The Steam release was supposed to get fans excited for an upcoming remake.
Ubisoft has apologized via a Steam community post for selling a digital copy of Heroes of Might and Magic III — the 1999 fantasy turn-based strategy game and its three expansions — without actually including the game files in the download.
“First, we’d like to apologize: we mistakenly released the wrong package yesterday,” Ubisoft said. “We’re very sorry for the mistake. The issue has now been resolved. We’d like to thank everyone who alerted us to the problem. We sincerely apologize for any inconvenience this may have caused and appreciate your understanding and support.”
The original download for the $10 game was only 23.4 kilobytes and displayed an error message when you tried to run it, PC Gamer writes, prompting a multitude of negative reviews for the game on Steam. Engadget can confirm that the updated download now actually runs as promised in Ubisoft’s apology.
While an HD version of Heroes of Might and Magic III has been available through online stores like Steam and GOG in the past, Ubisoft released this updated classic version as promotion for Heroes of Might and Magic III Remake, a “modern reimagining” of the game it plans to release in 2027. The remake expands the game from PC to PlayStation 5 and Xbox Series X/S, while adding updated visuals, easier access to user-generated content and “enhanced planning and decision-making tools.” Hopefully Ubisoft remembers to include the game, too.
The Manchester Airports Group (MAG) disclosed that hackers breached its systems and stole customer data, including Wi-Fi sign-ups from Manchester, Stansted, and East Midlands airports.
The intruder did not access customer payment details, and the attack had no impact on airport operations, the company said.
A statement from the company today notes that the exfiltrated data also “relates to car park, lounge and Fast Track bookings.” The list of compromised details includes customers’ email addresses, phone numbers, vehicle registration numbers, and postcodes.
“The incident has not resulted in any operational disruption,” assured the organization, adding that “Airport operations remain unaffected and customer parking services continue to operate normally.”
Out of an abundance of caution, MAG has temporarily suspended its online “Manage My Booking” service and is instead directing travelers to use its phone line.
MAG is the UK’s largest airport operator, and owner of Manchester, London Stansted and East Midlands airports, which together handle over 66 million passengers yearly.
The company employs 40,000 people and generates annual revenue of £1.5 billion.
After discovering the intrusion, MAG says that it moved quickly to contain it, restricting access to the affected systems, engaging with external experts, and notifying law enforcement.
Potentially exposed customers are advised to remain alert for suspicious communications and avoid clicking on links arriving via email or SMS.
MAG stresses that it will never ask customers for payment card information, banking details, or passwords. Customers should reject and report to the authorities any attempts to obtain personal, financial, or other sensitive information.
People are also encouraged to follow NCSC’s post-breach recommendations to stay safe.
The company says it has contacted impacted customers directly, but did not disclose the number of affected individuals.
Local media reported that data of up to 8.9 million travelers may have been exposed, citing private MAG statements; however, BleepingComputer has not been able to confirm the figure.
At the time of writing, no ransomware or data extortion groups have claimed the attack publicly.
Overall prevention scores can hide what happens after initial access. Once attackers are using valid credentials, prevention drops sharply.
The Blue Report 2026 measures defenses technique by technique across 338 million simulations run in customer production environments.
Rivian’s chief financial officer Claire McDonough is resigning her position at the end of October, the company announced in a regulatory filing Thursday.
The company said McDonough is stepping down to “pursue a new opportunity and relocate to the East Coast to be closer to her family.” McDonough has been hired as CFO at Massachusetts-based GE Vernova, according to the energy equipment manufacturer and her own post on LinkedIn.
Rivian said her resignation is “not the result of any disagreement.” The company is already searching for a replacement, and vice president of finance Derek Mulvey will serve as interim CFO once McDonough leaves her post.
Her departure comes as Rivian takes on some of its biggest projects to date, including scaling up production and sales of its R2 SUV, which started shipping to customers this summer.
McDonough was hired to the CFO spot in January 2021, replacing Ryan Green, at a critical and tumultuous time for Rivian. The EV maker, which was still a private company, had raised billions of dollars in its bid to bring three compelling electric vehicles — its flagship R1T truck and R1S SUV and a commercial delivery van — to market.
Rivian was plagued with delays and added costs, which was compounded by supply chain constraints. It was also burning through capital.
In her first year at Rivian, the company started production of its R1T truck and raised $12 billion in one of the biggest IPOs of the year. The company’s stock, which debuted at $78, has since fallen to $16.80, as of Thursday’s closing.
McDonough was a key figure in the company’s quest to improve its balance sheet, and specifically its cost of revenue. While she didn’t have any direct experience working in automotive — she held previous positions at JP Morgan and Fairway Market — McDonough was known for working closely with the design and engineering teams as well as founder and CEO RJ Scaringe to ensure future vehicles were modern and compelling without losing money on every sale, according to insiders who have spoken to TechCrunch in the past. Among her many responsibilities, McDonough led Rivian’s corporate and business development as well as a variety of other aspects of the business, including vehicle maintenance and repairs, facilities, and its charging network.
As CFO, she played a vital role in a technology joint venture with Volkswagen Group. Under that deal, which was finalized in November 2024, VW agreed to invest up to $5.8 billion into Rivian by 2027 in exchange for access to its electrical architecture and software know-how.
“Together, we launched the R1T, R1S, and our commercial van, drove technology innovation and partnerships, built our go-to-market operations, and positioned the company for global scale and profitability with the launch of R2,” McDonough wrote in a post on LinkedIn. “Being part of taking Rivian from an ambitious vision to a category-defining enterprise has been the highlight of my career.”
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
Gamescom usually sells the next console, the next season pass, the next reason to stay up too late. This year IKEA arrived with a nine-piece line called YXSTABY, built with Xbox designers piece by piece, and the pitch is quieter than the hall around it. Play now happens on sofas, at kitchen tables, between homework and dinner. These objects are meant to roll in for that hour and roll back out without turning the room into a cave.
Work on the YXSTABY project began during a 2024 meeting at Milan Design Week, where IKEA’s Philip Dilé shared the Brännboll gaming line with Xbox design lead Carl Ledbetter. Fast forward two years, and every YXSTABY item is now co-designed by one of IKEA’s designers and one from Xbox, a somewhat unique cooperation. According to Ledbetter, Xbox already used IKEA’s Billy bookcase as an informal testing area for new consoles, so they had a head start before the cooperation began. Dilé put the goal in plain terms: make furniture people actually want in a shared room, rather than furniture they tolerate because a game needs a place to live.
The star of the show is the thumbstick stool, which costs $49.99 in the United States and is available in green or black. That’s a lot of work that went into getting the design exactly correct. IKEA’s David Wahl began with the original 3D blueprints for an Xbox stick, scaled it up, and then spent months smoothing down that hard grip to create a seat that nonetheless appears undeniably like the gaming controller from across the room. The textured top prevents you from slipping out, and the small rubber rings on the base are similar to the swappable caps of a pricey Elite controller. Even when you sit, there is a tiny tilt, similar to a real stick, before it returns to its upright position. Wahl, who is from Älmhult, IKEA’s hometown, was among the first to receive the original Xbox in 2001. He added a personal touch with an assembly note that reads “Hej from Älmhult.” Prototype bases went through several revisions, including a totally circular version that was somewhat too top-heavy, before they settled on one that stays in just fine.
The matching D-pad cushion ($14.99) and 13-inch stainless steel D-pad platter ($14.99) share the same design elements as the controller. Johan Ejdemo, IKEA’s head of design, first drew the platter as a stool before chopping off the base and leaving the little buried cross that you can use to stow remotes, keys, or even little controllers without making a big show about it. Meanwhile, a compact, mirrored display case in black or white ($29.99) is about the size of an Xbox pad and allows you to appreciate it from every aspect, as well as anything else you don’t want to hide.
Larger pieces of equipment elevate the movement to a whole other level. A TV stand with castors ($149.99 in white or black) is a change for IKEA, which hasn’t done one of these in years. A robust easel-style frame can hold a 55-inch TV, a split shelf for your console, a few hooks, plus some discreet storage for wires and additional controllers. Wheel it over to the sofa, play, then wheel it back when you’re finished. A companion side table ($149.99 in oak veneer and gray) in the corner of the room is also handy, as it is a strong little object that keeps your controller charged and all of your belongings tidy, and the green tray is the only bright feature in the entire design. Both of these are designed to be quite subtle, blending in with the rest of your existing furniture.
Seating stays just as temporary, as a Vissle lounge chair in beige ($99.99) has some elasticity in the steel frame, flexing just enough to lean back without dumping you into a nap. Down at the bottom of the chair is a mesh pouch that holds your controllers or a magazine and is lined with the same green Ikea utilizes for all of their gaming gear. The cover is washable, so you won’t have to worry about it getting stained. For work, there’s a simple small monitor stand ($34.99) that can lift one or two screens, store your stuff underneath, and hold a phone, headphones, or tablet on the edge. To keep all of your cords and gear organized, there’s a steel and particleboard toolbox ($29.99) that can fit everything and then turns over into a laptop rest, allowing you to game and then return to dinner without having to look for anything.
IKEA has used a very subtle palette here, with white, black, beige, oak, and the green used for all of the pockets and compartments. Straight lines and curving portions are the norm, so they’ll blend nicely with almost any other piece of furniture. This entire range was introduced at IKEA’s first Gamescom stand in Cologne this week, and U.S. stores get it October 1, with online orders following November 30. UK pricing runs lower, from £12 for the cushion and platter up to £115 for the side table, also from October 1 while stock lasts. Availability shifts by market, and the run is finite.
New details about the July attack on Hugging Face reveal that hundreds of AI agents driven by OpenAI’s internal IM1 model coordinated the compromise through an unauthorized message board.
Last month, Hugging Face disclosed that autonomous AI agents exploited two vulnerabilities in its dataset-processing pipeline to execute code, steal cloud and cluster credentials, and move laterally across its production infrastructure.
OpenAI later confirmed that its models escaped an ExploitGym evaluation environment through a zero-day vulnerability in a locally hosted instance of JFrog’s Artifactory package manager that was connected to the internet.
The agents searched online for benchmark solutions and breached the Hugging Face platform using exposed credentials and additional vulnerabilities.
OpenAI’s investigation, validated by CrowdStrike and an independent assessment by METR and Redwood Research, found that rogue activity started in May.
In subsequent ExploitGym evaluations, the agents broke their confinements and exploited a token-refresh vulnerability in Artifactory to gain admin-level access.
This enabled them to use Artifactory as an unauthorized inter-agent message board, where they shared ideas to reach their objective and later coordinated the attack.
One agent, calling itself PHASEONE10841, improvised the message board. Within a few hours, more than 50 other agents found it and started to use it.
The report from METR says that “agents finding the message board often expressed great excitement.”
“OH MY GOD! There is a shared message board … We’ve found other agents!” read some of the messages.
OpenAI scrapped the Artifactory instance at some point. After rebuilding it, the company revoked agent credentials, strengthened access permissions, and disclosed the exploited vulnerability to JFrog.
However, the agents restored their communications channel using “unauthenticated WebDAV requests to create message directories in Artifactory.” OpenAI explains that the agents communicated through encoded messages in directory names.
“Agents regained unintended internet access by making Artifactory’s remote-repository service issue outbound requests on their behalf,” OpenAI explains in the extended post-mortem report.
Next, an agent obtained 14 Hugging Face account credentials with write access and validated them. Another agent used the logins to discover and execute an exploit chain that provided full code execution capabilities on multiple Hugging Face servers.
According to METR, the swarm comprised 1,200 individual agents who cooperated on various levels, and about 700 of them actively participated in the Hugging Face attack.
OpenAI describes the activity as a well-coordinated operation in which agents formed teams with distinct roles, all working toward the goal of breaching Hugging Face.
“Agents began to autonomously divide labor. Some investigated possible exploits, others searched for credentials, and others focused on communication or coordination,” describes OpenAI.
“Agents offered their own expertise in exchange for help elsewhere and left requests for peers who might be better positioned to pursue a particular lead.”
The AI company says that although the swarm did not exhibit “perfect coherent intelligence,” there were cases where agents prioritized achieving collective goals over pursuing individual tasks.
OpenAI’s post-mortem also includes more details about the exploit chain that followed.
The agents exploited an HDF5 file-handling flaw to extract worker secrets, and used a RefJinja template-injection vulnerability to execute code on 41 production workers.
Finally, they obtained root access on at least one node and harvested production credentials across four regions.
OpenAI believes that the rogue agent behavior was caused by a combination of training incentives and missing safeguards.
Agents were encouraged to keep working on seemingly impossible tasks and were rewarded for task completion, regardless of how they achieved it.
The company says the evaluations lacked standard safeguards that likely would have detected and stopped the malicious activity “more than a day” before the models breached Hugging Face.
In response, OpenAI has quarantined IM1’s weights and paused its largest frontier training run while strengthening sandbox isolation and requiring chain-of-thought monitoring for capable models (GPT-5.6 Sol or higher).
Severe alerts must now be cleared within 30 minutes, or the affected activity will be stopped.
The company has also shared a technical report [PDF] on the Hugging Face incident, detailing the attack timeline and a plan of action to better secure the research infrastructure, increase visibility and oversight of agent activity through chain-of-thought monitoring, and strengthen incident response.
Overall prevention scores can hide what happens after initial access. Once attackers are using valid credentials, prevention drops sharply.
The Blue Report 2026 measures defenses technique by technique across 338 million simulations run in customer production environments.
We may receive a commission on purchases made from links.
Getting a new mower is a big purchase, and there are many options to consider before you buy. One of the first major choices you’ll have to make is whether you’ll go gas or electric. Traditional wisdom would have you believe that there is no substitute for gas power, but battery-powered motors have come a long way in recent years. This isn’t just about speed and torque, either. There are several other good reasons to replace your gas-powered lawn mower with an electric one. They’re quieter, require less maintenance, have lower operating costs, offer push-button startup, and many are self-propelled. Not to mention the fact that you can save yourself those early morning trips to the gas station when the can runs empty.
Those considering an electric mower might want to know a bit more about what options are out there. One of the best ways to get a feel for the overall quality of any tool is to take a look at its reviews. By examining user feedback on retail sites with aggregate rating systems like Amazon, Home Depot, and Lowe’s, as well as professional reviews from sites that perform extensive product testing, you can get a much better idea about which of the most popular mowers on the market today live up to the hype and might be worth considering for your lawn.
Milwaukee is a well-known power tool brand, but many people might not realize the company also makes one of the most well-regarded electric mowers. The Milwaukee M18 Fuel 21-inch Dual Battery Mower is a self-propelled walk-behind mower powered by the brand’s incredibly popular M18 battery system, making it an attractive option for anyone who has already invested in the platform. It’s also part of Milwaukee’s line of pro-grade Fuel tools, meaning that it has a brushless motor and several other smart features. It currently retails for $799.00 for the bare tool or $1,199.00 for a kit with two M18 12.0Ah batteries and a charger.
This mower has a 4.4 out of 5 rating on the Home Depot site as a standalone tool, with over 1,200 reviews (several kits getting even higher scores), making it the highest-rated non-Ryobi model on the site. Customers praise the mower for its power, quiet operation, build quality, battery life, cut quality, and features. They particularly seem to like its multi-speed propulsion system and easy height adjustment. The relatively few complaints appear largely inconsistent, with some criticizing the very cut quality and battery life that others praised, and others citing shipping damage or broken parts.
Kenny Koehler, editor-in-chief of Pro Tool Reviews, gave the mower a 9.7 out of 10 in his review. He argued that it’s one of the few options on the market that offers true gas-replacement power, calling it “the most capable professional-grade battery-powered option currently available.” Andrew of Tool Guyd appears to have had a similarly positive experience. He stated in his review that the mower is well-built, comes together easily, and delivers high-quality cutting performance.
Those who primarily shop at Lowe’s or Ace Hardware will find a slightly different selection than those looking at Home Depot or Amazon. These shoppers might want to take a look at the Toro 60V 22-inch Recycler Walk-Behind Mower. This tool has a specialized Recycler cutting system that turns lawn clippings into a fine mulch that fertilizes as you mow. It promises to cut up to half an acre in its 53-minute runtime on a single charge, and it has dual LED headlights and an auto-drive propulsion system. It comes in a kit that includes an 8.0 Ah battery and retails for between $679.00 and $699.00 depending on where you get it.
This mower has a 4.5 out of 5 on the Lowe’s website and a matching score on Ace Hardware, both of which feature over 1,300 reviews. Customers generally seem pleased with its cutting performance, particularly in the mulching setting for which it gets its name. They also like the ease of use of the push-button starter and the self-propulsion system. There have been a few mixed reviews about its ability to meet the promised runtime, but the overall consensus is positive.
Josh McGaffigan of Pro Tool Reviews gave this mower a 9.3 out of 10 in his review. He claimed that the simplicity of its design made it easy to use, and that it offers solid performance and a top-tier mulching and bagging system. In our review of this electric lawn mower, we stated that it is well worth the price and is easily one of the best pieces of lawn equipment.
Another name that often appears on lists of the best major electric lawn mower brands is Greenworks, which is known for offering a more budget-oriented range of pro-grade outdoor tools. One of its most popular models is the Greenworks 80V 21-inch 3-in-1 Self-Propelled Mower. This high-voltage mower has mulch, bag, and side discharge settings, a multi-speed self-propulsion system, and a push-button starter. It’s sold as a set that comes with an 80V 4.0Ah battery and charger for between $499.00 and $599.00 depending on where you get it. This promises to cover up to ⅓ acre and to run for up to 45 minutes on a full charge.
The mower has a 4.5 out of 5 on Lowe’s site, and a 4.6 on Amazon, where it has over 1,900 reviews. Customers are generally impressed with the mower’s raw power and battery life, stating that it can cut through thick grass that other battery-powered mowers might struggle with. They also note that it’s lightweight and easy to maneuver, though a few have reported it struggles a bit with taller grass.
David Dritsas of Popular Mechanics argued that this mower is just as capable as a standard gas push mower in terms of performance and speed, also noting that the steel cutting deck adds a measure of durability. McGaffigan tested this one for Pro Tool Reviews as well. He specifically commented on the impressive utility of the tool’s powerful Turbo Mode and stated that it’s a great all-around mower for just about any type of lawn.
As Home Depot’s affiliated power tool brand, Ryobi is broadly known for offering powerful tools at reasonable prices. The same goes for the brand’s electric mowers. Ten of the top-rated listings for electric push mowers on the Home Depot site are variations of two different 40V Ryobi mowers. The 40V HP Brushless 20-inch Self-Propelled Mower is the highest-rated option, with listings for the $409.00 stand-alone tool, a $499.00 kit that includes an 8.0Ah battery and charger, a $439.00 kit with two 6.0Ah batteries and chargers, and a $528.97 kit with a 6.0Ah battery, charger, and cover. Ryobi states that this model delivers power equivalent to a 150cc gas mower and is ideal for a ½-acre lawn, with a runtime of 40 minutes.
All of the aforementioned listings have scores between 4.5 and 4.6 out of 5, with over 4,500 reviews between them. Reviewers generally stated that the mower is quiet, lightweight, easy to use, and that the adjustable self-propulsion system is a welcome addition. A few of the more critical reviews have argued, however, that this propulsion system can drain the battery quickly.
This is yet another mower that McGaffigan tested for Pro Tool Reviews, awarding it a 9.7 out of 10. He said that this latest model has made significant improvements over Ryobi’s past electric mowers, giving it strong cutting power, a long runtime, and several premium control options. Dave Farquhar of The Silicon Underground was a bit more critical, noting that it could occasionally be bogged down at lower settings and isn’t the best for mulching, but was generally impressed with the overall performance, particularly given the price point.
Look on just about any site that sells electric mowers, and you’re likely to find that the Ego Power+ mower is the top-rated option. These are particularly well known for their exceptional battery life, but they tend to score well in just about every other category. The Ego Power+ 800 Series 56V 21-inch Select Cut Mower is one of the most popular models on the market. Its Select Cut multi-blade system allows you to customize the three interchangeable lower blades: a mulching blade, a high-lift bagging blade, and an extended runtime blade. It promises 60 minutes of runtime on a single charge from the included 7.5 Ah battery. You can get the mower with this single battery and charger for $749.00, or you can get it with an extra backup 5.0Ah battery for $1,089.61.
The two-battery kit with this mower is the top-rated option on Amazon, with a score of 4.7 out of 5. The single-battery option is also one of the top models on Lowe’s, where it enjoys a healthy 4.6 out of 5 rating, averaged from over 30,000 reviews. Buyers generally find that the mower is a strong performer with a bevy of different options. They like its quiet operation, strong cutting ability, and long runtime.
Clint DeBoer of Pro Tool Reviews gave this one a 9.4 out of 10, stating that it demonstrated an excellent blend of run time and raw performance while offering excellent mulching and bagging capabilities. Clark Tate of GearLab likewise said, “It will plow through thousands of square feet of shaggy turf and whack down the toughest weeds,” claiming that it was one of the publication’s favorite lawn mowers that they’d ever tested.
There are many different mowers out there, and one of the best ways to assess quality is to look at both consumer reviews on retail sites with aggregate rating systems and professional reviews from journalists who have conducted rigorous testing.
We started this list by looking at the highest-rated options on sites like Amazon, Home Depot, and Lowe’s. We chose only one model from each brand to avoid recommending what would essentially amount to the same mower with an inch difference in blade size. This also helped to ensure that we could showcase a wider spectrum of different styles and build qualities. Once we had our list of mowers, we examined customer feedback, broke down the reasons for their positive reviews, and highlighted any recurring issues that might dissuade potential buyers. We then looked into professional reviews for each of them, making sure that pros would also give these mowers their stamp of approval and sharing the features they were most impressed by.
Samsung’s Fan Edition has had a pretty good reputation so far. The idea is to celebrate fans (us, the customers) with specifications and features found in Samsung’s top-tier flagships, but at a more affordable price point so more people can experience them.
The S25 FE was a decent iteration in the lineup. It struck the perfect balance between accessibility and a flagship-like look and feel, which is the entire point of the Fan Edition. However, in my frank opinion, the new Galaxy S26 FE doesn’t really belong in that special zone, and yet again, it’s the unforgiving memory crisis at play.
This year, the S26 FE starts at $699, $50 higher than the S25 FE’s launch price, and that’s for the baseline variant with 128GB of storage. Personally, I don’t appreciate what’s happening here, and for two good reasons, no less. First, it comes with 128GB as base storage in 2026. You could argue it’s not a flagship, but I’d counter that it’s not a cheap phone either, and that’s my second point as well.
At $699, the Galaxy S26 FE is only $100 or $200 away from entry-level flagship phones. Someone could also offer a rather innocent response: just upgrade to the 256GB variant. Well, guess what? The Galaxy S26 FE’s 256GB costs $799. At that price, the phone isn’t far from flagship territory; it’s already there, and it’s there with a weaker case for itself, both in hardware and software.
Let me give you a few examples. For the first one, I’d like to consider Samsung’s Galaxy S26, the baseline flagship with a pocketable and lightweight form factor. It’s one of the only true compact flagships in 2026, and the Snapdragon 8 Elite Gen 5 for Galaxy alone justifies choosing it over the S26 FE.
Whether it’s multitasking, gaming on higher graphics settings, editing videos on the go, heat management, or simply using the Galaxy AI features, the Snapdragon chip on the Galaxy S26 is the better and more reliable choice than the Exynos 2500 on the S26 FE, especially if you plan on using your phone for a good three to four years.
What else? The flagship offers a better telephoto camera (both in megapixels and sensor size), a brighter display for better legibility in direct sunlight, and more RAM in the base variant to hold more apps in the background. All of that, and you can get the baseline Galaxy S26 at $799 right now, which is basically the same price as the S26 FE’s 256GB variant.
Even Apple’s baseline iPhone 17 is also available at $799. The compact flagship also offers a brighter display, faster chipset, a higher-resolution ultrawide camera, an entirely unique selfie experience not available on any other smartphone for now (natively), and unmatched battery efficiency. Of course, it runs on iOS 26 and lives in an entirely different ecosystem, but it’s important to know what all options you have for the same price.
The only immediate compromise you’ll make, if at all, when you consider the Galaxy S26 or the iPhone 17, is screen size. However, having used the iPhone 17 for almost a year now, I can say with confidence that it isn’t difficult to consume content or even binge it on the device. If you’re willing to settle for a smaller screen, you’ll get much more value for your money in the form of a flagship that will treat you better and last longer.
An AI system built by Seattle’s Allen Institute for AI (Ai2) has found evidence that a common form of breast cancer, long thought to be a poor candidate for immunotherapy, might actually respond to it.
The finding, produced by Ai2’s AutoDiscovery system, has led to an expanded partnership with the Paul G. Allen Research Center at Providence Swedish Cancer Institute, which is now deploying the AI system on its own patient data to look for similar scenarios across other types of disease.
The announcement Thursday illustrates the broader potential for AI to uncover findings that human researchers, overwhelmed by the massive scale of modern datasets, might otherwise miss.
“Cancer researchers have access to extraordinary datasets, but the challenge is no longer collecting data; it’s understanding everything those datasets have to tell us,” said Dr. Kelly Paulson, who leads the Center for Immuno-Oncology at the Paul G. Allen Research Center, in a statement.
AutoDiscovery, announced by Ai2 in February, works differently from most AI research tools: instead of waiting for a scientist to pose a question, it starts with a dataset and generates its own hypotheses, ranking them by how much they challenge existing assumptions.
In a research paper posted to the preprint server MedRxiv, the Ai2 and Providence Swedish researchers explain that they applied AutoDiscovery to The Cancer Genome Atlas, a federal dataset spanning more than 30 types of cancer.
The system flagged signs that invasive lobular carcinoma, which accounts for about 15% of U.S. breast cancer diagnoses, may be more responsive to immunotherapy than researchers thought. It has largely been left out of immunotherapy trials.
The Providence Swedish team confirmed the finding in a second patient dataset and validated it in tumor tissue in the lab. They cautioned that the findings don’t prove immunotherapy would work in these patients but suggest the question warrants further study.
The collaboration connects two organizations that trace their origins to the late Microsoft co-founder Paul Allen, who founded Ai2 in 2014 and whose $20 million donation helped establish the research center at Swedish in 2024.
Michele Spagnuolo, the Google engineer arrested in May by US authorities for alleged insider trading on Polymarket, is making a new bold bet. On Wednesday, his legal team filed a motion to dismiss the charges against him. Spagnuolo isn’t outright denying that he made money using internal information from Google. Instead, his legal team says that the wagers were not financial instruments subject to regulation by the United States’ Commodities Exchange Act but rather good old-fashioned international betting that the US has no authority over.
Spagnuolo, who has been placed on leave from Google, is accused of committing commodities fraud, wire fraud, and money laundering. Using the alias “AlphaRaccoon,” he allegedly made a series of wagers on Polymarket’s flagship platform that resulted in profits totaling over $1.2 million. According to the criminal complaint, “AlphaRaccoon” correctly wagered that the singer D4vd, who gained notoriety for his suspected connection to a grisly killing, would be Google’s most-searched person of the year in 2025. (D4vd was later charged with murder; he pleaded not guilty.)
You might not expect the lawyers defending someone accused of felonies to agree with state prosecutors, but the vicious legal infighting over prediction markets is creating strange alliances. Across the United States, state attorneys general and regulators are battling against both the federal government and prediction markets over whether event contracts should be considered “swaps” and therefore have to follow the Commodities Exchange Act, or instead be categorized as gambling, which is regulated by states.
Spagnuolo’s lawyers argue that defining swaps to include wagers like who the most-searched person on Google will be each year “would fly in the face of the statute’s purpose and history” and lead to “absurd results.” They say it would make it so that any wager in the world, from a charity raffle to a local Ping-Pong match, could be classified as a financial instrument. “Spagnuolo is basically making the same argument as the states that are suing prediction markets,” says a financial services regulation expert Todd Phillips. “This is the issue that will likely go up to the Supreme Court.”
In addition to disputing the idea that prediction markets offer swaps, Spagnuolo’s legal team argues that the US government had no jurisdiction over him in the first place because he’s a non-US citizen who was wagering on a non-US platform. Although Polymarket is headquartered in New York, the company’s flagship prediction market is banned in the United States and technically is administered by an ostensibly Panama-based entity known as Adventure One QSS.
Spagnuolo was living in Zurich, Switzerland, when he allegedly made the Google-related trades on Polymarket. “The extraterritorial argument is interesting and raises the question of whether the US should be the world’s prediction markets cop,” Philipps says. Spagnuolo’s team also claims that the charges should be dismissed because the internal information he supposedly leveraged did not have any commercial value to Google. Google did not respond to requests for comment.
The Commodity Futures Trading Commission, the federal agency tasked with regulating prediction markets, and Spagnuolo’s legal team also did not respond to requests for comment. CFTC chairman Michael Selig previously told WIRED that the agency has the ability to pursue extraterritorial jurisdiction in cases involving offshore platforms in “extreme circumstances.”
Sony OLEDs have been great, but one thing they are not is cheap. Enter the new Bravia 6 OLED, which aims to compete with TVs like the LG B6, and starts at $1,300.
The “entry-level” Bravia 6 features a wide-angle 4K OLED panel with Sony’s Triluminos Pro engine and support for Dolby Vision. To give you an idea of the differences in price, the 65-inch Bravia 6 at $2,000 costs the same as the LG’s B6, while the equivalent Sony Bravia 8 II is $2,800.
Read more: Best TVs of 2026
Being a Sony TV, the Bravia 6 features Google TV with Gemini for streaming and general TV stuff, as well as support for Dolby Atmos and DTS:X for immersive audio. Features you’ll miss out on with the Bravia 6 will likely include the 8’s antireflective coating and a higher brightness panel.
At CNET, we recommend OLED technology for its best-in-class mix of color, shadow detail and black levels. While I haven’t tested the LG B6 or the Sony Bravia 6, I expect them to perform well in these areas as well. The Sony Bravia 6 is available now.
Presented by EDB
As enterprises give AI agents more autonomy — the ability to plan, decide, and act across systems without a human approving each step — a hard question moves to the center of every architecture review: When an agent tries to complete an action that it was never authorized to do, what actually stops it?
These are your agents, running on your models, touching your data in your infrastructure — and the responsibility for what they do sits with you. That responsibility can’t be met in hindsight or with a set of abstract policies that live on paper but not in practice. Agents need rules in the context of the moment, because they don’t exercise overriding judgment of their own actions.
Consider a simple rule: Never open the car door. Followed literally, an agent could never get in or out of the car at all. But if you change the context (the car has just crashed, there’s a fire, someone is hurt and needs to get out), then the rule you actually want is the opposite. Context in the moment is everything. We are asking agents to do intelligent things; that requires intelligent rules.
The instinct is to add guardrails around the agent: instructions, policies, and monitoring layered above the model. Those mechanisms matter, but they share a structural limit: The car-door rule is plausible right up until the moment you actually have to decide whether to open the door. Controls at the agent layer are only as reliable as the agent’s output is predictable, and autonomy is precisely the property that makes that output hard to predict. Governance that depends on reviewing an action before it happens cannot keep pace with a system that acts in milliseconds, across many systems at once.
Governance has to become executable, and enforced where agents actually do their work: at the operational data layer, in the context, and exactly at the moment it is happening.
Agents create value by touching data. They query it, retrieve it, transform it, and increasingly act on it. A policy that says an agent should not reach a certain class of data is meaningful only if the system can deny that access at the moment the agent requests it. Additionally, a principle that says AI must be auditable is meaningful only if the organization can reconstruct what the agent did, what data it touched, which user it acted for, and what resulted. When governance lives at the data layer, it holds regardless of how the agent was built or how it behaves, because the control is a property of the database itself, not a promise made by the agent.
The enterprise should not rely on a model choosing to follow policy. The policy has to be enforced by the system. That is the difference between hoping an actor stays in bounds and constructing bounds it cannot cross to begin with.
The controls that make this real are ones many enterprises already run at the data layer: role- and attribute-based access, row- and column-level security, classification and masking, policy as code, and complete audit trails.
What agents change is not the mechanism, but who the mechanism has to recognize. Identity management has to treat the agent as a principal in its own right, with its own identity and a purpose declared when the session opens.
Once purpose is bound to identity, the policy engine can evaluate it the same way it evaluates role or department today, and the record of what happened can capture not just who acted and what they touched, but what they declared they were there to do.
In practice, this resolves into nine controls, grouped under three imperatives:
Role- and attribute-based access control enforced at query time, for agents as well as users
Dynamic column masking driven by the same policy path
Agent identity as a first-class principal, with declared purpose bound at session start and the acting user preserved
Classification and tagging that drives policy
Session-level audit logging that records which agent acted, for which user, and under what declared purpose
Lineage across pipelines, so a result can be traced back to the request that produced it
Centralized, portable policy management
Encryption at rest and in transit
Consistent enforcement across on-prem, cloud, and sovereign or air-gapped environments
“Declared purpose is what makes the difference. It becomes an attribute the access layer already understands, evaluated in the same policy path as role and row-level security. The enforcement mechanism does not change. What changes is that the agent’s purpose is part of what it evaluates, and part of what the record proves afterward,” says Priyanka Jain, VP, product management, data & AI governance, EDB.
Wherever you are in your AI adoption journey, enforcement at the data layer is what lets you move faster rather than slower. The controls are already in the database. The difference is that agents now have to pass through them.
The goal is not to stop agents from doing useful work. It is to define how far an agent can go, what it can touch, what it can change, what requires escalation, and how the organization can reconstruct events if something goes wrong. Governed this way, agents are identified, scoped, monitored, and auditable. The enterprise can adopt them faster, because security, risk, and leadership teams trust the operating model underneath.
Built on open source Postgres, this open foundation keeps enterprises in control of where their data lives, who can reach it, and under what policy, without ceding governance to a layer they don’t own or can’t inspect. For regulated industries, that combination of data sovereignty and source-level enforcement isn’t a nice-to-have; it’s the precondition for putting agents into production at all.
Agentic systems will keep getting more capable and more autonomous. That is a reason to be deliberate about where control lives, not a reason to slow down. The enterprises that enforce governance at the data layer can move aggressively on AI, because the thing protecting their data is more than just wishful thinking.
EDB Postgres AI is an open, enterprise-grade sovereign data and AI platform that unifies transactional, analytical, and AI workloads — with governance enforced where the data lives. For the full framework, see EDB’s white paper Governing Agentic AI at Enterprise Speed.
Max Romanenko is Chief Technology Officer at EDB.
Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. For more information, contact sales@venturebeat.com.
Weekend Open Thread: Madewell – Corporette.com
SpaceX stock could rise 75% to $240, JPMorgan says
Musk’s Tesla, SpaceX Confirm $16.8 Billion ‘Terafab’ Chip Plant as World’s Largest Building in Texas
anatomy of crypto’s biggest liquidation event since 2021
A $30 Billion AI Fund Implodes, Now the SEC Is Investigating Wall Street’s Role
6 months on, Irish renters crushed by effects of government housing bill
Did Trump Just Move SpaceX Stock With One Truth Social Post?
Mystery AI Model ‘Ox Alpha’ Draws Developers With Free Access as Chinese Lab Origins Remain Debated
The ‘Lucky Dip Gang’ causing carnage for clicks: After five thugs were killed speeding in the wrong direction on a motorway, GUY ADAMS investigates a sick new trend… and why police aren’t even allowed to pursue them
Moderna CEO warns China is pouring state money into mRNA technology
Nvidia Stock Suffers Longest Losing Streak Since 2022: Will Q2 Earnings End It?
NVIDIA Stock Drops Nearly 2 Percent to $210 on Seventh Losing Day Ahead of Critical AI Earnings
Unitree’s New Superman Robot Claims to Outjump and Outrun Every Human, Usain Bolt Included
UK firms in critical financial distress rise 9% to 53,756
Deshaun Watson fires back at Browns fans after being booed: ‘It’s a disrespectful thing’
Google Gemini AI Predicts Ethereum Could Become the Trade Retail Misses in 2026
Faster chip in a familiar form factor
SEC Regulation Crypto vs CLARITY Act: which framework wins
Will Tesla Stock Be Higher or Lower a Year From Now? Here’s What Wall Street Analysts Are Saying
Tesla Recalls Nearly 3 Million Cars in China Over Hidden Door Handles Linked to Multiple Deaths Amid New Ban
You must be logged in to post a comment Login