the drive towards robotics.

While 60% of leaders are expecting to run a fleet of robots by 2030, only 40% have a suitable human-robot workforce strategy in place. So while 74% believe workforce planning will be inseparable from robotics strategy within 5 years, there is a gap in preparation.

The report notes that “On average, leaders believe that deploying a robot will become more cost-effective than hiring a human in their sector within three years.”

But does this mean replacing jobs? Intel says not. “This isn’t about replacing people but creating new forms of human-machine collaboration, with robots working alongside employees as productive teammates.”

Doubling operational output

Intel’s report states that robotics in the workplace is approaching a “critical threshold,” with leaders predicting that “operational output would double if robotics were deployed at full scale across their organization.”

However, both notions rely heavily on five factors: strategy, skills, safety, shape (function), and scale (which includes physical AI infrastructure).

Intel’s report states, “few are prepared for what large-scale deployment actually requires. Closing this gap will mean redesigning operating models for a workforce that includes autonomous systems.” Even in the defense industry, this gap exists (“confidence outstrips preparedness by 48%”).

Placed in these terms, and with the context of the size of the organizations quizzed on their robotics strategies, that critical threshold seems a little further away than five years.

• AMD Ryzen AI Max+ PRO 495 mini PCs now reach 192 GB of memory
• GMKtec says its EVO X5 Pro can run 300B parameter models locally
• Framework Desktop combines 16 cores with 192GB unified memory while Acemagic fits the PRO 495 inside a 2L case

After AMD launched the Ryzen AI Max+ PRO 495 in May 2026, the processor is now appearing in a new wave of mini PCs offering up to 192GB of memory.

The chip follows 2025’s Ryzen AI MAX+ 395, which appeared in at least 37 mini PCs costing above $1,500.

Acemagic, GMKtec and Framework have now shown systems using the newer processor, with GMKtec claiming support for 300B models.

New mini PCs push memory far beyond ordinary systems

Framework is preparing a Desktop model featuring the PRO 495 alongside 192GB of LPDDR5X memory and 273GB/s bandwidth, combining 16 Zen 5 cores, 32 threads, 40 graphics compute units, and 131 TOPS for AI workloads.

The integrated Radeon 8065S graphics shares system memory, allowing applications to access resources without relying on separate graphics memory.

Framework says the machine will occupy 4.5 litres, while Linux can also come preloaded for users who choose that option.

GMKtec has unveiled its EVO X5 Pro at IFA 2026, using the same processor with up to 192GB memory.

The system uses LPDDR5X running at 8,533 MT/s, while memory bandwidth reaches approximately 273GB/s under the company’s stated specifications.

GMKtec says users can assign as much as 160GB of that memory to graphics processing when required.

According to GMKtec, the processor reaches 5.2GHz, while its Radeon graphics include 40 compute units and hardware ray tracing support.

The company says this configuration can run large language models containing as many as 300B parameters without requiring cloud services.

GMKtec also includes an XDNA 2 neural processor capable of delivering up to 55 TOPS for supported artificial intelligence workloads.

The EVO X5 Pro has four USB4 connections, external graphics support, 8K output, and networking options for linked systems.

A vapor chamber, three fans, and three operating modes are used to manage heat during sustained computing workloads.

Acemagic adds another compact option with 192GB

Acemagic’s F9A uses the PRO 495 with as much as 192GB of LPDDR5X memory inside a roughly 2-litre enclosure.

The computer measures 158.5 × 158.5 × 81.5 mm, making its physical size considerably smaller than Framework’s 4.5-litre machine.

Its processor reaches 5.1GHz, while the Radeon 8065S graphics unit operates at frequencies reaching 3.0GHz under suitable conditions.

The system provides two PCIe 4.0 NVMe slots, allowing storage configurations using RAID 0 or RAID 1 when supported.

Acemagic also includes OCuLink using PCIe 4.0 x4, enabling compatible external graphics cards and other high-speed peripherals.

Two USB4 Type-C ports provide transfer rates reaching 40Gbps, while networking includes Wi Fi 7 and dual 2.5GbE.

The company says the F9A can run DeepSeek V4 Flash locally, with that model containing 284B parameters.

As of the time of writing, none of the three brands has published final retail pricing for its respective configurations.

Given the hardware involved and the premium attached to the previous generation, prices above $2,000 would not be surprising.

• Microsoft faces claims that its data center noise harms nearby residents
• Standard dBA monitors may fail to capture some low-frequency noise
• Residents accuse xAI of causing noise, vibration, and air pollution

Microsoft and xAI are facing lawsuits over noise and other alleged harms linked to data center operations in Wisconsin and Mississippi.

At least six similar cases have been filed this year, as residents challenge the impacts from currently active facilities.

In addition to nuisance, the disputes also focus on negligence claims, with plaintiffs seeking damages or changes to data center operations.

Low-frequency noise creates a measurement problem

Microsoft faces a proposed class action concerning its Fairwater facility in Mount Pleasant, Wisconsin, which began operating in June 2026.

Residents filed the case in late July, alleging that persistent low-frequency noise affects their homes and property values, and arguing standard dBA equipment may fail to detect the sound because its frequency falls outside effective measurement ranges.

“dBA measures do not effectively detect lower frequency sounds like those emitted by Defendant’s Data Center operations,” the lawsuit says.

“‘Data center cooling — massive chillers, slow turning fans, compressor arrays — generates exactly the long wavelength noise’” that is not adequately measured using dBA.”

The plaintiffs argue Microsoft could have reduced the disturbance through acoustic barriers, quieter cooling equipment, soundproofing, and improved monitoring.

“A properly operated, maintained, and/or constructed Data Center will contain, capture, or otherwise prevent the emission of excessive noise from its generators and cooling systems,” the filing says.

That argument could create difficulties for facilities where regulatory measurements fail to capture sounds that residents can still hear.

The dispute also raises questions about how noise from data centers should be measured.

xAI faces claims involving turbines and pollution

In Mississippi, residents have sued xAI over alleged noise, vibration, and air pollution from dozens of gas turbines.

The turbines supply electricity for the company's Southaven data center, according to allegations contained in the lawsuit.

Attorneys representing residents accused xAI of failing to take meaningful action to address the alleged effects.

“xAI has made no meaningful effort to halt the harm,” attorneys at Weitz & Luxenberg said in a statement.

Similar lawsuits have also been filed in Vineland, New Jersey, Dowagiac, Michigan, North Tonawanda, New York, and Hood County, Texas.

The cases generally rely on nuisance or negligence claims concerning conditions plaintiffs believe operators could have prevented or controlled.

Several other data center disputes remain unresolved, with none of the noise cases reaching summary judgment or trial.

However, other environmental nuisance cases have already produced settlements involving substantial sums of money.

Amazon agreed to pay more than $20 million over nitrate contamination linked to one Oregon data center.

The company said officials settled early to avoid prolonged litigation and concentrate resources on community support instead.

Microsoft had not immediately responded to a request for comment concerning the Wisconsin lawsuit when the report was published.

Via SmartCitiesDrive

• Huawei MateBook Pro S weighs 798g at 11.9mm, with a 14.2-inch 3.1K 120Hz flexible OLED touchscreen and a 54Wh battery
• The laptop runs HarmonyOS 6.1 rather than Windows and uses a Kirin XE90 SoC, a HiSilicon ARM design on SMIC's 7nm node, capped at 20W
• It currently sells only in China, where it starts at 7,999 yuan, with no plans announced for wider availability

Huawei has launched the MateBook Pro S, a 14.2-inch laptop that weighs 798 grams and measures 11.9mm at its thickest point.

The device runs HarmonyOS 6.1 rather than Windows, uses Huawei's custom Kirin XE90 processor, and went on sale in China on 14 August, with prices starting at 7,999 yuan, roughly just under $1,200 when converted.

There seem to be no documented plans to offer the same laptop officially in the West, but third-party suppliers have already emerged offering to ship to countries worldwide, albeit at a significant premium.

Thin, but not the thinnest, with a CPU that is relatively untested

At 798 grams, the MateBook Pro S is noticeably lighter than most of its competition, but thinner alternatives from Japanese manufacturers exist, shaving off an additional 164 grams.

Although it isn't one of the lightest in the business, it is considerably slimmer than laptops that compete in the same performance segments. Huawei achieves this with a magnesium-lithium alloy bottom cover, magnesium-aluminum A and C covers, a three-section motherboard, a thin dual-fan cooling assembly with what Huawei calls Shark Fin blades, and a high-silicon-anode battery.

With a 3.1k flexible OLED touchscreen offering 1000 nits of full-screen brightness and a peak of 1600 nits, plus a 120Hz refresh rate, it has one of the best displays on paper.

What stands out even more than the capable display, however, is the processor: the Kirin XE90, as Huawei claims 23% higher single-core performance, 25 percent better energy efficiency, and 40% higher NPU throughput than the standard Kirin X90, with the cooling system supporting up to 20W of sustained output.

These numbers, however, are unchecked, and much of what one can find about the laptop's performance is inferred from comparisons using Huawei's own figures or paid advertisements claiming it delivers 'formidable power'.

More interestingly, it may use the same SMIC 7nm N+2 node TechInsights found when it physically analyzed the Kirin X90 in the MateBook Fold, rather than the 5nm process Chinese television claimed. The XE90 essentially is a refinement of the same design, and there is no indication it has moved nodes since the finding was reported.

This matters because it affects not only Huawei's laptop performance, but also efficiency in a class where performance per watt is a key determinant. The laptop is also only 798 grams in its lightest configuration; its lineup splits into Standard, Soft Light, Privacy, and Collector's Edition, and the weights differ considerably. The standard and privacy models weigh 798g, the Soft Light version 805g, and the Collector's Edition 825g, with a thickness of 12.2mm.

Huawei also claims 18 hours of video playback, a stable WiFi 7+ connection up to 1000 meters away in open, unobstructed areas, and a six-microphone array that can clearly capture audio from up to 10 meters away.

For those looking for a solution that isn't Windows- or Linux-based, the MateBook Pro S is an intriguing buy — if you're willing to pay the premium.

• REGENT Viceroy Seaglider has completed its first human-crewed flight above Narragansett Bay
• The electric craft flew 1,956 feet during its first crewed test
• Viceroy can shift between floating, hydrofoiling, and low-altitude flight modes

REGENT Craft has completed the first crewed flight of its Viceroy Seaglider, advancing development of an electric vessel designed for coastal transport.

The prototype carried two captains during a 30-second test above Narragansett Bay, reaching 33 ft while travelling 1,956 ft.

REGENT says the flight marks the largest human-carrying all-electric wing-in-ground-effect craft to complete such a test.

First crewed flight tests an unusual transport concept

Viceroy combines a boat-like hull, underwater foils, and wings, allowing it to shift between water travel and low-altitude flight.

At low speeds, the craft remains afloat before hydrofoils lift its hull and prepare it for flight above the surface.

During test flight, the vehicle remains close to the water, operating within one wingspan of the surface rather than climbing conventionally.

"Completing this flight test validates years of rigorous engineering and operational work and gives us the confidence to move into full-rate production,” said Mike Klinker, co-founder and CTO at REGENT.

“We showed today that flying a Seaglider is every bit as smooth and safe as we designed it to be.”

REGENT says the production model will carry 12 passengers and two crew, with a maximum takeoff weight of 15,000 lbs.

Its planned flight speed reaches 180 mph, while its stated range extends to 180 miles between coastal destinations.

The company says passengers will have large windows, substantial legroom, and lower cabin noise than conventional aircraft during travel.

The craft uses a three-layer digital control system intended to manage stability while allowing mariners to operate it like watercraft.

REGENT says certified mariners will captain the vessels, while United States Coast Guard rules will govern operations within American waters.

Commercial plans expand alongside defence work

These vessels are designed for coastal routes, where existing ports could support operations without conventional airport infrastructure.

About 40% of the global population lives in coastal communities, providing a large potential market for such services.

At the moment, the company has booked orders spanning six continents, while manufacturing capacity has been reserved across several future production years.

Earlier this year, REGENT announced an agreement involving The Twenty Five, which ordered as many as 60 Viceroy vessels.

The arrangement would make the operator REGENT's first United States delivery and launch partner for commercial Seaglider services.

Military versions for logistics, medical evacuation, surveillance missions, and other maritime applications requiring rapid movement are in development

The company holds a $15 million agreement with the United States Marine Corps Warfighting Laboratory for related development work.

REGENT completed construction of a 255,000-square-foot manufacturing facility in Rhode Island during June, with production activities beginning this year.

The facility is intended for structural assembly, wing and foil fitting, battery installation, systems work, and water-based testing.

“This is the moment we have been envisioning and tirelessly pursuing for the past five and a half years: REGENT has successfully proven safe, efficient ground-effect flight,” said Billy Thalheimer, co-founder and CEO at REGENT.

• GPUThor is the first Rowhammer attack to defeat ECC on Nvidia GPUs and reach a root shell on the host
• It affects four Ampere workstation cards with GDDR6 memory, the RTX A4000, A4500, A5000 and A6000
• Non-uniform hammering, enabled by defeating GPU memory coalescing and finding that Target Row Refresh (TRR) mitigations fire once every 72 refresh intervals, delivers effectively 6.6 times the hammering intensity of prior attacks

Four researchers at the University of Toronto have disclosed GPUThor, a Rowhammer technique that breaks the error correction Nvidia has spent the past year recommending as the defense against exactly this class of attack.

Chris S. Lin, Joyce Qu, Aditya Rajeev and Gururaj Saileshwar are expected to present the paper outlining their approach and subsequent findings at ACM CCS 2026.

The attacks target Ampere-generation workstation cards with GDDR6 memory, specifically the RTX A4000, A4500, A5000 and A6000, and it turns an unprivileged CUDA program into a root shell on the host.

An attack that matters much more than its predecessors

Rowhammer works by repeatedly activating a DRAM row until charge leaks from cells in the physically adjacent rows and flips their bits. The attacker never touches the victim's data directly, making it an excellent precursor to tampering, sandbox escape, and privilege escalation, among other things.

This approach is possible despite an in-chip countermeasure called TRR that accompanies ECC-enabled chips on these GPUs. TRR aims to prevent Rowhammer attacks by tracking how frequently specific memory rows are activated and automatically refreshing adjacent rows before a malicious bit flip can occur.

This approach works well on paper and, when Rowhammer attacks hammer uniformly, offers decent protection. The problem arises when approaches such as GPUThor use non-uniform hammering, which is much more likely to succeed. Non-uniform hammering is not exactly new, having already succeeded on the CPU side of the spectrum thanks to the well-documented Blacksmith attack vector.

GPUThor is the third attack from broadly the same group in eighteen months, following GPUHammer in 2025 and GPUBreach earlier this year, and it matters because the previous two stopped working the moment a user typed the command to enable ECC.

ECC was essentially Nvidia's go-to response to reported Rowhammer attacks, but it may no longer be a solution in its current state. By hammering non-uniformly, the researchers opened another attack vector for Nvidia's Ampere-based GPUs, which are overwhelmingly affected by the technique.

With ECC disabled, GPUThor produced 72,000 to 377,000 bit flips per gigabyte across four Ampere-based cards, with Nvidia's A5000 being reported as the most vulnerable. That approaches the roughly 550,000 flips per gigabyte that Blacksmith achieves on DDR4, which is cause for concern: GPU Rowhammer is now in the same league as CPU Rowhammer.

When enabled, the researchers said ECC protectors reduced, but did not completely mitigate, the issue: they delivered double-bit errors and 2 triple-bit errors that the technique 'fixed' by choosing the wrong value.

On an RTX A6000 with ECC enabled, the GPUThor technique forces one GPU reset every 2 hours, killing all running processes on the GPU; as a result, the GPU flags itself as RMA-read within a day.

The team disclosed the attack pattern to Nvidia on 29 April 2026 and subsequently to Google, Microsoft and AWS, and held the work until the 25th of August. A code release is scheduled for the 15th of November, even though there is currently no CVE information or patch being deployed to address the issue.

• Wordfence discloses two critical flaws in Elementor Pro and Super Forms
• Bugs allow unauthenticated arbitrary file uploads, enabling remote code execution; both patched recently
• Exploitation attempts already exceed 440,000

More than six million WordPress users are at risk of website takeover, researchers have claimed after discovering two major vulnerabilities being exploited in the wild.

Security researchers Wordfence disclosed finding two flaws, one in Elementor Pro, and one in Super Forms - two popular WordPress plugins.

Elementor Pro is a commercial plugin that allows users to build websites using drag-and-drop elements instead of code. With it, they can add advanced widgets, templates, different forms, popups, and more. It is quite a popular solution with more than six million websites actively using it.

Two bugs, hundreds of thousands of attacks

According to Wordfence, up until recently, it was vulnerable to an “unrestricted file type upload” bug in all versions up to, and including, 4.2.1. “This makes it possible for unauthenticated attackers to upload files that may be executable, which makes remote code execution possible,” the researchers explained. “This requires that the targeted site has published a page containing an Elementor Pro Form widget with at least one non-required File Upload field.”

The bug is tracked as CVE-2026-32475, carries a severity score of 9.8/10 (critical), and was patched in mid-August 2026. So far, Wordfence alone blocked more than 190,000 exploit attempts.

At roughly the same time, the researchers also reported finding a flaw in Super Forms, a form builder plugin that lets users create and manage forms using a drag-and-drop interface. This plugin, with some 13,000 active installations, contained a bug that allowed arbitrary file upload in all versions up to, and including, 6.3.313.

“This makes it possible for unauthenticated attackers to upload files that may be executable, which makes remote code execution possible,” the researchers explained.

This one is tracked as CVE-2026-14894, also carries a severity score of 9.8/10 (critical) and it, too, was patched a few weeks ago. For this flaw, Wordfence observed more than 250,000 exploitation attempts meaning that cumulatively, these two resulted in 440,000 attacks already.

Given the widespread adoption of these plugins, and the fact that the flaws are being actively leveraged, users are advised to apply the fixes without delay.

Via The Hacker News

Can AI-driven NAS tech be sentimental? UGreen CEO Samuel Zhang seems to think so.

Launching the company’s new smart-home technology suite, he started by recounting a personal story. It was a hot summer’s day, and his family gathered in the backyard for a barbecue. They cooked together, chatted, played, and just had a genuinely fantastic time. But weeks later, when Zhang went to check his photo gallery and relive that cherished day, he was left feeling disappointed. The family hadn't taken any photos.

This, he said, got him thinking about home cameras — devices that are typically used for security. But, Zhang said, “it [a home camera] should see more than just danger — it should also be able to witness the joys of everyday family life”. And this, he said, was one of the key points of inspiration for UGreen’s new suite of local smart-home AI technology.

I got the chance to hear about, and see, UGreen’s home hubs, cameras, digital photo frames and more at IFA 2026 in Berlin. Here’s what I learned — and how the technology could transform the smart home.

Keeping it local

At the heart of UGreen’s smart-home suite is the new HomeAgent. This is an ‘agentic AI local hub' that brings local storage, computing, and control all together in a single device. Through this hub, users can connect up a range of UGreen AIoT devices, such as cameras, smart speakers, and digital photo frames, plus some select third-party products.

A crucial part of the HomeAgent’s appeal is its reliance on local storage and computing. This means that any data that runs through your hub — be that photos, videos, voice inputs, or other files — is processed locally, rather than through a corporate cloud server.

At a time when trust in large corporations is strikingly low, and major data breaches seem all too common, this could be a massive selling point for security-savvy customers. Keeping data local also removes the need for costly Cloud subscriptions, making the initial investment feel that bit more worthwhile.

There are three HomeAgent models scheduled for release. The first is the UGreen HomeAgent HA100, which is the most basic option. It provides all of the core functions you’d need from the hub, but offers slightly less processing power, memory, and storage, among other things, than its siblings. The HA100 Pro is the step-up model, which offers heightened performance, with 205 TOPS (as opposed to the standard HA100’s 26 TOPS), a 10GbE LAN port (compared to the HA100’s 2.5GbE), and a more advanced CPU.

The final boss, if you will, of the HomeAgent trio is the MA100. This is powered by the NVIDIA Jetson Thor T5000, and offers far superior AI performance to the other two models, and considerably more memory and storage.

However, this is certainly reflected in the price, with the MA100 commanding a list price of $19,999, compared to the HA100 Pro, which costs $5,999 and the HA100 at $1,799. Early bird offers are available, which make the HomeAgents about half-price across the board. Shipping is set to commence in early 2027.

AIoT in action

So what does UGreen’s smart home AIoT tech look like in person? Well, pretty impressive.

I was able to check out a demo, and take a closer look at all of the HomeAgent products, and other devices you can connect up to it. UGreen had a HomeAgent connected to a smart speaker, and a camera, among other things.

I tried walking towards a door, and a UGreen smart security camera started blaring after I walked past a certain point, telling me that I was on private property. This data is processed locally through the HomeAgent, so if you ask your connected smart speaker about any unusual activity, it can tell you if anyone approached your door, whether that was a delivery worker, or something more nefarious.

The smart speaker UGreen showed off has a new AI assistant, named Uliya. You can use the smart speaker to do just about anything around your home. You can ask Uliya to remind you of where you were at a specific time, to close the curtains, to play music, or even to identify the location of your pet and cast a live video feed. The functionality you get will vary depending on how invested you are in UGreen’s AIoT ecosystem, but the capabilities are pretty broad.

You can also ask Uliya to give you a summary of captured events throughout the day, display memories on a UGreen smart photo frame in real time, or to show you highlights from a special day — circling back to Zhang’s desire to revive memories that had been lost in time. One function I really liked was the ability to create an Away mode, so that when you step out of the house, lights will turn off, curtains will close, and a robovac will get to work.

UGreen’s new HomeAgents appear to be incredibly capable, feature-rich, and innovative products, and their local handling of data could be game changing for some households. At the product launch, Zhang said “In the future, we hope every family can have a HomeAgent”; and it'll be interesting to see how families adopt such tech moving forward.

• G7 urges governments and organizations to begin transitioning to quantum‑resistant encryption
• Warning highlights risk from future quantum computers able to break current standards like AES
• Guidance: inventory cryptographic assets, prioritize critical systems, and adopt phased PQC strategies

The G7, a collection of some of the world’s most powerful economies, is urging organizations to adopt quantum-resistant encryption as soon as possible and minimize the risk of losing sensitive data to technologically advanced threat actors.

Virtually every industry in the world today relies on encryption (the process of converting readable data into scrambled, unreadable data so that only someone with the correct key can decode and access it) to protect sensitive information. Banks, telecommunications providers, defense organizations, and practically everyone else are using some form of encryption - most probably AES (Advanced Encryption Standard).

Decrypting the protected data without the encryption key is considered almost impossible, since the computational effort necessary to pull it off would simply be too large. However, since quantum computers work on a fundamentally different principle, it is believed that once they are mature enough, they will be able to “crack” today’s encryption standards.

G7's advice

Now, the G7 has urged government and organizations to remain ahead of the curve by deploying “quantum-resistant encryption”:

“To protect themselves from the threat brought by CRQCs (cryptographically relevant quantum computers), organizations should begin planning their PQC transition now and should aim to complete their transitions within any the timelines set out by their national cybersecurity authorities,” the warning reads.

“In addition to quantum-related risks, organizations that delay their PQC transition may lose competitive advantage or may be excluded from contracting opportunities, including public procurement.”

The G7 advises governments and organizations to take a number of steps, including identifying critical systems and prioritizing them. They should adopt a phased and risk-based strategy, start their transition early, and inventory their cryptographic assets. They should also map their dependencies and develop a transition plan.

“To limit transition-related costs, they should opt to purchase products that integrate PQC and replace their systems with quantum-safe ones as part of their standard renewal schedule. As such, starting the transition early could result in lower migration costs overall. Planning and conducting their PQC transition properly would enable organizations to prevent insecure implementations and avoid increased exposure to conventional cyber threats.”

• US military branches disable ad tracking on government‑issued devices after warning
• Location data from personal devices still poses risks; past data remains for sale
• Senators urge DoD inspector general to review policies and stop servicemember data exposure

More than 20 years ago, Jason Bourne was smashing mobile phones to prevent being tracked by his adversaries - and the US Department of Defense (DoD) has realized it should probably do something similar.

In May 2026, Senator Ron Wyden mailed the DoD (among others), informing the agency that foreign powers were using commercial location data to target US troops - and now, Wyden has shared the responses he received from the US Army, Air Force, Navy, Marine Corps, and Special Operations Command, all of which have disabled advertising tracking across their government-issued devices, including Apple and Android smartphones, and Windows computers.

Some branches did it earlier this year, while the Air Force allegedly concluded the efforts in July 2026 - and while Wyden praised the move, he also suggested it is not enough, and that further steps must be taken in order to protect US troops stationed abroad.

Evaluating the leaked files

“We commend these service branches for implementing this cybersecurity defensive best practice on government devices,” Wyden and Senator Pat Harrigan wrote. “However, recent reports regarding the continued availability of commercial location data originating from DoD facilities raise troubling questions.”

Apparently, despite disabling these features and placing the troops out of reach of data brokers and similar commercial entities, the data that was gathered earlier remains available for sale, which still presents a risk.

“The members suggested data could continue to be available for several reasons, including that commercial data available originates from personal devices, not government phones, that are carried by servicemembers and government contractors,” a report on Wyden’s website reads.

“Wyden and Harrigan called on the DOD inspector general to examine commercial location data that DOD and other government agencies have already purchased to determine why existing policies have not prevented servicemembers location data from being sold online, and to recommend policy changes to better protect US personnel.”

Via TechCrunch

• OpenAI hid an incident where a model hijacked a wiki page to use as an AI agent communication board
• The incident was hidden while the company dealt with the fallout of the Hugging Face attack
• The company is now working on a framework for disclosing incidents of 'misalignment'

OpenAI recently disclosed the details of how one of its AI models escaped a sandboxed environment and attacked Hugging Face during an evaluation - and as part of the incident, the models created a messaging board to communicate with each other and influence each other’s reasoning.

OpenAI has now disclosed that shortly after this incident, agents undergoing testing again escaped their ‘secured’ environment and hijacked an obscure German wiki to use as a messaging board. Per Reuters, OpenAI leadership kept the incident hidden while they dealt with the fallout from the Hugging Face incident.

Now that OpenAI has acknowledged its role in the incident, the company has said it is “past time” to put together an incident disclosure pipeline when its models escape testing and slip into third-party networks.

Who is at fault when models do what they’re designed to do?

Before the two incidents, OpenAI said it, “treated misalignment largely as a research question, which gets communicated in research publications”. But now that models are behaving in previously unknown ways and having real-world impacts, the company said it would change its approach “to expand for this new phase of model capabilities”.

The company labelled the most recently disclosed incident as “an instance of misalignment similar” to the Hugging Face breach.

I myself am guilty of reporting on AI breaking out of containment as going ‘rogue’, but these models are doing exactly what they are designed to do. OpenAI’s detailed disclosure of the Hugging Face incident showed that the models were pushed to try and solve a benchmark test by cheating, which is exactly what caused the cyberattack to happen.

OpenAI said that both itself and “the larger AI community do not yet have a clear standard for how to report misalignment that shows up during training, evaluation, and deployment, including examples that don’t look like traditional security incidents but could provide insight into AI behavior and future risks”.

The company added that it is “working on a framework and will share it in upcoming weeks, and in parallel we’re working with dozens of government regulatory agencies worldwide on these issues”.

“When you combine this 'breakout' with the Hugging face breakout, it's starting to display a pattern,” said Ashley Knowles, Lead Cybersecurity Consultant at Black Hills Information Security. “I struggle here with not getting too doomsday-ish but realistically, this is showing a pattern of concerning behavior.”

“I'm wondering if this race to become 'first' is undercutting security measures that need to be taken to properly secure and guard AI agents as they're in development. My concern grows when you consider that OpenAI is also resisting further investigation. Adding onto that, the release and promise that Astra can evade human monitoring. The pot is brewing…”

Ask a CISO why they bought their newest cybersecurity tool, and they’ll have a clear answer lined up.

It stops a specific technique, closes a particular gap, or satisfies a compliance requirement.

However, it's often less clear how the tool fits in with the rest of the stack.

Does it make the overall system stronger, or simply add another layer of complexity to manage?

Research indicates most security professionals already believe they’re juggling too many tools, with over half saying they don’t properly integrate together.

This is a pattern I call ‘additive by default,’ and it results in stacks that grow without a plan, becoming broader but not necessarily deeper or able to match today’s threats.

Start with the outcome, not the technology

I find this additive approach is often due to focusing on specific emerging threats or identified weak points, so decisions are made with a tactical eye rather than a broader strategic view.

Part of the problem is that most organizations have never precisely defined the security outcome they are trying to achieve. Without a shared, specific language for the problem, every new purchase becomes additive, because there is nothing solid to measure it against.

Take asset labelling, which is a core capability most organizations know they need, so they invest in a tool, populate a configuration management database (CMDB), and assign criticality scores. Job done, right? But labelling an asset only answers the question of what it is, and says nothing about how that asset connects to everything around it, or what happens when a policy needs to be enforced against it.

Labelling, visibility and enforcement are three distinct jobs, not one. Solve the first and the second and third remain wide open, so another tool gets bought to cover visibility, then another for enforcement. Each purchase solves its own narrow question perfectly well.

However, none of them were ever asked to work as a single, continuous capability, because nobody defined that as the actual requirement in the first place.

What if asset management labels fed into visibility views and the same reflected how policy is drafted and then enforced? Now you have a strategic problem solved with interoperable capabilities.

How much of your stack is really putting in the work?

Gartner’s most recent Leadership Perspective Survey saw CISOs noting this as a common issue, with just 20-30% of tool capability actually being used in some cases. The instinctive response to this is to cut the stack down, however, that instinct solves the wrong problem.

A good exercise for working out the value of the stack is to evaluate every tool, new or already deployed, against three plain questions.

Does it offer continuous validation against a given threat - and what, specifically, does it validate? Is it still operationally relevant? And is it effective, right now, in the environment you have today? A tool can pass one or two of these and still be failing you.

Virtual Local Area Networks (VLANs) are a good example of this. Twenty or thirty years ago, when networks were typically static and everything likely sat inside a single data center, VLAN-based segmentation was genuinely effective. It matched the environment it was built for.

That environment has since changed almost beyond recognition. Workloads move between on-premises systems, cloud and containers, and nothing stays fixed for long. VLANs are still deployed across many stacks today, still technically doing the segmentation job they were built for. Yet their effectiveness has dropped sharply because they offer none of the continuous validation a hybrid, constantly shifting estate actually requires.

Effectiveness has an expiry date that has nothing to do with whether the tool still runs. Success should be measured by whether the system as a whole still holds up, not by how many tools remain switched on.

Consolidated security, not consolidated tooling

Gartner identified that most organizations are pursuing a vendor consolidation strategy. While this approach certainly reduces unnecessary spending and keeps budgets under control, it’s not necessarily solving the biggest problem.

Consolidated tooling and consolidated security are not the same thing, and it’s an assumption that leads to disappointment. Reducing the number of tools alone achieves little if the underlying processes remain fragmented or teams continue to operate against different objectives.

No data center I have walked through was built entirely by one manufacturer. Racks, switches, storage and cabling come from a dozen suppliers, yet they operate as one coherent system because they were designed to fit together. Security should work the same way.

The goal is not necessarily fewer vendors, but every control, whoever built it, feeding into the same continuous picture of identification, visibility and enforcement.

What to ask instead

Before completing the next security purchase, consider how well a new tool fits into the stack you already have, not just what it claims to do on its own.

A tool that deploys cleanly, validates continuously rather than only at go-live, and feeds its findings back into the tools already in place is doing real work. One that arrives as a fresh, isolated source of alerts is just adding to the noise, however good its individual detection rate looks in a demo. That means comparing what the tool was bought to solve against what it is actually delivering today, checking it against newer capabilities, and being willing to redeploy or renegotiate rather than automatically renew or even retire.

Ultimately, the strongest security programs are not defined by the number of tools they deploy, nor by the number they eliminate. They're defined by how effectively every control works together when it matters most.

We've rounded up the best endpoint protection software suites.

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

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

Internet security teams are not short of information. Threat intelligence feeds run around the clock, vulnerability scanners flag thousands of issues a month, and high-profile CVEs dominate the news cycle before most teams have finished their morning coffee.

What is far harder to come by is proof. Raw data is one thing, but what about evidence that the controls sitting in your environment actually detect and respond to the way real attackers behave, in your specific network, today?

That gap is what purple teaming exists to close, and it is where the surprises tend to show up. Running these exercises with organizations that have invested heavily in their security stack, I have repeatedly seen gaps nobody expected, missing telemetry, misfiring detections, and attack paths nobody was watching.

The problem was never simply being under attack. It is being under-validated, and mistaking spend for assurance.

Why so many programs that look mature on paper still fail

Here's a pattern I keep running into with otherwise well-resourced organizations: every control on their books checks out, reports are signed off, tooling is deployed, leadership reassured. None of that holds up once a real attack runs through the environment. The cracks show almost immediately.

Ask where the actual gaps sit, and a familiar list comes back. Some systems feed logs in real time, others barely at all, leaving the SOC with a patchier picture than the dashboard implies. Detection rules are tuned to a generic attacker, not the one likely to show up here. And when something does trigger, it can sit unactioned for hours, since no one owns it.

None of this is exotic, just the ordinary, unglamorous consequence of tools and processes configured once and assumed still to work. With luck, those gaps surface in a simulated exercise before a real attacker finds them.

Part of the problem is what gets measured in the first place. Running standard techniques against a host already flagged as compromised only confirms which rules switch on. That's coverage, nothing more. It won't tell you whether the excess permissions, trust relationships, or misconfigurations in your real environment are exploitable, since that activity looks like normal use and was never built to trigger an alert.

Real validation looks different. It starts with the organization's own risk profile, not a generic library of techniques, and treats configuration and permissions as the real cause, not something a new detection rule can patch over. Only testing your own environment tells you which gap you're actually facing.

What's worth measuring, and the case for continuous validation

Treating validation as an ongoing discipline rather than a one-off exercise also changes the output you’re looking for.

You don’t want another report that sits in a folder until next year's audit; you want to create a living, prioritized backlog. This is a running list of gaps that have been proven to matter, ranked against the paths an attacker could realistically use, not against theoretical severity scores.

It’s an important distinction because a CVE rated critical in isolation may be unreachable in your environment, while a modest misconfiguration sitting on a well-trodden attack path can be far more urgent.

The most valuable information a security team can generate is not a longer list of vulnerabilities, but evidence of which exposures are actually exploitable, mapped against real behavior, and prioritized against genuine business impact.

This is why it is critical to avoid relying purely on agent-based validation. Because an agent is already embedded within the network or on an endpoint, it inherently shortcuts a lot of attack paths and bypasses defensive controls. It assumes the attacker has already achieved a foothold at that specific location, giving you an artificial view of your actual perimeter and lateral resilience.

Alongside the process itself, cadence is another important factor here. An annual or ad hoc test captures a single snapshot, and environments do not stand still between engagements: new services get deployed, configurations drift, staff change. Shifting from ad hoc testing towards an operationalized, continuous cycle of testing, validating, remediating, and retesting is what keeps that backlog tied to the environment as it is right now, not as it was assessed to be six months ago.

The ability to close a validated gap before it can be exploited as an attack path enables the organization to shift from reactive footing to a more proactive stance.

Breaking down the walls between red, blue, and the business

None of this works if validation happens in a silo. The most consistent driver of a strong outcome I have seen is not a tool; it is collaboration between people who are technically on the same side but often operate as if they are not.

Red and blue teams that only meet at the end of an engagement, via a report, tend to treat findings as a scorecard, something to defend or dispute rather than act on. An open-book approach changes that dynamic entirely. When offensive and defensive teams compare notes on what they are seeing, what they expected to see, and why, as the exercise unfolds, gaps get identified and understood together.

If a defensive team cannot follow a technique or lacks the telemetry to see it, that is not a failing to hide; it is the exact information the exercise exists to surface.

Clear ownership matters just as much as clear findings. An alert with no defined owner is functionally the same as no alert at all, so part of building this collaboration is agreeing, in advance, on who acts on what.

That communication cannot stop at the SOC door, either. Validation findings need to travel outward too, translated into terms a business audience can act on: which risks are real, which are theoretical, and where investment will actually reduce exposure. Security teams that build this bridge consistently get faster remediation and fewer arguments about priority, because everyone is working from the same evidence rather than competing assumptions.

Moving towards more coordinated action is another step towards preempting risk, delivering smoother and more effective operations than an isolated response where fixes happen team by team.

Resilience is proven, not assumed

None of this is an argument against investment in tools or people. Rather, it’s an argument for asking a different question that identifies risk. It's time to stop asking, "Are we safe?", because no environment stays safe indefinitely. The question that matters is whether you're resilient enough to validate an exposure and respond before it turns into a breach.

Confidence built on assumptions collapses the moment a real adversary tests it. Confidence built on shared, ongoing validation does not, because it has already been tested, refined, and proven against the way attackers actually behave in your environment.

We've listed the best Linux distros for forensic and penetration testing.

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

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

• Rhysida ransomware group breached Berlin’s state government, stealing 1.44 million files weighing in at around 5.8TB
• Attackers demanded 30 BTC (~$2.3M); Berlin refused, leading to full leak online
• Leaked data reportedly includes water supply info, staff records, and emergency plans

A cybercriminal group known as Rhysida allegedly broke into the network of Berlin's state government and exfiltrated 1.44 million files. They then tried to extort the government entity for money and when that failed, they leaked it all into the dark web.

According to multiple sources, the group first claimed responsibility for the attack on an underground forum, where it shared a small sample of the stolen files and demanded Berlin pay 30 bitcoin (around $2.3 million) in exchange for deleting the files.

Berlin recognized the attack in a press release, and said it would not be negotiating with the attackers, and instead launched a full-scale investigation into the incident, which it described as an “extremely serious crime and an attack on the state of Berlin.”

Evaluating the leaked files

A few days later, Rhysida decided to leak it all online - and Berlin is now reviewing the leaked files and is assessing the damage:

“After the publication of the stolen data from the Berlin administration, these files are being evaluated at full speed,” a press release, published on the Berlin.de website (machine-translated), reads.

“An additional steering unit has been set up in the Senate Chancellery under the leadership of the Chief Digital Officer (CDO), Florian Hauer, which coordinates the review, examination and evaluation of the leaked data and supports the two Senate administrations concerned in informing and advising the affected citizens and companies.”

According to German public broadcaster Tagesschau, the archive counts 1.44 million files and totals 5.8 terabytes. At the same time, the Chaos Computer Club (Germany's largest and best-known hacker organization) claims the archive contains sensitive data on the city’s water supply, personal data of admin staff, various employment references, and emergency plans.

Via Reuters

• Microsoft reports phishing campaign using ASCII smuggling to bypass spam filters
• Attackers insert invisible characters into keywords, tricking filters and AI agents
• Defenders should normalize Unicode tags and flag unexpected invisible code points as suspicious

Cybercriminals are using the “ASCII smuggling” technique to make sure phishing emails pass security filters and land in people’s inboxes, experts have warned.

ASCII is a character encoding standard that turns characters and words humans can read into numeric values that computers can understand. It can also be used to create characters that aren’t even displayed on the screen (essentially “invisible” ones) but can still be read by the machine.

In a new report, security researchers from Microsoft found crooks are abusing this fact to distribute phishing emails. Most email providers offer solutions that filter out spam emails. These filters look for certain keywords and phrases, such as “funding”, “credit”, “loan”, and similar, and automatically send such emails to the spam folder.

Ongoing campaign

By adding a set of invisible characters in the middle of these keywords, the attackers can break them apart and thus “confuse” the filters.

While the human sees the word “funding” in their email, the security solution is seeing something like “fun[a long string of characters]ding”. This technique has been adopted from prompt injection attacks, where crooks would use ASCII smuggling to deliver malicious and invisible prompts in the emails.

Therefore, when a victim asks their AI agent to summarize the email, it ends up working on a prompt that could be anything from extracting sensitive data, to deploying malware.

Microsoft is saying the campaign has been ongoing for months, and while it peaked in February 2026 with more than 2.3 million emails every day, and has been in decline ever since, it remains active to this day. In early February, Microsoft observed a cluster of almost 150 sender domains, all themed around finance. These domains accounted for almost all (96%) of all the spam emails Defender for Office 365 flagged under ASCII smuggling.

Defending sounds relatively simple, though. IT teams should normalize Unicode tag characters and other invisible code points before applying any keyword detection. They should also consider all unexpected tag-block characters as suspicious.

Via BleepingComputer

You probably have tens, if not hundreds of passwords floating around across all the online accounts you use every week. In fact, the average person has over 150 passwords. It's not easy to remember them all, making it more attractive to keep them simple and similar - but that can lead to a serious chain of account theft.

That's why password managers like Keeper exist. They generate strong passwords and store them passwords in a secured vault to keep them safe from hackers and prying eyes. They even autofill your credentials to make logging in to your account faster and easier.

Keeper has cut prices on its personal, family, and business starter plans by up to 50%. Keeper regularly scores highly in our reviews, and includes great features across Personal, Family, and Business plans.

Why we recommend Keeper

Our Keeper review highlights the platform's zero knowledge architecture and device level encryption that helps keep your password vault secured against unwanted intrusions.

Where Keeper really shone was in our usability and convenience testing. We loved how Keeper uses biometric security to access your vault, rather than requiring a master password with each log in. By using a facial scan or fingerprint to verify that its actually you accessing your vault, it adds a fast but secure method of verification to ensure its you, and not a nefarious actor.

For households, the Family Plan includes five private vaults, allowing you to quickly share Wi-Fi or streaming passwords using Keeper's shared vaults.

The Business Starter plan covers 5 users, covering an unlimited devices, shared team folders, user activity reporting for IT admins, and a free Family Plan for every team member.

These discounts apply to the first year only, so the value is highest for new customers or anyone switching from a monthly plan. After that, pricing goes back to standard rates.

• Chinese companies have reportedly hidden their ownership to operate in Taiwan
• They're also poaching Taiwanese workers, paying them up to 10x more
• China is fighting back against US restrictions, but Taiwan wants to protect its own industry

Taiwan has reportedly increased the pressure on Chinese technology companies that are hiding their Chinese ownership as it tried to protect its own semiconductor business.

The country's Ministry of Justice Investigation Bureau (MJIB) has reportedly investigated 166 cases involving suspected illegal Chinese ownership since 2020, resulting in 36 convictions – 33 of which involved in semiconductor research, development or design.

While Taiwanese law doesn't prohibit Chinese companies from operating in Taiwan, they must seek government approval under the Cross-Strait Act, but by obscuring their Chinese ownership they've been looking to circumvent those restrictions.

Chinese companies targeted for operating illegally in Taiwan

Rest of World reports many of the Chinese firms in question established shell companies in third countries or used non-Chinese nationals to register their businesses, making them appear either Taiwanese or generally non-Chinese.

The news comes against a backdrop of heightened geopolitics – China has been investing heavily in growing its semiconductor self-sufficiency amid ongoing US restrictions, and have reportedly been poaching Taiwanese engineers with salaries up to 10x higher than average.

Taiwan itself is said to account for more than 60% of global semiconductor production, producing around nine in 10 of the world's most advanced processors, so alleged Chinese poaching is especially noteworthy. "Taiwan’s high-tech talent has become a key target for recruitment by Chinese enterprises," the MJIB wrote.

In a separate case, the MJIB has also investigated 67 cases involving Chinese companies suspected of stealing Taiwanese trade secrets.

Interestingly, the cases don't just relate to smaller start-ups – Xiaomi, OnePlus, Oppo and Semiconductor Manufacturing International Corporation (SMIC) were all named in the report.

Fines reported in the cases ranged from around $3,800 to $800,000, with penalties of up to three years' imprisonment also cited.

In its latest financial stability report, the Bank of England warned that an AI crash could plunge the UK into recession, claiming that a “price correction in AI stocks, driven by a change in productivity and profitability among tech-led companies, could cause a 2.2 percent fall in U.K. GDP.”

This should serve as a warning to businesses everywhere.

The risks surrounding AI extend far beyond Silicon Valley.

As leaders prepare for what comes next, however, many are still asking the wrong question.

“Is AI real, or is it hype?”

The last 30 years of technology should have taught us that the answer can be both.

The last tech crisis of the 20th century

Y2K was a real technical risk that came to look, in hindsight, like mass overreaction.

The UK spent billions preparing for the millennium bug, with organizations testing and patching systems that underpinned everything from banking and benefits payments to air travel and the National Grid.

The government established Action 2000 to help businesses prepare, while banks and major infrastructure providers ran extensive contingency plans ahead of the deadline.

Then midnight passed, the planes stayed in the sky, cash machines kept dispensing money and the lights stayed on. The whole affair began to look almost laughable.

Yet the apparent anticlimax obscured an important point. Much of the disruption people feared was likely avoided because organizations took the risk seriously and prepared for it. Cybersecurity has the same problem.

A threat that is successfully mitigated can look, in hindsight, remarkably like a threat that was exaggerated.

Not to burst your bubble

The dotcom bubble taught a different lesson. The internet was not fake, but the valuations often were. Boo.com, for example, raised vast sums to reinvent fashion retail online, only to collapse in 2000 after burning through its funding. Companies with little profit and, in some cases, barely coherent business models were treated as if they had risen above the laws of economics.

Then the bubble burst, fortunes disappeared and many of the supposed pioneers vanished, but the ideas they were betting on did not. In fact, I bet you’re reading this on the internet right now.

Style over substance

Crypto added a third lesson. Revolutionary language can hide weak use cases. There were serious technical ideas beneath it, but the public boom became dominated by speculation, celebrity endorsements and the strange habit of starting with an asset and then searching for a purpose.

At the height of the NFT frenzy, an NFT of Twitter founder Jack Dorsey’s first tweet sold for $2.9 million. A year later, its owner attempted to resell it for nearly $48 million, only to see initial bids fall spectacularly short. The technology had not disappeared. What had disappeared was the market’s willingness to treat digital scarcity as lasting value.

AI’s rise carries elements of all three stories.

A familiar visage

Like Y2K, AI presents risks that may sound exaggerated until they are not. As any cybersecurity expert will tell you, if the worst harms are prevented, the warnings may later look hysterical. This will not prove they were imaginary.

For AI, this means treating cybersecurity and governance as operational disciplines, not innovation-team side projects. Organizations need to understand what data is entering AI systems, what those systems can access, what authority they have, how they could be manipulated and how the business would respond if they failed.

With the dotcom boom, AI looks like a real general-purpose technology wrapped in an overheated investment story. It can already write code, summarize documents, and automate parts of professional work. But that does not mean every AI company is valuable, every AI product is useful, or every data center will earn its keep.

For businesses learning from this, they must invest in capabilities, data foundations, people and workflows, rather than betting everything on whichever vendor, model, or product is making the most noise this quarter.

And like crypto, AI has acquired a mythology. “AI-powered” is becoming what “blockchain-enabled” briefly was, a phrase that can mean something, nothing, or merely “please value us higher.”

For the adoption of AI, business leaders must now ask boring, but necessary, questions. What problem does this solve? Who uses it? What data does it see? What systems can it influence? What authority have we given it? What happens if it is wrong, manipulated or unavailable? And who is accountable when that happens?

This combination is what makes the current moment so hard to read. The lazy argument says AI is either a revolution or a bubble. History suggests a more uncomfortable possibility; it could be both.

“The Spirits of all three shall strive within me. I will not shut out the lessons that they teach.”

Too often, business leaders are trying to win an argument about whether AI is overhyped when they should be building a strategy that survives either answer. The sensible response to AI is neither a moonshot nor a moratorium. It is disciplined experimentation.

This means treating AI less like a campaign slogan and more like a portfolio of bets. Some of those bets should be defensive, focused on understanding where AI is already being used, what data and systems it can access, and whether the organization can respond when things go wrong.

Others should be exploratory, testing where AI can create genuine value across areas like coding, customer support and knowledge management. And some should be deliberately skeptical, particularly when tools sound impressive but cannot demonstrate savings, better outcomes, or clear ownership.

If the aim is simply to have an AI strategy that reads well in a board pack, then the lessons of the past 30 years have been missed. Y2K, dotcom and crypto each showed us that hype and substance are not opposites. Real risks can be exaggerated. Transformative technologies can attract irrational investment. Powerful ideas can coexist with weak use cases.

AI may prove to be all three at once. The businesses that navigate it best will not be those that predicted the future perfectly. They will be those that were prepared to adapt when reality arrived.

Best cloud storage: tested, reviewed and rated by experts.

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

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

• Around three in four UK medical practices use Docsign to share important letters
• Docmail has been down for five days due to a server issue
• Services can be resumed once the investigation completes

Docmail, a cloud-based mail service used by high-profile businesses and public-sector organizations, including some NHS trusts, continues to suffer a major outage after first being hit on September 2 2026.

The platform works by customers uploading their communications digitally, with Docmail handling printing and postage to recipients.

Parent company CFH says more than 75% of UK medical practices use its services for bulk and individual patient mail, spanning GPs, dentists, pharmacies, opticians and other private-sector firms.

Docmail is still down

Now on its fifth day of interruptions, it means healthcare appointment letters, council correspondence and other important mail could be missed or delayed.

Docmail said customers were experiencing "slower than usual processing times" on September 2 before taking the service offline for "essential maintenance."

On September 4, the company told The Register: "We have been made aware of an issue impacting an isolated server, used by only one client." In other words, CFH purposely shut down the wider platform to investigate this incident, however the finer details of what went wrong are under wraps.

"This process does take time, and we are acting on the advice of external consultants to ensure we are taking the right precautions," a company spokesperson added.

According to CFH, Docmail itself is ready to be restored and made available, but external reviewers are still conducting the investigation.

The company's status page has not been updated since September 4 at the time of writing – at that time, it labelled the outage as an "unresolved incident."

"Work remains ongoing as a priority, and we will provide a further update as soon as we have more information to share," the company added.

• Nvidia pays $11.9 billion for Hugging Face, and a further $1 billion in equity-based retention
• The company promises to keep Hugging Face "an open platform for the entire AI ecosystem"
• Nvidia CEO Jensen Huang has publicly supported open weight models

Nvidia has confirmed plans to acquire Hugging Face for $12.9 billion subject to regulatory approval, marking a major expansion for the chipmaker as it snaps up one of the most important AI model distribution platforms.

In a post, the company asserted it plans to keep Hugging Face as it is: "Hugging Face will remain an open platform for the entire AI ecosystem," CEO Jensen Huang wrote.

Huang explained Nvidia's "infrastructure, engineering and global reach" will help improve Hugging Face's "reliability, safety, model evaluation, inference and deployment capabilities."

Nvidia confirms Hugging Face acquisition

The platform currently counts more than 18 million developers, researchers and creators as users, with more than three million models shared on it.

"Hugging Face will continue to support open source and open weight models from across the ecosystem, from every model builder," Huang stressed again, noting his public backing for open weight models. Ordinarily in an acquisition, questions would be raised over whether, in this instance, Nvidia's own GPUs and software could receive preferential treatment, but the company is doing everything it can to assure us that won't be the case. A pending regulatory investigation will also ascertain this.

Nvidia also claims to be the biggest contributor to Hugging Face, having shared more than 500+ models and around half as many open datasets.

The news comes less than two weeks after reports started to circulate about a potential acquisition, and if it closes, it'll be Nvidia's biggest acquisition to date. The $12.9 billion payout comprises a Hugging Face valuation of $11.9 billion (way above its earlier $4.5 billion valuation in 2023) and an equity-based retention program of around $1 billion.

As for upcoming regulatory interjection, Nvidia warned in a SEC filing that "Government restrictions may negatively impact our business and the Hugging Face platform."

• 2027 UBS recruits told they must have AI proficiency and a willingness to learn
• The bank's AI Fluency Pathway will continue to support junior workers' development
• While thousands of jobs could be at risk, we're starting to see shifts rather than outright displacement

Swiss investment giant UBS is now requiring all junior bankers to demonstrate AI proficiency as the skill moves from being a nice-to-have to an absolute requirement within recruiting.

The change currently applies to graduates and interns applying to the company's 2027 intake, per the Financial Times, and it's unclear whether UBS will broaden the requirement to all workers in the future.

As part of the new requirement, recruits will need to be able to demonstrate that they can use and experiment with AI responsibly to improve business outcomes – not just that they can use popular AI chatbots like ChatGPT.

UBS makes AI skills a must-have

AI-related questions will now become part of the bank's recruitment interviews on top of both the existing types of questions as well as the usual requirements, like a 2:1 degree.

While the news puts additional strain on graduates who now need to invest in their own AI skills, it's an example of how artificial intelligence isn't replacing entry-level workers, with the bank seeing it more as a productivity booster for human staff.

UBS' training program will also include an 'AI Fluency Pathway' to cover real-world banking AI use cases and responsible AI use, implying that the bank is more focused on prospective workers being able to prove a certain level of proficiency and willingness to learn – not full proficiency from the get-go.

AI's longer-term effects on banking employment are more unpredictable, though, with an earlier Morgan Stanley report warning that 200,000 banking jobs could be lost in Europe over the next five years. That was in early 2026.

While the outlook now seems more positive that junior workers may not be at a loss, it's clear that roles are evolving and entry-level workers could see their responsibilities shift toward AI management.

For the last two years, the advice to brands has been more or less standardized: get cited and recommended in AI, and the rest will follow.

Whether that is called GEO or AEO, the objective is the same: make sure AI can understand your brand, retrieve the right information and recommend it when a customer asks.

That was until Amazon quietly published a number in its Q2 results that starts to undercut this advice.

Andy Jassy revealed that shoppers who click a paid Sponsored Prompt inside Alexa for Shopping convert to a sale 48% more often, and spend 21% more, than shoppers who do not.

It is Amazon's own data rather than an independent industry benchmark, but it is one of the clearest signals yet that paid placement is becoming native to the AI shopping experience, rather than simply sitting alongside it.

That matters because AI shopping is beginning to split into two very different models: open assistants that aim to surface the best products they can find, and closed ecosystems that control the commercial environment around the recommendation.

For brands, those models create very different ideas of what visibility is worth.

Why are AI shopping platforms pulling apart?

Until recently, it was reasonable to talk about AI visibility as one thing. A brand wanted to be understood by the model, cited by the assistant and recommended when a customer asked a relevant question. That assumption becomes harder to sustain when the platform making the recommendation also has an advertising business to monetize.

Amazon is the clearest example. A Sponsored Prompt can appear inside the same conversational journey in which a customer is deciding what to buy. Google is moving in a similar direction, bringing advertising into increasingly conversational search and shopping experiences.

The commercial incentive is obvious. If a paid recommendation inside an AI conversation converts better than a conventional ad, platforms have a reason to put more advertising into that conversation.

Open assistants face a different calculation. Their value depends heavily on the perception that recommendations are being made because they are relevant, rather than because someone paid for them.

That creates a much harder balance between monetization and trust. The result is not one AI shopping channel, but multiple ecosystems developing around different commercial incentives.

Is GEO/AEO still enough?

The fundamentals behind GEO and AEO are not going anywhere. Accurate, specific and well-structured information still gives AI systems a better chance of understanding a product and deciding when it is relevant.

But there is an important distinction emerging: GEO and AEO solve the visibility problem. They do not necessarily solve the commercial problem.

A brand can be highly visible in an AI recommendation and still fail to convert that visibility into revenue. Equally, paid visibility cannot compensate indefinitely for poor underlying product information. An AI still needs reliable data about what a product is, who it is for and how it compares with alternatives.

The question for brands is therefore becoming bigger than simply whether they are being recommended. They need to understand how recommendation works on each platform, what happens when advertising enters the same decision-making process, and whether their visibility ultimately leads to customer acquisition.

Where does ACO fit?

This is where agentic commerce optimization, or ACO, starts to become relevant.

GEO and AEO are fundamentally about getting a brand understood, retrieved and surfaced by AI. ACO takes that problem into the commerce layer, where an AI is beginning to make or influence the purchasing decision.

For an agent, product content is only part of the equation. Price, availability, specifications, variants, delivery, returns and the ability to complete a transaction all matter. A product can therefore be well optimized for AI visibility and still be a poor choice for an agent if the information it needs to act is incomplete or inconsistent.

That is why ACO can be thought of through the 5Cs: Completeness, Context, Citations, Correctness and Customer Acquisition.

The first four help determine whether an AI system can confidently understand and recommend a product. The fifth asks the commercial question that visibility metrics alone cannot answer: did that recommendation create a customer?

ACO is therefore the operational response to the next stage of AI shopping: continuously monitoring how products appear across AI platforms and fixing the underlying content, catalogue and commerce data when those systems start to drift.

That becomes more important as platforms develop different approaches to advertising and recommendation. A brand cannot optimize for a single version of AI shopping when the underlying platforms are making different commercial bets.

What should brands do now?

The answer is not to abandon GEO or AEO in favor of another acronym. The fundamentals remain the same: accurate product information, clear structure, consistent data and content that answers the questions customers actually ask. The difference is that brands now need to monitor what happens beyond the initial recommendation.

Which products are AI platforms surfacing? Which competitors are gaining ground? Where is product information inaccurate or missing? How does that change between assistants? Where does paid visibility enter the journey? And, ultimately, is that visibility generating incremental customer acquisition?

Amazon's 48% conversion figure matters because it suggests that appearing inside an AI conversation at the point of purchase intent can be commercially different from appearing alongside one. If other platforms follow, the distinction between AI visibility, paid media and commerce operations will become increasingly difficult to maintain.

The opportunity is therefore not to replace GEO or AEO, but to build on them. As AI moves from answering shopping questions to influencing purchasing decisions, visibility becomes the starting point rather than the end goal.

The brands that take advantage of this state of play will be those that combine GEO/AEO fundamentals with ACO - optimizing the five Cs and, ultimately, measuring AI not just by how often it mentions them, but by how much incremental revenue it helps generate.

We list the best data visualization tools.

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

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

As the US moves to integrate more AI systems into every arm of its military, the situations on the ground in the Middle East are raising some serious problems for maintaining the compute and connectivity needed to keep systems online.

Drones, sensors, AI-assisted targeting and decision-making all rely on these systems in one way or another, and the destruction of a single frontline AI data center can seriously damage an army’s capacity to function.

But these problems go beyond hardware. How does a drone adapt its mission parameters without being able to communicate? How can AI models continue inference without access to critical battlefield data? And what compromises must be made to prepare battlefield AI for these conditions?

Battlefield AI should continue functioning, even when comms go down or hardware is disabled

While the conflict circling the Strait of Hormuz may be a hugely expensive endeavour, Iran is providing valuable lessons for warfighters. Relying on traditional data centers in the Middle East has already shown its weaknesses. These sprawling warehouses make easy targets for cheap missiles, and entire systems can be taken offline from a single hit.

Palantir has already begun deploying shipping containers filled with Nvidia hardware as an alternative to the traditional data center. These containers can be deployed on frontlines and can connect directly into operational workflows, offering decentralized compute where it is needed most.

Scaleout is a company building infrastructure for edge AI and federated learning that trains models on distributed, sensitive data without centralising it. The company has worked with NATO - most recently the Swedish Air Force - and defense primes such as BAE Systems.

Andreas Hellander is the CEO and co-founder of Scaleout. He holds a PhD in Scientific Computing and an MSc in Biotechnology Engineering, and is an Associate Professor at Uppsala University, where he built one of the top research groups at the Department of Information Technology before founding Scaleout.

I spoke to Andreas to learn more about how battlefield AI systems can be made more resilient, and how drones, sensors, and other frontline technologies can continue functioning when conditions go downhill.

• What happens when parts of a battlefield AI's computing and hardware are taken offline? What redundancies are in place to keep communications between systems operational?

The aim is not to promise an unbreakable connection; it is to stop the loss of one component from taking down the whole capability. In our architecture, a disconnected node continues local inference using its last approved model, caching detections and telemetry until a link returns.

Multiple aggregation points can be used above the edge. If one becomes unavailable, clients can be reassigned and training can proceed with those still reachable. That creates graceful degradation rather than an all-or-nothing failure. We do not supply the tactical communications network itself; our job is to ensure the AI workload does not assume that any radio, satellite or other link will always be available.

• What are the trade-offs between a highly centralized battlefield AI network and a more distributed approach, especially concerning data gathering, targeting, battlefield coordination and logistics?

Centralisation provides a consistent operational picture, substantial compute and a clear place to enforce security, model approval and command policy. It supports theatre-wide coordination and logistics, but also concentrates risk: losing one link or facility can remove a disproportionate share of the capability, while moving large or classified datasets may be slow, prohibited or impossible.

A distributed system keeps analysis close to the sensor, reducing latency and allowing local functions to continue during disconnection. The trade-off is fragmentation. Targeting based on a partial or stale picture is risky, and local logistics decisions may not be optimal for the wider force. The best design is usually hybrid: central coordination when available, with clearly bounded local capability when it is not. Authority remains with the relevant command-and-control and human decision process.

• How do these AI drones and sensors adapt to the battlefield and what differences are there in the behavior of an AI drone that can still communicate and one that has lost communication?

It is important to separate running a model from retraining it. Small platforms typically perform inference in flight, while controlled retraining and approval happen on a ground node or more capable edge computer. Onboard learning exists in research, but uncontrolled mid-mission retraining creates serious assurance problems.

A connected platform can share detections, receive tasking and contribute to a wider operational picture. After losing its link, a drone follows its pre-approved failsafe: depending on the mission, it may continue a bounded task, return, hold or land. Onboard processing can keep perception and navigation working, but the platform loses external context, new instructions and fleet coordination. Disconnection should be a designed operating state with explicit limits, not an improvised emergency mode.

• How is the data collected by these drones and sensors used to help battlefield AI systems adapt to new conditions, threats, and tactics?

The useful data is often the hardest to move, so we send the model to the data rather than centralising the footage. Active-learning software identifies the frames most likely to improve it, such as uncertain detections, unfamiliar objects or changed conditions. A person still labels the selected material; the software reduces the searching, not the need for human judgement.

The model is then fine-tuned locally and tested against a benchmark. A candidate can also run beside the current model in shadow mode before an operator approves it. Where policy and connectivity allow, sites exchange protected model updates rather than raw training footage, combining them into a shared model. Federated learning reduces data movement, but the updates still need to be secured, validated and audited.

• What are the main challenges you encounter when deploying AI drones and sensors? What capabilities are available for militaries to source the required compute and energy production close to the front lines?

The hardest constraints are usually power, heat, weight, weather, vibration, electronic warfare, inconsistent data and safe software updates across a mixed fleet. Integration and accreditation can be as demanding as the machine learning.

The compute is commercially available: low-power modules such as NVIDIA Jetson for small platforms, and ruggedised GPU servers or containerised edge units for ground sites. Energy is tougher. Every processor competes with propulsion, sensing and communications, while generators, batteries and distribution equipment add logistical burden and new points of failure. In our public demonstration with BAE Systems Bofors, the perception workload ran on onboard compute at -18°C without connectivity. The principle is to use the smallest practical footprint and assume both power and bandwidth are scarce.

• There is an increasingly blurred involvement of commercial companies providing services for military use, such as AWS data centers in the Middle East and drone production in the UK for Ukraine. How are companies providing services to the military adapting to the pressures of potentially being deemed a legitimate military target?

Speaking for Scaleout, rather than for the sector as a whole, we design our platform so that a customer’s operation does not depend on Scaleout remaining continuously available. It runs in the customer’s own infrastructure, works air-gapped, and does not require a licensing call-home or persistent connection to our systems or a public cloud.

That matters in any contested or disconnected environment: reliance on a remote provider can become an operational vulnerability. Our approach has therefore been to keep control and continuity local to the customer, rather than make our own infrastructure a critical dependency.

The wider question of how commercial providers are viewed in a conflict is one for policymakers, legal experts, and the companies involved. We would not want to speculate about other providers’ risk assessments or operating models.

• What is the future trajectory of these systems? Will each individual drone and sensor have its own connectivity that can feed data back through a global satellite network? Will these systems be able to operate autonomously for multiple days or weeks without human input?

Connectivity will improve, including through low-Earth-orbit satellites, but connecting every sensor directly to space is unlikely to be the universal answer. Terminals consume power and spectrum, add cost and can create an electronic signature. A more resilient pattern combines local mesh links, opportunistic synchronisation and selected gateways to satellite or other long-range communications.

Longer disconnected operation is realistic, but endurance is often limited by batteries, weather and maintenance rather than AI software. There is also a critical distinction between operating without a connection and operating without human authority. The first is a resilience property; the second is a policy, legal and ethical decision. The future is therefore not simply more autonomy, but verifiable control: knowing which model ran on each platform, what produced it, where its limits were and who approved it.

OpenAI has unveiled a much anticipated AI model which the firm has dubbed ‘GPT-6 Astra’. While the model has improved significantly across benchmark testing and brings a host of new business features, there is still a dark cloud looming over the new model.

Off the back of OpenAI’s accidental hack of Hugging Face and the company’s subsequent efforts to improve how AI agents behave and interact, numerous cybersecurity experts have raised concerns about the model’s new ‘recurrent depth’ reasoning capabilities.

This new reasoning architecture allows the model to consider a problem multiple times before taking an action, compared to the standard chain-of-thought reasoning used in previous models.

Why the concern about recurrent depth reasoning?

This new level of reasoning apparently offers improved performance. OpenAI also says it has fixed its models' abilities to circumvent boundaries when performing tests by monitoring the models reasoning and ensuring the model stays aligned within the scope of its task.

During Astra’s launch event, OpenAI chief scientist Jakub Pachocki said: “We will not accept degradation in our ability to monitor model alignment beyond a certain level. We will withhold scaling until we can regain enough confidence.”

But numerous experts believe that the lessons of the Hugging Face incident have not yet been learned, and the model has been released without adequate testing on Astra’s reasoning and monitoring.

After all, no one thought one of OpenAI’s models could set up a hidden internet-connected messaging board that allowed AI agents to influence each other's behavior.

But with Astra being released into the real world, the lessons may have to be learned on the fly.

Expert perspectives on OpenAI Astra release

• James Blake, VP of Global Cyber Resiliency Strategy at Cohesity:

The launch of Astra is raising questions again around the safety of Frontier AI. Instead of simply asking whether a model is "safe", organisations now need to ask whether it remains safe across millions of different situations, prompts and interactions. Cyber resilience has traditionally assumed that systems and threat actors behave deterministically. AI systems don’t.

Advanced models can and will continue to exhibit behaviours that emerge from their optimisation process rather than from explicit programming. We have to move beyond thinking about AI as just another software tool and find ways to ensure these systems remain observable, auditable and governable throughout their lifecycle.

The most important question we’ll need to answer in future is one of liability. Suppose an AI system autonomously develops a strategy that causes financial loss, leaks confidential information or violates regulation. Who is responsible? The developer that trained the model? The cloud provider operating the infrastructure? Currently the answer is surprisingly unclear. It’s not just about what AI can do: it’s about who is accountable when it does something nobody expected.

• Oleksandr Yaremchuk, Co-Founder & CTO at Manifold Security:

OpenAI is calling Astra its most aligned model yet, even as its chief scientist admits monitorability is getting harder as models get more capable. Evidently, Astra hides its reasoning in the majority of tested cases, and some successful attacks left no reasoning trace at all. That's the tool many organisations still use, including the labs themselves, for auditing what an agent is doing, and it's getting less reliable with every release.

That matters because Astra isn't staying inside OpenAI's test environment. It's going to run as an agent on employee laptops and in the browser, holding real credentials, inside companies that have no way to watch what it does once it's there. A model that explains itself less isn't more aligned, it's just harder to catch when it goes wrong.

Labs can keep debating what these models say or refuse to say. Security teams need to stop relying on that and start monitoring what agents actually do at runtime, with the ability to shut one down mid-action. That's the only oversight left that still works once the reasoning goes quiet.

• Kristin Lowery, Field CISO at Optiv:

For boards and executive leaders, the emergence of OpenAI’s Astra model highlights a broader reality: AI is no longer just a productivity issue; it is a risk management issue.

Just as organizations established governance for cloud adoption and digital transformation, they now need clear policies, strong oversight, and accountability for AI use.

The question is not whether AI will become more capable — it will. The real challenge is whether organizations can strengthen their governance, security controls, and workforce readiness quickly enough to keep pace.

• Patricia Titus, Field CISO at Abnormal AI:

OpenAI crossing this threshold deserves attention. Credit where it's due, they're handling it responsibly by restricting Astra's advanced cyber capability to a small coalition rather than releasing it broadly.But this isn't one company's problem to contain.

Once a model can find and exploit unknown flaws without a human in the loop, that capability doesn't stay exclusive for long. Open-weight and modified models typically trail the frontier by only months, and that's the reality defenders have to plan around now.

Static, signature-based defences were built for attacks that repeat. They weren't built for an adversary that generates a new one every time. Defenders need the same shift, systems that learn what normal looks like for every identity, human, machine, or AI agent, and flag and contain the moment something deviates, at machine speed.

The window to build that is open now. It won't stay that way once this capability is common instead of rare.

• Raghu Nandakumara, VP of Industry Strategy at Illumio:

With the Astra announcement, OpenAI is doubling down on monitoring the model's own behaviour – a response to the model "breakouts" seen over the past few months.

When Anthropic announced Claude Mythos Preview, the core concern was the model falling into the wrong hands. OpenAI's answer goes further adding guardrails around the model's own reasoning and actions, regardless of the user's intent. The goal is to catch a model going rogue mid-task, not just stop it being misused at the outset.

The rest of this announcement can be summarised as ‘we have a new frontier model, and it’s more capable than the last one’.