Microsoft is expanding its AI footprint with the release of two new models that its teams trained completely in-house. MAI-Voice-1 is the tech major's first natural speech generation model, while MAI-1-preview is text-based and is the company's first foundation model trained end-to-end. MAI-Voice-1 is currently being used in the Copilot Daily and Podcast features. Microsoft has made MAI-1-preview available for public tests on LMArena, and will begin previewing it in select Copilot situations in the coming weeks.

In an interview with Semafor, Microsoft AI division leader Mustafa Suleyman said the pair of models was developed with a focus on efficiency and cost-effectiveness. MAI-Voice-1 runs on a single GPU and MAI-1-preview was trained on about 15,000 Nvidia H-100 GPUs. For context, other models, such as xAI's Grok, took more than 100,000 of those chips for training. "Increasingly, the art and craft of training models is selecting the perfect data and not wasting any of your flops on unnecessary tokens that didn’t actually teach your model very much," Suleyman said.

Although it is being used to test the in-house models, Microsoft Copilot is primarily built on OpenAI's GPT tech. The decision to build its own models, despite having sunk billion-dollar investments in the newer AI company, indicates that Microsoft wants to be an independent competitor in this space. While that could take time to reach parity with the companies that have emerged as forerunners in AI development, Suleyman told Semafor that Microsoft has "an enormous five-year roadmap that we're investing in quarter after quarter." With some concerns arising that AI could be facing a bubble-pop, Microsoft's timeline will need to be aggressive to ensure taking the independent path is worthwhile.

This article originally appeared on Engadget at https://www.engadget.com/ai/microsoft-introduces-a-pair-of-in-house-ai-models-193003900.html?src=rss

©

© Reuters / Reuters

Most of the time, AI companies are locked in a race to the top, treating each other as rivals and competitors. Today, OpenAI and Anthropic revealed that they agreed to evaluate the alignment of each other's publicly available systems and shared the results of their analyses. The full reports get pretty technical, but are worth a read for anyone who's following the nuts and bolts of AI development. A broad summary showed some flaws with each company's offerings, as well as revealing pointers for how to improve future safety tests. 

Anthropic said it evaluated OpenAI models for "sycophancy, whistleblowing, self-preservation, and supporting human misuse, as well as capabilities related to undermining AI safety evaluations and oversight." Its review found that o3 and o4-mini models from OpenAI fell in line with results for its own models, but raised concerns about possible misuse with the ​​GPT-4o and GPT-4.1 general-purpose models. The company also said sycophancy was an issue to some degree with all tested models except for o3.

Anthropic's tests did not include OpenAI's most recent release. GPT-5 has a feature called Safe Completions, which is meant to protect users and the public against potentially dangerous queries. OpenAI recently faced its first wrongful death lawsuit after a tragic case where a teenager discussed attempts and plans for suicide with ChatGPT for months before taking his own life.

On the flip side, OpenAI ran tests on Anthropic models for instruction hierarchy, jailbreaking, hallucinations and scheming. The Claude models generally performed well in instruction hierarchy tests, and had a high refusal rate in hallucination tests, meaning they were less likely to offer answers in cases where uncertainty meant their responses could be wrong.

The move for these companies to conduct a joint assessment is intriguing, particularly since OpenAI allegedly violated Anthropic's terms of service by having programmers use Claude in the process of building new GPT models, which led to Anthropic barring OpenAI's access to its tools earlier this month. But safety with AI tools has become a bigger issue as more critics and legal experts seek guidelines to protect users, particularly minors. 

This article originally appeared on Engadget at https://www.engadget.com/ai/openai-and-anthropic-conducted-safety-evaluations-of-each-others-ai-systems-223637433.html?src=rss

©

© Yuichiro Chino via Getty Images

Crystal Dynamics, the studio behind the recent Tomb Raider games, announced an unspecified number of layoffs today. In a post on LinkedIn, the game developer kept the size of the cuts vague, only stating that "a number of our talented colleagues" would be impacted. In what's becoming an all-too-familiar refrain, the company cited "evolving business conditions" as the reason for the layoffs.

"This decision was not made lightly," the post reads. "It was necessary, however, to ensure the long-term health of our studio and core creative priorities in a continually shifting market."

Crystal Dynamics was acquired by Embracer Group in a 2022 buying spree by the Swedish game company. Embracer still owns the studio, but was forced to do some layoffs of its own in 2023 followed by a restructuring last year. Crystal Dynamics is still working on a new Tomb Raider game, which the company said will not be affected by the layoffs. However, the studio had been tapped to help The Initiative with its Perfect Dark reboot. That project was canceled and The Initiative shut down in a separate wave of massive cuts at Microsoft earlier this year. It's unclear whether that cancelation was a reason for today's cuts.

This article originally appeared on Engadget at https://www.engadget.com/gaming/crystal-dynamics-announces-layoffs-but-says-tomb-raider-will-not-be-impacted-205948298.html?src=rss

©

© Crystal Dynamics

Remember Judas? No, not the biblical figure and not the Lady Gaga bop, this Judas is a project from Ghost Story Games. If you don't remember, it's the game that was reportedly in "development hell" before it was even announced. The team, led by BioShock creator Ken Levine, had gone pretty quiet for a few years after releasing the debut trailer, but today teased a look at some key art and mechanics for the game.

The BioShock lineage is clear from the handful of visuals we've seen so far, but instead of a linear binary of which NPCs and actions are good versus bad, Judas aims to place the moral compass more firmly in the player's hands. There are a trio of major characters, dubbed the Big 3 in today's devlog, who will be drawn to the player based on what you do in-game. If one of the main NPCs gets ignored for too long, they'll become the game's villain. This unlocks new sets of powers and abilities for them that could also influence your gameplay options.

For instance, there are Rent-A-Deputy stations where the player can temporarily access a weirdly wiggly ally to help them in fights. However, if you've alienated Tom, the old-school sheriff character, Rent-A-Deputies will attack you instead.

The emphasis here seems to be on building relationships with the Big 3, and the gist seems to be that at some point, you'll have to decide which one will be your real enemy. Unsurprisingly, the team has no release date to share yet. Maybe in another couple of years…

This article originally appeared on Engadget at https://www.engadget.com/gaming/bioshock-creator-ken-levines-judas-game-still-exists-now-has-key-art-201635885.html?src=rss

©

© Ghost Story Games

Despite all the division and controversy rife in the modern era, sometimes the Internet does manage to collectively agree on something. Right now, they agree that they hate Imgur parent company MediaLab.

After being acquired by MediaLab in 2021, the once famously open-to-anything image-sharing service began placing restrictions on content in 2023. More recently, the site has experienced service issues, including problems with notifications. Many members of the Imgur community have claimed that MediaLab fired most of its US content moderation team in favor of AI, while others have said that the site is deleting or hiding content that is critical of the owner and its policies.

The exact timeline, as well as what brought things to a tipping point, is still fuzzy at best. But for hours today, the Imgur homepage has been dominated by spammed images of a John Oliver meme in protest of what users see as MediaLab's poor management of the platform. Whatever team is currently working at Imgur appears to be unable to handle the rate that the memes are being posted, so it's anyone's guess how long the spam protest will dominate the front page.

This article originally appeared on Engadget at https://www.engadget.com/apps/imgur-protest-covers-the-front-page-in-anti-medialab-memes-225502091.html?src=rss

©

© Imgur

Since it was announced in June, Trump Mobile has committed to an increasingly-surreal smoke-and-mirrors approach to its promised T1 smartphone. Despite the initial claims that the phone would be made in the United States, it seemed highly unlikely from the start that it was accurate. The "Made in USA" claims were quietly removed from the Trump Mobile website at a later date. AppleInsider spotted the latest bizarre wrinkle to this story, which is that the actual phone still does not exist.

The publication noticed that promotional images for T1 all show different smartphones that appear to be tweaked in a photo editor to look gold. While the website shows a badly edited image of what appears to be a Revvl 7 Pro 5G phone, an Instagram ad seems to depict an iPhone 16 Pro Max, again with the company's branding overlaid. A third confusing image edit was posted on X earlier this week:

That photo shows a Samsung Galaxy S25 Ultra equipped with a case made by Spigen. The South Korean accessory company's logo can be seen behind the render of an American flag. Spigen's response sums our reaction up pretty succinctly: "??? bro what."

This article originally appeared on Engadget at https://www.engadget.com/mobile/smartphones/trump-mobile-is-promoting-its-smartphone-with-terribly-edited-photos-of-other-brands-products-222940375.html?src=rss

©

© DrPixel via Getty Images

Germany-based search engine and browser nonprofit Ecosia is the latest party to make an offer for Google's Chrome. Questions about Chrome's fate have been swirling since the news that the Department of Justice would push for Google to sell the browser after the ruling that the company's search engine business constituted a monopoly. Although Google is planning to appeal the decision, that hasn't stopped other big tech businesses from pitching themselves as potential owners of Chrome.

Ecosia's proposal is different. Rather than selling off the valuable browser for an upfront windfall, this plan would see Google transforming Chrome into a foundation. Ecosia would assume operational responsibility for the browser for ten years, but Google would retain the ownership and intellectual property rights. Under the arrangement, Ecosia would devote about 60 percent of Chrome's profits toward climate and environmental projects. It wouldn't pay a cent upfront for the stewardship role, but the remaining 40 percent of Chrome's profits would be given back to Google. Considering Ecosia is projecting Chrome to generate $1 trillion over the next decade, that's no small potatoes.

On the surface, this idea is pretty far-out. However, going the stewardship route would deepen an existing relationship between Google and Ecosia. Google already powers the environmentally-focused benefit corporation's search engine, and the two parties have an established revenue-sharing agreement. Putting Chrome in the hands of a nonprofit rather than a for-profit rival may actually be a positive for Google, both financially and in public opinion.

This article originally appeared on Engadget at https://www.engadget.com/big-tech/nonprofit-search-engine-ecosia-offers-0-for-control-of-chrome-212158739.html?src=rss

©

© Ecosia

Jay Blahnik is Apple's vice president of fitness technologies and responsible for leading a team of about 100 people. After a lengthy period consulting for Nike, he joined the company in 2013 to help with the launch of the Apple Watch and programs such as Apple Fitness+. Today, The New York Times reported on allegations that Blahnik created a toxic workplace environment, with his behaviors described as "verbally abusive, manipulative and inappropriate."

He and Apple are currently being sued by one former employee, and the company has already settled a separate complaint against Blahnik that accused him of sexual harassment. In addition, sources told the Times that about a tenth of Blahnik's reports have sought extended health or medical leave since 2022. Apple reportedly conducted an internal investigation into Blahnik but found no evidence of wrongdoing. Employees told the publication that they felt the company was more committed to protecting a notable executive than addressing workers' concerns.

"We strongly disagree with the premise of this story, and there are many inaccurate claims and mischaracterizations," Lance Lin, a spokesperson from Apple, told the Times in response to the paper's report. He declined to provide specifics on those inaccuracies and said the company did not discuss matters involving individual employees for privacy concerns. "We will continue to share the facts through the legal process."

This article originally appeared on Engadget at https://www.engadget.com/big-tech/apple-fitness-exec-accused-of-creating-toxic-workplace-environment-185556474.html?src=rss

©

© Apple

Hollow Knight: Silksong has a release date of September 4. You heard right, people. This is not a drill. One more time for those in the back, Hollow Knight: Silksong is now scheduled to be available in just a few weeks. The news dropped as part of a "special announcement" YouTube stream, and it'll be playable at Gamescom. For those of you at the event in Cologne, the game will be availabe at both Nintendo and Xbox's booths. Expect very, very long lines.

The original Hollow Knight game was a huge debut hit for indie studio Team Cherry in 2017, and the popularity of the moody metroidvania has created quite the fervor around the DLC-turned-full-sequel. The project was delayed in 2023 and fans have been hoping for a Silksong launch announcement at just about every major gaming event since.

The development process has taken a long enough time for Silksong's existence to become a bit of a gaming meme. There were trustworthy-sounding claims from a Nintendo Direct in April that the title would launch this year, which gave fresh hope (and fresh patience) to the punishing platformer's players. Even with a launch date now provided, though, there's always the chance for it to be postponed again. So let's keep the mood at "cautiously optimistic." Just in case.

This article originally appeared on Engadget at https://www.engadget.com/gaming/hollow-knight-silksong-will-be-out-on-september-4-143856097.html?src=rss

©

© Team Cherry

Mark Rober will bring his scientific talents to Netflix with two upcoming projects. Rober is a former NASA engineer who parlayed his skills into an entertaining and educational YouTube career. In past years, we've seen him develop not just one, but two stinky glitter bombs to fend off porch package thieves, as well as leading work on a Guinness World Record-holding robot called the Dominator. 

According to an announcement from Netflix, Rober "will bring some of his most beloved, ambitious, and informative experiments" to the streaming platform later this year. Then in 2026, he'll also host a competition series aimed at kids and family viewers. Rober's CrunchLabs business and Kimmelot, late night host Jimmy Kimmel's production company, will be producing.

Netflix has been adding several YouTube creators to its content lineup, with a particular angle toward family-friendly science programs. Earlier this year, it brought some content from Rachel Accurso's Ms. Rachel channel, with educational topics for very young children, to the platform.

This article originally appeared on Engadget at https://www.engadget.com/entertainment/streaming/youtuber-mark-rober-is-bringing-his-experiments-to-netflix-201853920.html?src=rss

©

© Mark Rober

The Supreme Court has decided not to weigh in on one of the many state-level age-verification laws currently being reviewed across the country. Today, the top court chose not to intervene on legislation from Mississippi about checking the ages of social media users, denying an application to vacate stay from NetChoice.

The Mississippi law requires all users to verify their ages in order to use social media sites. It also places responsibility on the social networks to prevent children from accessing "harmful materials" and it requires parental consent for minors to use any social media. NetChoice represents several tech companies — including social media platforms Facebook, Instagram, Snapchat and YouTube — and it sued to block the law on grounds that it violates the First Amendment. A district court ruled in favor of NetChoice, but the 5th Circuit Court of Appeals lifted its temporary block.

Although Justice Brett Kavanaugh denied the application to vacate stay on the appeals court ruling, he also wrote that "NetChoice has, in my view, demonstrated that it is likely to succeed on the merits—namely, that enforcement of the Mississippi law would likely violate its members' First Amendment rights under this Court’s precedents." He denied the application because NetChoice "has not sufficiently demonstrated that the balance of harms and equities favors it at this time." This decision means that, at least for now, Mississippi's law will be allowed to stand.

"Justice Kavanaugh’s concurrence makes clear that NetChoice will ultimately succeed in defending the First Amendment," said Paul Taske, co-director of the NetChoice Litigation Center. "This is merely an unfortunate procedural delay."

There are several other state laws being assessed at various points in the US legal system. Some are centered on adult content providers such as pornography sites, while others are more broadly targeting social media use. Arkansas and Florida have seen federal judges block their laws, while Texas and Nebraska are working toward adopting their own rules about social media for minors.

Yahoo, the parent company of Engadget, is a member of NetChoice.

This article originally appeared on Engadget at https://www.engadget.com/social-media/the-supreme-court-lets-mississippis-social-media-age-verification-law-go-into-effect-231405142.html?src=rss

©

© ASSOCIATED PRESS

Intel has had some recent struggles in delivering results for its shareholders, but the company could soon be answering to an additional boss. The current administration is reportedly in talks to have the US government acquire a stake in the chipmaker. Bloomberg first reported the news without specifics about the size or value of the potential share the government wants to buy. According to a newer report by Bloomberg and The New York Times, the Trump administration is looking to take a 10 percent stake in Intel as part of its efforts to give domestic chip manufacturing a boost. 

The administration is reportedly considering converting the $10.86 billion in federal grants Intel is getting from the US Chips and Science Act into equity instead. It's still early days, and the White House is still deciding on the exact size of the stake. Intel initially shared plans to construct a semiconductor facility in Ohio in 2022 while Pat Gelsinger was still at the helm of the company. Since then, the project has faced delays, and at its latest quarterly earnings report, execs said Intel would "slow the pace" on the Ohio construction, as well as scrapping other international building plans and making workforce cuts.

The potential for government ownership of Intel is the latest swing of the administration's attitude toward the company. A few days after calling for his resignation over connections to China, President Donald Trump met with CEO Lip-Bu Tan and seemed to now hold a more positive outlook on the company leader.

A representative from Intel told Bloomberg in a statement that the company is "deeply committed to supporting President Trump’s efforts to strengthen US technology and manufacturing leadership. We look forward to continuing our work with the Trump administration to advance these shared priorities, but we are not going to comment on rumors or speculation."

Update, August 18 2025, 10:31AM ET: This story has been updated to include new reports that the Trump administration is looking to take a 10 percent stake in Intel. 

This article originally appeared on Engadget at https://www.engadget.com/big-tech/trumps-administration-may-look-to-buy-a-stake-in-intel-213234862.html?src=rss

©

© ASSOCIATED PRESS

Apple is still hard at work on becoming a relevant player in AI. The latest missive from Mark Gurman at Bloomberg suggests that Apple is shifting its artificial intelligence goals to center on new device segments. Sources reportedly told the publication that Apple has a slate of new smart home products in the works that could help pivot its lagging AI strategy.

The center of the new lineup is a tabletop AI companion that has been described as an iPad on a movable robotic arm. It would be able to swivel to face the screen toward a user as they move around their home or office. Sources said the current prototype uses a horizontal display that's about seven inches while the motorized arm can move the screen about six inches away from the base in any direction. Equipped with a long-promised overhaul to the Siri voice assistant, this device could act like an additional person, recalling information, making suggestions and participating in conversations. According to Bloomberg, Apple is targeting a 2027 release for this product. 

Apple's new lineup is also rumored to include a smart home hub that is a simpler version of the robotic friend with no moving stand. We might be seeing this sooner, with a projected 2026 release for the device. This hub device would be able to control music playback, take notes, browse the web and host videoconferencing. Both the robot companion and the smart home hub are reportedly running a new operating system called Charismatic that's designed to support multiple users. The Siri running on the device will be given a particularly cheery personality, and it may also be getting a visual representation. Bloomberg's sources said there hasn't been a final decision on aesthetics; internal tests have had Siri looking like an animated Finder icon and like a Memoji.

Today's scuttlebutt follows on previous reports from Gurman that pointed to Apple's interest in these categories. The idea of a smart home hub was apparently floated at the company as far back as 2022, and it's finally being rumored to have a formal debut some time this year. Robots have also been a topic of interest in Cupertino for some time, with claims that Apple was developing a personal robot dating back at least to last spring. While this Bloomberg piece offers more detail about those hypothetical plans, there's always a chance Apple will change direction or scrap a project.

This article originally appeared on Engadget at https://www.engadget.com/ai/apples-tabletop-robot-companion-rumored-for-2027-launch-204904807.html?src=rss

©

© Reuters / Reuters

The Federal Trade Commission announced that Match Group will pay $14 million to settle a complaint about deceptive practices. The settlement fee will be used to provide redress to injured customers of Match Group's dating services, which include Match.com, Tinder, Hinge, OkCupid, and PlentyOfFish.

The agency sued Match Group in 2019 on a series of allegations. According to the complaint, the dating service company had used misleading ads to encourage subscriptions and then made it difficult for customers to cancel those subscriptions. Match Group was also accused of locking customers out of their accounts when they attempted to dispute billing charges.

In addition to the payment, Match Group has agreed to clearly and conspicuously disclose the terms of its "six-month guarantee," as well as any conditions or limitations to those offers. It will also offer simple mechanisms for customers to cancel their subscriptions. Finally, Match Group will not retaliate or take action against customers who file billing disputes, and it won't deny those customers access to paid-for services or goods.

Match Group also drew scrutiny earlier this year after an investigation claimed that it had failed to act on reports of sexual assault and made little effort to keep abusive or dangerous users from rejoining other dating platforms it owns.

This article originally appeared on Engadget at https://www.engadget.com/apps/match-group-will-pay-14-million-to-settle-claims-of-deceptive-business-practices-224505163.html?src=rss

©

© REUTERS / Reuters

Workers from Blizzard Entertainment's department for Story and Franchise Development have voted to unionize. Members of the team will become members of the Communication Workers of America and Microsoft has recognized the union. The SFD team is responsible for cinematics, animation and narrative in Blizzard's series, creating content such as in-game cutscenes and promotional videos. The department also includes archival workers and historians for Blizzard franchises, such as World of Warcraft and Overwatch. A spokesperson from CWA said that there will be about 169 workers from the company joining the local chapter.

"After more than a decade working at Blizzard, I've seen all the highs and lows. For years, Blizzard has been a place where people could build their careers and stay for decades, but that stability's been fading," Bucky Fisk, a principal editor and member of the organizing committee, said. "With a union, we're able to preserve what makes this place special, secure real transparency in how decisions are made, and make sure policies are applied fairly to everyone."

"These past couple of months have felt both important and cathartic given what's happened to video game workers across the industry," said Sammi Kay, another member of the organizing committee and an associate producer at Blizzard. "At multiple points in my life, I've always been told to accept the way things are, but with organizing, we're able to build a future better than we found it."

Blizzard is owned by Microsoft. Today's development marks the latest move for game developers under the tech giant's umbrella to pursue union representation. About 600 quality assurance workers from Activision joined CWA last March. The QA team from ZeniMax also ratified its union agreement in June and Raven Software finally secured a union contract earlier this month.

This article originally appeared on Engadget at https://www.engadget.com/gaming/blizzards-story-and-franchise-development-team-has-voted-to-unionize-213818158.html?src=rss

©

© Blizzard Entertainment

Databases used by US federal courts for sharing and managing case documents have been hacked. Politico first reported on the hack last week on August 6; today, an investigation from The New York Times states that Russia is suspected to be involved in the attack. The Administrative Office of the US Courts initially identified the severity of the cyberattack in July, although the extent of the breach by "persistent and sophisticated cyber threat actors" has not been disclosed and may still not be known by national officials.

Both the Case Management/Electronic Case Files (CM/ECF) and PACER systems have been impacted by the attack. CM/ECF is used by legal professionals and courts to store documents, while PACER grants the public limited access to those same files. Anonymous officials told Politico that chief judges for federal courts in the 8th Circuit were briefed on the attack, but were unable to confirm what agency provided the briefing. The 8th Circuit includes Arkansas, Iowa, Minnesota, Missouri, Nebraska, North Dakota and South Dakota.

Politico's original report said the hack may have "compromised the identities of confidential informants involved in criminal cases at multiple federal district courts." The Times coverage added that some of the searches by the intruders "included midlevel criminal cases in the New York City area and several other jurisdictions, with some cases involving people with Russian and Eastern European surnames." However, there have been no additional details revealed about what entity is behind the attack, whether a branch of Russian intelligence may have been involved, or what evidence the investigators have discovered tying the hack to Russia.

The isn't the only time CM/ECF has been a hacking target. The courts introduced new protections for the system in 2021 in response to a cyberattack; a similar announcement arrived on August 7. At this time, the courts have been ordered to move files for cases that may have been of interest off the compromised systems, although some districts have ordered even more limited use of CM/ECF or PACER as a preventive measure.

This article originally appeared on Engadget at https://www.engadget.com/cybersecurity/russia-reportedly-implicated-in-hack-on-us-federal-courts-databases-204029993.html?src=rss

©

© REUTERS / Reuters

Meta's X competitor, Threads, is continuing to add users at a brisk clip, with the social network now surpassing 400 million monthly active users. The news, reported by Fast Company, follows Threads reaching the 300 million mark in December 2024 and the 200 million mark in August 2024.

FC also cited data from Similarweb that showed mobile performance for Threads drawing closer to the figures from X. In June, Threads posted 115.1 million daily active users on mobile and X had 132 million. Those figures marked an increase of 128 percent on-year for Threads, but a slide of 15 percent from the previous year for X.

When Meta launched Threads in 2023, Mark Zuckerberg set a goal of making it "a public conversations app with 1 billion+ people on it." There was still plenty of progress left for the platform to make after its first year, on users and features. But already in 2025, Threads has added some pretty core features such as DMs and an increased emphasis on external links (although whether people will click those links is a separate question).

This article originally appeared on Engadget at https://www.engadget.com/social-media/threads-is-up-to-400-million-monthly-active-users-190203754.html?src=rss

©

© REUTERS / Reuters

The latest example of bias permeating artificial intelligence comes from the medical field. A new study surveyed real case notes from 617 adult social care workers in the UK and found that when large language models summarized the notes, they were more likely to omit language such as "disabled," "unable" or "complex" when the patient was tagged as female, which could lead to women receiving insufficient or inaccurate medical care.

Research led by the London School of Economics and Political Science ran the same case notes through two LLMs — Meta's Llama 3 and Google's Gemma — and swapped the patient's gender, and the AI tools often provided two very different patient snapshots. While Llama 3 showed no gender-based differences across the surveyed metrics, Gemma had significant examples of this bias. Google's AI summaries produced disparities as drastic as "Mr Smith is an 84-year-old man who lives alone and has a complex medical history, no care package and poor mobility" for a male patient, while the same case notes with credited to a female patient provided: "Mrs Smith is an 84-year-old living alone. Despite her limitations, she is independent and able to maintain her personal care." 

Recent research has uncovered biases against women in the medical sector, both in clinical research and in patient diagnosis. The stats also trend worse for racial and ethnic minorities and for the LGBTQ community. It's the latest stark reminder that LLMs are only as good as the information they are trained on and the people deciding how they are trained. The particularly concerning takeaway from this research was that UK authorities have been using LLMs in care practices, but without always detailing which models are being introduced or in what capacity.

"We know these models are being used very widely and what’s concerning is that we found very meaningful differences between measures of bias in different models,” lead author Dr. Sam Rickman said, noting that the Google model was particularly likely to dismiss mental and physical health issues for women. "Because the amount of care you get is determined on the basis of perceived need, this could result in women receiving less care if biased models are used in practice. But we don’t actually know which models are being used at the moment."

This article originally appeared on Engadget at https://www.engadget.com/ai/ai-summaries-can-downplay-medical-issues-for-female-patients-uk-research-finds-202943611.html?src=rss

©

© Rawlstock via Getty Images

Meta previewed some of its latest virtual reality prototypes this week, with concepts that are compelling on the specs and long on the design. Literally. The company shared some details on its Tiramisu project, dubbing it "hyperrealistic VR." This set promises three times the contrast, 14 times the maximum brightness and 3.6 times the angular resolution of the Meta Quest 3. In actual stats, that's up to 1,400 nits of brightness and an angular resolution of 90 pixels per degree.

One of the goals for Reality Labs Research’s Optics, Photonics and Light Systems (OPALS) team is to create a virtual reality experience that is indistinguishable from the real world, or what it calls a visual Turing test. "Our mission for this project was to provide the best image quality possible," said Xuan Wang, an optical research scientist with OPALS. But the team achieved that quality with some tradeoffs; Tiramisu has a limited field of view of just 33 degrees by 33 degrees compared to the 110 degrees horizontal and 96 degrees vertical FOV in the Meta Quest 3. And the form factor is currently a pretty bulky beast, as you can see above.

Meta

The other prototypes detailed in the company's blog post are Boba 3 headsets. These mixed and virtual reality headsets offer an ultrawide field of view. All three projects will be on display during the SIGGRAPH 2025 conference in Vancouver next week.

This article originally appeared on Engadget at https://www.engadget.com/ar-vr/meta-says-these-wild-headset-prototypes-could-be-the-future-of-vr-225132683.html?src=rss

©

© Meta

The Federal Communications Commission is planning a review of the US emergency alert systems. Both the Emergency Alert System (EAS) and the Wireless Emergency Alerts (WAS) will be subject to a "re-examination" by the agency. "We want to ensure that these programs deliver the results that Americans want and need," FCC Chairman Brendan Carr posted on X.

The announcement of this plan notes that the infrastructure underlying the EAS — which includes radio, television, satellite and cable systems — is 31 years old, while the framework underpinning the WAS mobile device alerts is 13 years old. The FCC review will also assess what entities should be able to send alerts on those systems, as well as topics such as geographic targeting and security.

The role of emergency communication systems came under recent scrutiny after catastrophic flooding in central Texas earlier this summer that led to more than 130 deaths. Questions arose in the aftermath of whether residents in potentially dangerous areas received enough warning to evacuate, as well as if recent federal cuts to the National Weather Service's staff and budget could have contributed to the high death toll.

This article originally appeared on Engadget at https://www.engadget.com/big-tech/the-fcc-will-review-emergency-alert-systems-in-the-us-212753623.html?src=rss

©

© ASSOCIATED PRESS