Welcome to Video Games Weekly on Engadget. Expect a new story every Monday or Tuesday, broken into two parts. The first is a space for short essays and ramblings about video game trends and related topics from me, Jess Conditt, a reporter who's covered the industry for more than 13 years. The second contains the video game stories from the past week that you need to know about, including some headlines from outside of Engadget.

Please enjoy — and I'll see you next week.

---

Let’s all agree to stop talking about that awful conservative activist group out of Australia. You know the one — like a parasite, it attached itself to the censorship campaign that erased thousands of adult games from Steam and Itch.io, and successfully positioned itself at the center of the delisting narrative. However, logic and evidence suggests this group had very little to do with the mass removals.

This Australian anti-porn organization led a movement in April to remove the edgelord simulator No Mercy from Steam, and since everyone agreed that game sucked, the campaign worked and the title disappeared from the storefront. This is where I believe the organization’s involvement in the current drama ends.

It seems No Mercy spurred payment processors including Visa, Mastercard and PayPal to turn their attention to the PC gaming market (an irony that I would find funnier if it weren’t actively eroding an industry I love). These institutions took the opportunity to dictate the types of games they would support, and in response on July 16, Steam added a clause to its ruleset banning content that "may violate the rules and standards set forth by Steam’s payment processors," including “certain kinds of adult only content.” Censorship on Steam is not new; Valve has for years had rules banning mislabeled adult content, hate speech, anything violating local laws and many other regulations. But outsourcing censorship to payment processors is new, and hundreds of games were removed from the platform following the rule change. On July 24, Itch.io rolled out its own changes and summarily de-indexed every adult and NSFW game it hosted, which amounted to roughly 20,000 titles being hidden from search and browse pages.

The conservative Australian group claimed responsibility for the Steam bans on July 19, three days after the platform’s rule change went live. The organization said the censorship was the direct result of two of its recent efforts: an email campaign that sent 1,067 messages to Visa, Mastercard, PayPal and others claiming Steam and Itch.io were hosting illegal sexual content, and an open letter addressed to the same financial institutions, signed by faith-based, anti-sex work and anti-queer activist groups.

There is no evidence that these campaigns were directly responsible for payment processors’ renewed enforcements. It’s actually ludicrous to suggest that roughly 1,000 emails or an open letter would even register at companies the size of Visa, Mastercard or PayPal. What’s more, after taking credit for the removal of hundreds of Steam games, the Australian group has attempted to distance itself from the whole shebang. Following intense scrutiny from players, industry watchdogs and media outlets, every group in this situation is trying to avoid accountability, in fact. Valve says Mastercard made this happen, while Mastercard says it’s just following the law, and PayPal says it’s simply doing what companies like Mastercard tell it to do. Meanwhile, the Australian group is trying to avoid blame for the sweeping Itch.io delistings while simultaneously attempting to exploit the Steam bans and gain momentum for its conservative bullshit.

In the most likely scenario, the Australian activist group saw these PC gaming audits coming and, in a strange act of stolen puritanical valor, took steps to center itself in the conversation. We can stop helping it do so. Forget its name and, as the IGDA suggests, direct your ire toward the organizations with power in this situation, namely Mastercard and Visa.

Still. It’s notable that an organization backed by conservative Christian groups that loudly oppose sex work, queer rights and freedom of expression was able to so cleanly align itself with financial companies censoring content on Steam and Itch.io. This uncontested endorsement is especially worrisome in a political and social climate where women, the queer community, people of color and those who don’t conform to a traditional conservative lifestyle are under attack. At a time like this, subversive and raw art is more necessary than ever, but it’s also in its most vulnerable position. Choosing this moment to activate a censorship campaign is not only dangerous for our most vulnerable communities, it’s cowardly.

There’s been some additional misinformation wrapped up in this censorship mess, of course. Three games were incorrectly reported as delisted or removed from Itch.io or Steam as part of this situation: Mouthwashing, Trials of Innocence and Console Me. One game that was unjustly removed during the chaos was the psychological horror game VILE: Exhumed — read my interview with creator Cara Cadaver right here.

---

The news

BioShock 4 enters a new circle of development hell

In most contexts, I’d be pretty stoked on the thought of a hell-based BioShock, but this is the worst possible iteration of that idea. According to Bloomberg, BioShock 4 failed a recent review with executives at its publisher, 2K Games, and it’s heading back to developers at Cloud Chamber for a narrative revamp. Plus, Cloud Chamber studio head Kelley Gilmore is gone and creative director Hogarth de la Plante was moved to a publishing role. Not much is known about the game that’s assumed to be BioShock 4, but it was revealed in 2019 alongside the formation of Cloud Chamber, so it’s already been in development for quite a while.

In related Rapture news, 2K’s remake of the original BioShock was canceled earlier this year, Bloomberg reports. Ken Levine, the creator of the BioShock series, is currently working on a familiar-looking FPS called Judas at his own studio, Ghost Story Games.

GOG gave away millions of games to protest censorship

GOG partnered with developers to release 13 games with adult themes for free from August 1 to 3, in protest of all the censorship going down on Steam and Itch.io. None of the free titles were specifically banned in the censorship campaign, but they featured sexual, queer or violent content that could easily be targeted by similar efforts. GOG handed out its free games to more than 1 million players.

Itch.io is reindexing free NSFW games

After deindexing all of its adult games on July 24 — like, all of them — Itch.io on August 1 relisted all free games in this category. Itch.io is currently auditing thousands of adult and NSFW games that it swept up in the payment processor ban, and it’s unclear how the platform will support titles with these themes going forward. One of Itch.io’s longstanding partners, Stripe, said it will no longer facilitate transactions of titles "designed for sexual gratification," but there’s apparently room for negotiation in the future.

Battlefield 6 will land on October 10

In a shocking twist, EA also revealed that Battlefield 6 will have multiplayer content.

Age verification is coming to an Xbox near you

The video game world is feeling the effects of the UK's Online Safety Act. Platforms including Discord and Xbox are implementing new age-verification methods to comply with the law in the UK, and Microsoft is planning to expand its program to other regions. It's in no rush, though, saying it'll use the UK as a guinea pig for these systems first, and then implement what it learns across the globe.

Sony is suing Tencent over its blatant Horizon clone

At its unveiling in 2024, viewers instantly called out Polaris Quest’s Light of Motiram for looking an awful lot like Guerrilla Games' Horizon series — our headline called it a "pretty blatant Horizon ripoff," even. It took the better part of a year, but PlayStation's lawyers have finally kicked into gear. Sony is now suing Tencent, which owns Polaris Quest, over what it calls a "slavish clone" of its IP.

Raven finally has a union contract with Microsoft

Recent layoffs at Microsoft have only heightened the importance of proper labor organizing in video games, and one of the industry’s first unions is finally making things official. Three years after initiating the process, Call of Duty support studio Raven Software has ratified its union contract with Microsoft. Raven Software initially voted to organize under Activision Blizzard, but after Microsoft completed its acquisition of the studio in October 2023, it continued negotiations with the tech titan.

Play VILE: Exhumed, the game that Steam doesn't want you to see

After VILE: Exhumed was removed and permanently banned from Steam for reasons that don't actually apply to the game, developer Cara Cadaver and publisher DreadXP have rolled out their own distribution model, and prices start at $0. You can download and play the game right now for free, and there's an option to throw some money Cara's way. A portion of the profits will benefit the Toronto-based charity Red Door Family Shelter.

Additional reading

• Kris Holt’s indie game roundup

• Sam Rutherford’s review of the Legion Go S powered by SteamOS

---

Have a tip for Jessica? You can reach her by email, Bluesky or send a message to @jesscon.96 to chat confidentially on Signal.

This article originally appeared on Engadget at https://www.engadget.com/gaming/video-games-weekly-censorship-and-stolen-puritanical-valor-234220878.html?src=rss

©

© Cara Cadaver

After her game was banned from sale on Steam in a baffling decision with no appeal option, solo developer Cara Cadaver has made VILE: Exhumed available as shareware under a Creative Commons license. The project can be downloaded for free, but players can opt to donate in support of the solo developer's work. Both Cara and publisher DreadXP will pay forward those donations, with 50 percent of the game's profits being given to the Toronto-based charity Red Door Family Shelter. The group aids families, refugees and women who are escaping violence.

Both Steam and Itch.io have recently adopted sweeping and vague policies regarding their approach to projects with adult content due to pressure from payment processors. Itch.io has begun re-indexing some projects, but only free ones.

These changes have disproportionately impacted projects by underrepresented and queer creators, according to a statement from the International Game Developers Association that condemned the broad delisting of adult games. In her post announcing the new distribution plan for VILE: Exhumed, Cara summed up the situation pretty aptly: "What this actually results in is taking power and storytelling away from women, other marginalized artists, and ultimately, from everyone."

This article originally appeared on Engadget at https://www.engadget.com/gaming/banned-steam-game-vile-exhumed-is-back-as-a-free-shareware-title-225220847.html?src=rss

©

© Cara Cadaver

Sonos is having overheating issues with a "very low" number of its Roam speakers. Bloomberg reports that some users' portable Bluetooth speakers are overheating around the device's USB-C port. The company has not recalled any products as a result but it aware of the issue happening.

"We’ve closely tracked a limited number of reports involving the USB-C charging connection on some first-generation Sonos Roam speakers," Sonos said in a statement shared with Engadget. "While the overall incidence rate is very low, and environmental conditions appear to play a role, we’ve taken several proactive steps to even further reduce the likelihood of this issue, including software updates and accessory improvements."

The issue appears to be centered on the original model of the Sonos Roam that debuted in 2021; the Sonos Roam 2 offered some upgrades over that version when it rolled out last spring.

This article originally appeared on Engadget at https://www.engadget.com/audio/speakers/a-very-low-number-of-original-sonos-roam-speakers-are-overheating-205722779.html?src=rss

©

© Nathan Ingraham for Engadget

Florida Attorney General James Uthmeier has sued multiple pornography platforms on allegations that they fall afoul of age verification laws. The state passed HB 3 in March 2024 and the law took effect in January 2025. HB 3 placed new requirements on services to confirm the ages of their users if they contain "material harmful to minors" and to ensure nobody younger than 18 accesses their content.

The lawsuit today targets the companies behind several porn sites, including XVideos, XNXX, BangBros and Girls Gone Wild, as well as adult advertising network Traffic Factory. "We are taking legal action against these online pornographers who are willfully preying on the innocence of children for their financial gain," Uthmeier said.

Although today's lawsuit focuses on pornography providers, many of the provisions in HB 3 also center on teen use of social media. In June, a judge temporarily blocked the law after NetChoice and the Computer and Communications Industry Association — groups representing several social media platforms — sought a preliminary injunction. Uthmeier has appealed that injunction to the Eleventh Circuit.

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

This article originally appeared on Engadget at https://www.engadget.com/big-tech/florida-is-suing-several-porn-companies-over-age-verification-190251850.html?src=rss

©

© Joe Raedle via Getty Images

Intel is reportedly still struggling with a chipmaking process crucial to its future. Reuters reports that the company's 18A process is still producing low yields and high defect rates. Intel has invested billions of dollars in the manufacturing process, on which it's pinned its hopes of gaining ground on TSMC.

This isn't the first concerning news about 18A. Last year, a report stated that Broadcom was unhappy with the results of a test run for a potential order. However, Intel insisted at the time that 18A was on track to make its upcoming Panther Lake chips at volume later this year. "Our performance and yield trajectory gives us confidence this will be a successful launch that further strengthens Intel's position in the notebook market," Intel said last month.

Intel has typically aimed for a yield of at least 50 percent of usable chips before scaling up production. The company is said to make the bulk of its profit after reaching 70 to 80 percent. Last year, 18A's usable Panther Lake chips had reportedly only reached a five percent threshold. Intel had aimed for 10 percent by this summer. Tuesday's report doesn't state 18A's current yield, only describing it as a small percentage.

In a statement sent to Engadget, an Intel spokesperson said it is pleased with 18A's current state. "We feel very good about our trajectory on Intel 18A, and it will be the foundation of multiple generations of client and server products in the coming years," the spokesperson wrote. "Panther Lake is going to be a great product for Intel and our partners," the company continued, adding that its launch is still on track for later this year.

In a July interview with Reuters, Intel's Chief Financial Officer David Zinsner suggested that 18A's yields were better than reports claimed. He added that yields tend to "start off low and improve over time."

The company's 18A process is a risky bet, combining manufacturing changes with a next-gen transistor design. Intel embraced the challenge with an aggressive timeline that one of Reuters' sources called a hail mary. 18A going well would also help attract business for its upcoming 14A process. Last month, Intel warned investors that it may have to leave chip manufacturing altogether if it doesn't land 14A contracts.

The company needs all the help it can get. It recently confirmed that it would cut around 20 percent of its workforce by the end of this year. That follows 20,000 job cuts from June 2024 to July 2025. Earlier this year, it took on a new CEO to try to right the ship.

This article originally appeared on Engadget at https://www.engadget.com/big-tech/intels-next-gen-manufacturing-process-is-reportedly-still-struggling-184146350.html?src=rss

©

© Intel

AI startup Eleven Labs just announced a service called Eleven Music, which generates fake songs that are cleared for commercial use. It's a prompt-based affair, so it can create just about anything users dream up. 

The songs can feature vocals and lyrics. The Washington Post gave examples of prompts like "a smooth jazz song with a ‘60s vibe and powerful lyrics, but relaxing for a Friday afternoon." The service reportedly only takes a few minutes to generate music.

The company has been quietly testing the platform for some time, with WSJ indicating it has given 20 of its customers access to the model and that they've used it to make stuff for films, TV shows, video games and apps. ElevenLabs hasn't specified who the 20 customers are, likely because people get angry about AI slop.

As for training, the company has inked deals with two digital rights agencies for smaller music labels called Merlin Network and Kobalt Music Group. ElevenLabs co-founder and CEO Mati Staniszewski says he's aiming to get major labels on board. He also says that "the model is strictly created on data that we have access to." This is good news for the company, as other music-generation platforms like Suno and Udio have been sued for alleged use of copyrighted works.

AI-generated music is having something of a moment right now. The "band" Velvet Sundown is completely made up and managed to amass millions of listens on Spotify. It remains to be seen how much of that interest was based on actual fandom or morbid curiosity.

ElevenLabs is primarily known for its voice-generation technology. It created a news app that reads stories to consumers with AI-generated voices based on celebrities like Judy Garland and James Dean. One of its tools was used to emulate Joe Biden's voice in robocalls urging voters not to participate in a primary. It's also been used to create deepfakes for other celebrities.

This article originally appeared on Engadget at https://www.engadget.com/ai/elevenlabs-launches-its-own-royalty-free-ai-music-service-183033630.html?src=rss

©

© ElevenLabs

Motorola and Swarovski are teaming up to create the crossover they knew we always wanted: a crystal-laden edition of the Razr. For those who just felt that their retro cell phone wasn’t special enough, the fancy new Pantone Ice Melt Razr features a 3D quilted leather-inspired finish with 35 hand-positioned Swarovski crystals, including one larger, 26-facet crystal on the hinge.

The Pantone Ice Melt will cost a cool $1,000, though it will include a pair of the Moto Buds Loop. These open-ear buds are designed to be worn like jewelry — fitting, given the Swarovski treatment. Motorola is calling this bundle the Brilliant Collection. A stand-alone Motorola Razr will normally run you $700, and a pair of Moto Buds Loop will set you back $300, so the package deal for $1,000 basically means free crystals.

Motorola has been expanding the Razr lineup lately, and the company now offers the Razr, Razr Plus and Razr Ultra. The Pantone Ice Melt edition is a design spin on the base model and retains the same 3.6-inch cover display, 6.9-inch foldable AMOLED main display, MediaTek Dimensity 7400X processor, 4,500 mAh battery and 50-MP main camera.

Motorola said that this release is "just the beginning for Motorola Collections," so we expect to see more curated collaborations down the line. The Brilliant Collection will become available on August 7.

This article originally appeared on Engadget at https://www.engadget.com/mobile/smartphones/motorola-and-swarovski-team-up-for-a-super-fancy-razr-175308229.html?src=rss

©

© Motorola

A power bank with a wireless charging option is a very handy device for anyone who's typically on the go. It's probably worth having a fully charged one on hand in case of a power outage too. Anker makes some of our favorite power banks, and one with a Qi2 charging function and 10,000mAh capacity is currently on sale. You can snap it up for $60.

That's a 25 percent discount on the list price of $80. It is also the lowest price for this model so far in 2025. The power bank is currently available in four colors for this price: black, green, pink and white.

This particular model doesn't feature as a top pick in our roundup of the best power banks. Anker charging gear is solid for the most part, though the company did recently issue recalls for certain older models.

In any case, this MagGo Power Bank offers 15W wireless charging that can top up an iPhone 16 Pro's battery from zero to 30 percent of its capacity in 25 minutes, according to Anker. When you plug a USB-C cable into the power bank and an iPhone 16 Pro, the brand claims that you can charge the phone to 50 percent of its battery capacity in 26 minutes. 

It has enough juice to charge an iPhone 16 1.75 times over, an iPhone 16 Pro 1.71 times and an iPhone 16 Pro Max 1.33 times, per Anker. You can also fully charge the power bank's battery in about 2.5 hours.

Follow @EngadgetDeals on X for the latest tech deals and buying advice.

This article originally appeared on Engadget at https://www.engadget.com/deals/this-anker-10k-qi2-magnetic-power-bank-is-on-sale-for-60-right-now-171552456.html?src=rss

©

© Anker

Step aside, old boy. The ThinkPad X9-14 Aura Edition is making me forget all about the ThinkPad X1 Carbon, the most renowned business laptop for over a decade. Of course, Lenovo isn’t doing away with the X1 Carbon, but the X9 series certainly seems like the company’s experimental project for the same premium category. It’s not perfect, but I like where Lenovo is going with it.

Much love has been packed into the X9-14’s impressively ultraportable design, including a stunning 14-inch OLED screen, strong speakers, long battery life and solid performance for casual business users. The problem is that it costs a chunky $1,283 for low-end specs, it doesn’t come with a USB Type-A port and it’s failing the legacy of excellent ThinkPad keyboards.

That doesn’t stop the ThinkPad X9-14 Aura Edition from doing the best with what it’s got, and it’s got a lot.

Lookin’ sleek and ready for travel

I’ve always been a fan of bold, colorful laptops, but I understand that in a business setting you’d want to blend in. The ThinkPad X9-14, however, manages to do so in style — it won’t draw your coworkers in from across the room, but you’ll hear a good ol’ “ooo” when they get a closer look. The dark gray aluminum frame looks like it was carved out of a space rock and then blasted with sand (or oxide, both give that sparkly grainy look). Of course, there’s the iconic ThinkPad logo in the top left corner of the lid with a red LED, but there’s also a X9-14 logo at the very top, which sits on an oblong island giving machine-look vibes.

The interior sports more of the same minimalist design, with my favorite visual aesthetic being the keys. It looks like someone carved away the bottom of each one to add a bit of flair, and I’m all about it. And if you’re not a fan of bezels, they take up very little room on the display. There’s a 4K webcam on the lip, which you can disable with the F9 button when you’re out of meetings. I’d do it in meetings too, because of its poor video quality — my features were fuzzy and my face was red.

The ThinkPad X9-14’s chassis maintains a remarkably thin 0.26-inch profile up until you get to the ports, where it expands to a neat 0.68-inch block that also holds the fans. There’s only four ports, but most of the essentials are covered, including two Thunderbolt 4 ports, an HDMI slot and a headphone jack. The critical miss is the lack of a USB Type-A port, which the 15-inch model supports.

Despite that drawback, the X9-14’s thin frame combined with its 2.8-pound chassis make for an exceptionally ultraportable business laptop that won’t be a pain to pull out at a TSA checkpoint.

Pretty on the eyes, music to my ears

Nothing will make me gush more than a pretty OLED screen, and the ThinkPad X9-14 had me blushing like a pink Starburst. In my experience, OLED panels are less straining on the eyes (although this topic is somewhat of a debate), but I prefer them when I’m working all day. Even when you’re all done with your business, you won’t be able to pull away from this 14-inch screen with its 2,880 x 1,800 resolution. That’s sharp enough to capture each strand of hair on your favorite furry Disney sidekick. And whether you’re scrolling through endless rows of data on your spreadsheets or making your way through your indie-game library, the X9’s 120Hz refresh rate will enable smoother visuals than the average business laptop. The one thing I hate, though, is the rounded edges because it cuts off some of the visuals.

I watched an episode of Steven Universe, which is iconic for its pastel color palette, and my eyes practically swam in the crystal blue ocean of Beach City. The dialogue sounded a bit quiet, but tinkering with the Dolby Atmos settings gave it the boost it needed.

I couldn’t stop myself from jumping into a match of Marvel Rivals to keep my weekly challenge streak running, and the ThinkPad’s OLED allowed me to soak in all of the game’s vivid colors. I tried out Phoenix for the first time and blasted enemies with her cosmic flames, creating a visual spectacle of orange and yellow. The cacophony of explosions were distinct and well-balanced on the audio side as well, I could hear that jabroni (Star Lord) screaming “Legendary!” on my backline. The dual-speaker setup produced well-rounded audio for a business laptop. Tracks like “Golden” from KPop Demon Hunters (my new obsession) popped with crisp vocals and bassy synths, so feel free to take off your sweaty headphones and vibe at the end of the workday.

Rami Tabari for Engadget

Mushy click-clack and haptic feedback

Like Rumi’s relationship with Jinu (I told you I’m obsessed), my relationship with this keyboard is complicated. I’m clicking away at this review, and the ThinkPad X9-14 is comfortable to type on, but it’s not as clicky as I want it to be. There’s some decent travel, but the keys are just… mushy. Sure, I can type on it all day as the buttons are very well-spaced, but they’re not satisfying to use. I imagine for most folks that might be fine, but after testing hundreds of laptops, I can say for certain there are better keyboards out there.

On the other hand, the touchpad was so smooth that my fingers slid across its glassy surface like a figure skater at the Olympics. Its haptic feedback offers enough response to mimic the real thing, and I’ve actually grown to like haptics more than real touchpads, which are terribly inconsistent.

In addition to the trackpad and keyboard, you can also use the ThinkPad X9's touchscreen to get work done. I experienced some resistance when rearranging some of my browser windows, so it’s not great if you’re just using your fingers. But I’m sure folks that need to sketch will appreciate it, especially since the OLED display will do justice to your art. (For some reason, Lenovo’s website doesn’t say that its pens are compatible with the ThinkPad X9-14, but a Lenovo representative confirmed they are, and the laptop itself also features the Lenovo Pen Settings app.)

The ThinkPad X9's performance: Business casual

I’ve enjoyed my time with the ThinkPad X9-14 so far, and that’s with its bottom-of-the-barrel CPU, the Intel Core Ultra 5 226V, plus 16GB of RAM and a 512GB SSD. I expected some lackluster performance, maybe even a little slowdown, but it actually felt pretty smooth.

I used the ThinkPad X9-14 to do a little work on the fantasy novella I’m writing, which, if you know anything about the genre, requires me to have an obscene number of tabs open. I split my screen in three, between Campfire, Google Docs and Legend Keeper, my unholy trinity of world building. The ThinkPad kept up with my shenanigans, so it’ll do well with most casual office workflows.

If you’re really crunching numbers, though, the Intel Core Ultra 5 226V is the lowest performing CPU you can get in the X9-14, which isn’t necessarily a bad thing, until you see the price (I’ll go into that below). Its benchmark scores aren’t far off from the Zenbook A14, but ASUS’ chosen Snapdragon X X1 CPU got marked for its slow performance. That’s likely because the Zenbook needs to emulate Windows apps due to its ARM architecture. While the Zenbook A14’s scores are a smidge better than the ThinkPad X9-14, at least with the latter you don’t have to worry about the latency of emulating apps or running into compatibility issues.

Since the MacBook Air has been getting cheaper, I’d be remiss not to consider it as a competitor to the ThinkPad X9, either. If you’re shopping for any flavor of ThinkPad as a business laptop, you’re likely looking for security features, or remote management tools (if you’re in IT), and MacBooks offer both. The MacBook Air M4 13-inch is roughly 50 percent more powerful than the ThinkPad and the Zenbook. Again, the ThinkPad X9-14 still offers decent performance, just don’t expect to do anything taxing like video-editing.

Now, if you’re thinking about cutting out some work and doing more play, the X9-14’s Intel Arc Graphics iGPU is surprisingly capable. You’re not going to be playing AAA games on max settings, but I hopped into a few matches of Marvel Rivals and managed a decent 40 to 55 frames per second. I maintained the native resolution but cranked everything down to Low with Ultra Performance and Frame Generation enabled.

The ThinkPad X9-14 crushed the Zenbook A14 in terms of graphics, and while the MacBook Air delivered stronger performance overall for video-editing and designing, it can’t compare to the gaming compatibility that the X9-14 offers.

And like most modern laptops, the ThinkPad X9-14 has an NPU. If your job requires you to work with natively-run AI features, the ThinkPad X9-14 will aggregate those spreadsheets for you (and create cursed memes of, say, your boss’ face on a dog’s body).

Rami Tabari for Engadget

A battery for business hours and beyond

I spent quite a bit of time with the ThinkPad X9-14 before it eventually ran out of steam. It survived me working on this review, jumping to my book at the end of the day and a late night Discord chat with friends. On our video rundown battery test, it lasted 11 hours and 51 minutes. That’s more than enough to carry on your business and even add some overtime (or, you know, kick back and relax to some chill videos).

The competition is stiff, though. Last year’s ASUS Zenbook S14, which has an Intel chipset, clocked over 4 hours longer than the ThinkPad X9-14. Meanwhile, Snapdragon X laptops blow all those Intel machines out of the water, with systems like the ASUS Zenbook A14 coming in at 18:16. Still, while Lenovo's notebook might fall short of those impressive runtimes, at least it doesn't have the same app limitations that hold those devices back. It’ll get the job done, and that’s what matters.

Is it worth the business tax?

So how much does this all cost? Well, the model I reviewed comes with an Intel Core Ultra 5 226V, 16GB of RAM (though only the 32GB configuration is currently available) and a 512GB SSD. That totals out to $1,283 on Lenovo's website (with 32GB of RAM). For context, I’ve seen more powerful gaming laptops that cost less. But you’re not buying a business laptop for its power, you (or your IT manager) are buying it for its features.

While many non-business laptops are designed to be durable, you are still getting a military-grade chassis, which has been tested against drops, shocks and extreme temperatures. Then there’s Lenovo’s ThinkShield Security suite. That’s just a fancy term for saying the ThinkPad X9-14 includes a standard dTPM security chip, a fingerprint reader, a IR webcam for Windows Hello, an e-shutter for the webcam (F9) and Intel’s Threat Detection Technology (TDT) that uses AI to uncover cyberattacks. Intel TDT is included with Intel vPro, which can be used for remote management.

Now, if you want that laundry list of business features and top-tier performance, it’s going to hurt. To upgrade to an Intel Core Ultra 7 268V with a 1TB SSD, it’ll cost $1,939. Yep, that’s quite the leap for specs we’ve seen in significantly cheaper laptops (like the Dell 16 and 14 Plus).

It’s important to note, however, that business laptops like the ThinkPad X9-14 aren’t really intended for individual consumers but rather businesses looking to buy in bulk. There are business laptops priced at a level that might appeal more to the average consumer, like the $1,399 ASUS ExpertBook P5 (P5405), but those are more rare than I’d like. For that price, the ExpertBook comes with a Core Ultra 7, 32GB of RAM and a 1TB SSD.

Wrap-up

I thought I’d be more disappointed with the ThinkPad X9-14 considering its low-end CPU, but it’s perfectly suited for everyday tasks and made for a great writing companion. What really sold me was the OLED screen and audio, which are great not just for watching training videos or slideshows but also (and equally important) for after work when you get some play in, whether that’s literally playing games or binging your favorite new series.

I also appreciate the battery life and portability, plus all of Lenovo’s security features packed into a military-grade chassis.

However, I can’t justify the average consumer paying $1,283 for the ThinkPad X9-14 with specs you’d find in laptops that cost hundreds of dollars less (minus the RAM, of course). For small business owners, I recommend waiting for a sale or consider a business laptop marketed more toward you (again, the ASUS ExpertBook P5 (P5405) is a great choice). But if you’re the IT person at a company looking for a laptop to buy in bulk, then yes, the ThinkPad X9-14 is a worthy purchase.

This article originally appeared on Engadget at https://www.engadget.com/computing/laptops/lenovo-thinkpad-x9-14-aura-edition-review-a-solid-business-laptop-with-some-quirks-171551190.html?src=rss

©

© Rami Tabari for Engadget

Elon Musk’s chatbot is seldom out of the news. When it isn’t vomiting antisemetic rhetoric or fixating on conspiracy theories without invitation, it’s inviting you to romance its anime-inspired AI companions. Grok’s NSFW credentials now extend to xAI’s new image and video generator. Grok Imagine is exclusive to paying SuperGrok and Premium+ X subscribers and, as reported by TechCrunch, has a baked-in text-to-image and video feature with a number of modes that dictate its results. 

As you might expect, submitting prompts to Grok Imagine while in the embarrassingly named "Spicy" mode results in sexualized content in the form of images or short clips. TechCrunch was able to test the feature and said that while some of its requests came back blurred out or moderated, it was able to generate "semi-nude imagery" without resistance from the ever-obliging bot. Images reportedly only take a few seconds to produce and more are auto-generated as you scroll through the results.

Grok Imagine can also generate content featuring celebrities, but there appears to be additional restrictions in place for this. TechCrunch was unable to prompt it to produce an image of a pregnant Donald Trump, for example, with the chatbot instead generating an image of Trump standing next to a pregnant woman, or holding a baby.

Elon Musk is unsurprisingly very enthusiastic about his company’s latest toy, claiming on X that the number of images generated using Grok Imagine grew from 14 million to 20 million in the space of a day. He has also talked up its potential as a meme motherload and promised its existing capabilities will "improve radically" over time.

This article originally appeared on Engadget at https://www.engadget.com/ai/surprising-no-one-groks-image-and-video-generator-now-has-an-nsfw-spicy-mode-171057865.html?src=rss

©

© NurPhoto via Getty Images

For the first time since GPT-2 in 2019, OpenAI is releasing new open-weight large language models. It's a major milestone for a company that has increasingly been accused of forgoing its original stated mission of "ensuring artificial general intelligence benefits all of humanity." Now, following multiple delays for additional safety testing and refinement, gpt-oss-120b and gpt-oss-20b are available to download from Hugging Face.

Before going any further, it's worth taking a moment to clarify what exactly OpenAI is doing here. The company is not releasing new open-source models that include the underlying code and data the company used to train them. Instead, it's sharing the weights — that is, the numerical values the models learned to assign to inputs during their training — that inform the new systems. According to Benjamin C. Lee, professor of engineering and computer science at the University of Pennsylvania, open-weight and open-source models serve two very different purposes.

"An open-weight model provides the values that were learned during the training of a large language model, and those essentially allow you to use the model and build on top of it. You could use the model out of the box, or you could redefine or fine-tune it for a particular application, adjusting the weights as you like," he said. If commercial models are an absolute black box and an open-source system allows for complete customization and modification, open-weight AIs are somewhere in the middle.

OpenAI has not released open-source models, likely since a rival could use the training data and code to reverse engineer its tech. "An open-source model is more than just the weights. It would also potentially include the code used to run the training process," Lee said. And practically speaking, the average person wouldn't get much use out of an open-source model unless they had a farm of high-end NVIDIA GPUs running up their electricity bill. (They would be useful for researchers looking to learn more about the data the company used to train its models though, and there are a handful of open-source models out there like Mistral NeMo and Mistral Small 3.)

With that out of the way, the primary difference between gpt-oss-120b and gpt-oss-20b is how many parameters each one offers. If you're not familiar with the term, parameters are the settings a large language model can tweak to provide you with an answer. The naming is slightly confusing here, but gpt-oss-120b is a 117 billion parameter model, while its smaller sibling is a 21-billion one.

In practice, that means gpt-oss-120b requires more powerful hardware to run, with OpenAI recommending a single 80GB GPU for efficient use. The good news is the company says any modern computer with 16GB of RAM can run gpt-oss-20b. As a result, you could use the smaller model to do something like vibe code on your own computer without a connection to the internet. What's more, OpenAI is making the models available through the Apache 2.0 license, giving people a great deal of flexibility to modify the systems to their needs.

Despite this not being a new commercial release, OpenAI says the new models are in many ways comparable to its proprietary systems. The one limitation of the oss models is that they don't offer multi-modal input, meaning they can't process images, video and voice. For those capabilities, you'll still need to turn to the cloud and OpenAI's commercial models, something both new open-weight systems can be configured to do. Beyond that, however, they offer many of the same capabilities, including chain-of-thought reasoning and tool use. That means the models can tackle more complex problems by breaking them into smaller steps, and if they need additional assistance, they know how to use the web and coding languages like Python.

Additionally, OpenAI trained the models using techniques the company previously employed in the development of o3 and its other recent frontier systems. In competition-level coding gpt-oss-120b earned a score that is only a shade worse than o3, OpenAI's current state-of-the-art reasoning model, while gpt-oss-20b landed in between o3-mini and o4-mini. Of course, we'll have to wait for more real-world testing to see how the two new models compare to OpenAI's commercial offerings and those of its rivals.

The release of gpt-oss-120b and gpt-oss-20b and OpenAI's apparent willingness to double down on open-weight models comes after Mark Zuckerberg signaled Meta would release fewer such systems to the public. Open-sourcing was previously central to Zuckerberg's messaging about his company's AI efforts, with the CEO once remarking about closed-source systems "fuck that." At least among the sect of tech enthusiasts willing to tinker with LLMs, the timing, accidental or not, is somewhat embarrassing for Meta.

"One could argue that open-weight models democratize access to the largest, most capable models to people who don't have these massive, hyperscale data centers with lots of GPUs," said Professor Lee. "It allows people to use the outputs or products of a months-long training process on a massive data center without having to invest in that infrastructure on their own. From the perspective of someone who just wants a really capable model to begin with, and then wants to build for some application. I think open-weight models can be really useful."

OpenAI is already working with a few different organizations to deploy their own versions of these models, including AI Sweden, the country's national center for applied AI. In a press briefing OpenAI held before today's announcement, the team that worked on gpt-oss-120b and gpt-oss-20b said they view the two models as an experiment; the more people use them, the more likely OpenAI is to release additional open-weight models in the future.

This article originally appeared on Engadget at https://www.engadget.com/ai/openais-first-new-open-weight-llms-in-six-years-are-here-170019087.html?src=rss

©

© Igor Bonifacic / Engadget

Google has an event lined up for August 20 and has promised to reveal "new Pixel phones, watches, buds" and more. However, rumors have been swirling that the company won't have many of these products ready to go after being unveiled.

Winfuture has reported that the Pixel Watch 4, Fold Pro 10 and Buds 2a will be announced at the Made by Google event, but won't be available until October. The Pixel 10 line of smartphones will reportedly be available on August 28, which is just around a week after the announcement.

The holdup for the other products is reportedly due to supply chain issues, according to sources who spoke to Winfuture. Google hasn't confirmed this delay. As a matter of fact, the company hasn't even officially confirmed that the products in question exist. That's what the Made by Google event is for.

There have been plenty of leaks, so we likely know what's coming and some specs. The splashy Pixel 10 Pro Fold reportedly features a new display and hinge, which could account for the supply chain delay. The Pixel Watch 4 reportedly includes new internal hardware and a modified charging system which both could be contributing to the delay.

We'll just have to wait for August 20. This isn't a streaming event, so follow along with our liveblog of the entire proceedings.

This article originally appeared on Engadget at https://www.engadget.com/big-tech/googles-pixel-watch-4-fold-pro-10-and-buds-2a-are-rumored-to-launch-later-than-the-rest-of-its-new-gear-165020931.html?src=rss

©

© Sam Rutherford for Engadget

Fox Corporation has announced the launch of Fox One, a streaming service that will carry Fox-branded news, sports and entertainment content for $19.99 per month or $199.99 per year. The service will feature both live and on-demand access from the entire slate of Fox properties including Fox News, Fox Business, Fox Sports, Fox Weather and more. Subscribers will also be able to bundle the new offering with Fox Nation, the media corporation's on-demand streaming service with original shows.

“In bringing together the full power of the FOX content portfolio in one service, we have created a great value proposition and user experience that will appeal to the cord-cutter and cord-never fans currently not served by conventional pay TV packages," said Pete Distad, CEO of direct-to-consumer at Fox Corporation.

Fox One joins a growing list of TV streaming services aimed at cord-cutters that bring channel portfolios traditionally found on cable to a convenient app format. The new service also means access to NFL and MLB games aired on Fox networks without the need for a cable subscription. Fox was previously attached to the launch of Venu Sports, a sports-focused streaming service that never came to fruition. Fox One will launch on August 2.

This article originally appeared on Engadget at https://www.engadget.com/entertainment/streaming/fox-one-streaming-service-launches-august-21-164512558.html?src=rss

©

© Fox

Google's NotebookLM (NLM) is designed to be the ultimate study guide. So, with the new school year already beginning, it's fitting that the AI tool is now available for younger users. Just be sure to check its work, kids.

For consumers, anyone 13 and older can now use the AI-powered learning tool. However, any minimum age restrictions in your country will override that. NLM is also now available as a core service for all ages as part of the Google Workspace for Education suite.

The Gemini-powered NotebookLM lets you upload documents and take an AI-fueled crash course on them. The tool can train on text files, PDFs, websites or Google Docs / Slides. (You can combine sources, too.) Within a few seconds of uploading, you'll see a Notion-style digital notebook on the topic.

You can then ask questions about it, view summaries and generate mind maps. It can even spit out video explainers and podcast-style audio summaries.

Like any generative AI, NLM can make mistakes. But the nice thing is it's super easy to check its work: Each output includes citations that link back to the source material.

Fortunately, Google says it doesn't train on your chats or sources you upload, and no humans review it. The company recently added demo notebooks so you can take NLM for a spin without uploading anything.

This article originally appeared on Engadget at https://www.engadget.com/ai/googles-notebooklm-is-now-available-for-younger-users-163008926.html?src=rss

©

© Google

WhatsApp has shared a new update about how it prevents scams, as well as unveiling a new feature aimed at protecting users from possible swindles. People will now receive alerts when they are added to a new WhatsApp group by someone not in their contacts.

This safety overview will include key details about the group chat, such as the number of members, whether any other members are contacts and the chat's start date. It also offers some common sense reminders of how to avoid scams. Users can exit the group from that alert without ever looking at the chat if they choose, or can look at the chat to double-check whether it is a group they wish to participate in.

WhatsApp said it will "continue to test new approaches" for delivering similar alerts on individual direct messages within its service. These tools seem to still be in development, but the company indicated it is working on ways to catch scammers who initiate contact on a different platform before moving a conversation to WhatsApp.

In addition to the in-app tools, WhatsApp said that it has also identified and blocked many accounts used to perpetrate scams. During the first half of 2025, the company said it detected and banned more than 6.8 million accounts linked to scam centers. 

The Federal Trade Commission has published multiple reports over the years about the prevalence of scams on social media platforms.

This article originally appeared on Engadget at https://www.engadget.com/social-media/whatsapp-adds-new-warning-about-potential-group-chat-messaging-scams-160013367.html?src=rss

©

© WhatsApp

Roku just revealed a new ad-free streaming service called Howdy. The service costs $3 per month, making it an enticing prospect for budget-conscious viewers. The company says this isn't an introductory price and should be permanent.

Howdy will stream a whole bunch of content, but it's mostly pre-existing movies and shows. The company has inked deals with Warner Bros. Discovery, Lionsgate and other providers, so the platform will host movies like Elvis and Mad Max: Fury Road and shows like Weeds and Party Down. It'll also be home to Roku Originals, like the stellar Weird Al "biopic."

Company founder and CEO Anthony Wood spoke to Deadline and called Howdy "a response to the reality that many consumers are interested in a service that is ad-free and low-cost." He may be right. Consumers are likely to flock to something that's $3 a month, especially when the other streamers keep raising prices and stuffing ads everywhere.

This isn't Roku's only recent foray into the world of paid streaming, after years of relying on free, ad-supported television (FAST) channels. The company acquired the service Frdnly, which streams live TV from a number of noteworthy cable channels. Prices for this platform range from $7 to $10 per month.

Howdy will be available later today on Roku devices. The company promises a "rollout on mobile and additional platforms in the near future."

This article originally appeared on Engadget at https://www.engadget.com/entertainment/tv-movies/rokus-new-howdy-service-offers-ad-free-streaming-for-3-per-month-151550247.html?src=rss

©

© Roku

Illinois Governor JB Pritzker has signed a bill into law banning AI therapy in the state. This makes Illinois the first state to regulate the use of AI in mental health services. The law highlights that only licensed professionals are allowed to offer counseling services in the state and forbids AI chatbots or tools from acting as a stand-alone therapist.

HB 1806, titled the Wellness and Oversight for Psychological Resources Act, also specifies that licensed therapists cannot use AI to make “therapeutic decisions” or perform any “therapeutic communication.” It also places constraints on how mental health professionals may use AI in their work, such as specifying that its use for “supplementary support,” such as managing appointments, billing or other administrative work, is allowed.

In a statement to Mashable, Illinois State Representative Bob Morgan said, “We have already heard the horror stories when artificial intelligence pretends to be a licensed therapist. Individuals in crisis unknowingly turned to AI for help and were pushed toward dangerous, even lethal, behaviors.” The law enshrines steep penalties in an effort to curb such outcomes, with companies or individuals facing $10,000 in fines per violation.

“This legislation stands as our commitment to safeguarding the well-being of our residents by ensuring that mental health services are delivered by trained experts who prioritize patient care above all else,” said Mario Treto Jr., secretary of the Illinois Department of Financial and Professional Regulation.

The bill passed the Illinois House and Senate unanimously in a sign of overwhelming bipartisan support. The legislation is particularly notable as the Trump administration’s recently-revealed AI plan outlines a 10-year moratorium on any state-level AI regulation. It also comes as OpenAI has said it is improving the ability for its models to detect mental or emotional distress and will ask users to take a break during unusually long chats.

This article originally appeared on Engadget at https://www.engadget.com/ai/illinois-is-the-first-state-to-ban-ai-therapists-145755797.html?src=rss

©

© REUTERS / Reuters

Microsoft has revealed the first batch of Xbox Game Pass additions for August, and there are several tasty options in store over the next couple of weeks. The first one lands today on Game Pass Ultimate, PC Game Pass and Game Pass Standard in the form of Rain World, a difficult survival platform game. In case that’s not your tempo, though, you won’t have to wait long for other things to play.

On August 7, Assassin’s Creed Mirage will hit Game Pass Ultimate and PC Game Pass. I enjoyed this tighter, more focused take on the Assassin’s Creed formula quite a bit, especially considering how bloated these games have become in recent years (that said, AC Shadows is one of my favorite games of this year so far). Mirage is worth checking out if you dig the series but haven’t taken a leap of faith with that installment yet.

Several titles are joining the Game Pass Standard tier in the first two weeks of August, including Citizen Sleeper 2 — an acclaimed RPG that debuted earlier this year — on Wednesday. Survival shooter Aliens: Fireteam Elite returns to Game Pass on August 12 as well, while roguelike kingdom builder 9 Kings arrives in game preview on Game Pass Ultimate and PC Game Pass, but only for PC for the time being.

Here's the full breakdown of the first wave of August Game Pass additions, including the tiers on which they'll be available:

Today

• Rain World (Cloud, Console and PC) — Game Pass Ultimate, PC Game Pass, Game Pass Standard

August 6

• Citizen Sleeper 2: Starward Vector (Xbox Series X/S) — Game Pass Standard, already on Game Pass Ultimate and PC Game Pass

• Lonely Mountains: Snow Riders (Xbox Series X/S) — Game Pass Standard, already on Game Pass Ultimate and PC Game Pass

• MechWarrior 5: Clans (Xbox Series X/S) — Game Pass Standard, already on Game Pass Ultimate and PC Game Pass

• Orcs Must Die! Deathtrap (Xbox Series X/S) — Game Pass Standard, already on Game Pass Ultimate and PC Game Pass

August 7

• Assassin’s Creed Mirage (Cloud, Console and PC) — Game Pass Ultimate, PC Game Pass

• EA Sports Madden NFL 26: 10-hour early access trial (Cloud, Console and PC) — Game Pass Ultimate

August 12

• Aliens: Fireteam Elite (Cloud, Console and PC) — Game Pass Ultimate, PC Game Pass, Game Pass Standard

August 14

• 9 Kings (Game Preview) (PC) — Game Pass Ultimate, PC Game Pass

Along with the new additions, there are three titles leaving Game Pass on August 15. Those are Anthem (Console and PC) via EA Play, Farming Simulator 22 (Cloud, Console and PC) and Persona 3 Reload (Cloud, Console and PC). The beleaguered Anthem is shutting down for good this January in any case, and as it's an online-only game, there won't be any way to play it after that time — unless fans find a way to get community servers up and running.

This article originally appeared on Engadget at https://www.engadget.com/gaming/xbox/xbox-game-pass-additions-for-august-include-assassins-creed-mirage-144428943.html?src=rss

©

© Ubisoft

They might be nearly three years old at this point, but Apple’s second-generation AirPods Pro remain some of the most popular wireless earbuds around for a reason, and right now they’re down to $169 on Amazon. That’s a not-insignificant 32 percent off the usual $249 price of the noise-canceling in-ears.

Apple’s flagship AirPods were $100 off — a record-low $149 — during Prime Day, so we have seen them even cheaper than this, but for a non-sale period, Amazon’s current deal is worth considering. The AirPods Pro 2 are our top choice for iPhone users looking for wireless earbuds, thanks to much beefier sound than the original model, rock solid ANC and transparency modes, and seamless switching between Apple devices.

Throw in spatial audio support, hands-free Siri and more than acceptable battery life, and the AirPods Pro 2 are hard to top. Not everyone gets on with the sometimes fiddly touch gestures, but a lot of wireless headphones struggle with this. If there’s anything to give you pause, it’s that we might be getting the AirPods Pro 3 sooner rather than later. But if that’s the case, you can be confident you’ll pay a lot more than $169 for the privilege.

The AirPods Pro aren’t the only discounted AirPods right now. The AirPods 4 (without ANC) are down to $99. Usually priced at $129, you’re saving $30 on a much improved entry-level AirPods offering. They fit and sound better than any non-Pro AirPods before them, and inherit some previously Pro-only features thanks to the H2 chip. If you travel a lot you’ll also certainly rue the lack of ANC, but if that isn’t a deal breaker for you, the sub-$100 price point might be tempting.

Check out our coverage of the best Apple deals for more discounts, and follow @EngadgetDeals on X for the latest tech deals and buying advice.

This article originally appeared on Engadget at https://www.engadget.com/deals/apples-airpods-pro-2-drop-to-169-on-amazon-142401241.html?src=rss

©

© Engadget

At start of December, Google DeepMind released Genie 2. The Genie family of AI systems are what are known as world models. They're capable of generating images as the user — either a human or, more likely, an automated AI agent — moves through the world the software is simulating. The resulting video of the model in action may look like a video game, but DeepMind has always positioned Genie 2 as a way to train other AI systems to be better at what they're designed to accomplish. With its new Genie 3 model, which the lab announced on Tuesday, DeepMind believes it has made an even better system for training AI agents.   

At first glance, the jump between Genie 2 and 3 isn't as dramatic as the one the model made last year. With Genie 2, DeepMind's system became capable of generating 3D worlds, and could accurately reconstruct part of the environment even after the user or an AI agent left it to explore other parts of the generated scene. Environmental consistency was often a weakness of prior world models. For instance, Decart's Oasis system had trouble remembering the layout of the Minecraft levels it would generate. 

By comparison, the enhancements offered by Genie 3 seem more modest, but in a press briefing Google held ahead of today's official announcement, Shlomi Fruchter, research director at DeepMind, and Jack Parker-Holder, research scientist at DeepMind, argued they represent important stepping stones in the road toward artificial general intelligence.

So what exactly does Genie 3 do better? To start, it outputs footage at 720p, instead of 360p like its predecessor. It's also capable of sustaining a "consistent" simulation for longer. Genie 2 had a theoretical limit of up to 60 seconds, but in practice the model would often start to hallucinate much earlier. By contrast, DeepMind says Genie 3 is capable of running for several minutes before it starts producing artifacts.

Also new to the model is a capability DeepMind calls "promptable world events." Genie 2 was interactive insofar as the user or an AI agent was able to input movement commands and the model would respond after it had a few moments to generate the next frame. Genie 3 does this work in real-time. Moreover, it’s possible to tweak the simulation with text prompts that instruct Genie to alter the state of the world it’s generating. In a demo DeepMind showed, the model was told to insert a herd of deer into a scene of a person skiing down a mountain. The deer didn't move in the most realistic manner, but this is the killer feature of Genie 3, says DeepMind.

As mentioned before, the lab primarily envisions the model as a tool for training and evaluating AI agents. DeepMind says Genie 3 could be used to teach AI systems to tackle "what if" scenarios that aren't covered by their pre-training. "There are a lot of things that have to happen before a model can be deployed in the real world, but we do see it as a way to more efficiently train models and increase their reliability," said Fruchter, pointing to, for example, a scenario where Genie 3 could be used to teach a self-driving car how to safely avoid a pedestrian that walks in front of it.

Google DeepMind

Despite the improvements DeepMind has made to Genie, the lab acknowledges there's much work to be done. For instance, the model can't generate real-world locations with perfect accuracy, and it struggles with text rendering. Moreover, for Genie to be truly useful, DeepMind believes the model needs to be able to sustain a simulated world for hours, not minutes. Still, the lab feels Genie is ready to make a real-world impact.

"We already at the point where you wouldn't use [Genie] as your sole training environment, but you can certainly finds things you wouldn't want agents to do because if they act unsafe in some settings, even if those settings aren't perfect, it's still good to know," said Parker-Holder. "You can already see where this is going. It will get increasingly useful as the models get better."

For the time being, Genie 3 isn't available to the general public. However, DeepMind says it's working to make the model available to additional testers.

This article originally appeared on Engadget at https://www.engadget.com/ai/google-deepminds-genie-3-can-dynamically-alter-the-state-of-its-simulated-worlds-140052124.html?src=rss

©

© Google DeepMind