We may earn a commission from links on this page. Deal pricing and availability subject to change after time of publication.

TCL is exploring making tech beyond QLED TVs, and the RayNeo Air 4 Pro AR/XR Smart Glasses are its latest project. These glasses project your phone, laptop, or gaming system into a 201-inch virtual display that only you can see. Amazon has them available for $249 (originally $299) when you use the on-page coupon. At $249, they would be at their lowest price since their recent release date, according to price-tracking tools.

RayNeo Air 4 Pro AR/XR Smart Glasses

201″ HDR10 Display, Bang & Olufsen Audio

$299.00
at Amazon

$299.00
at Amazon

The RayNeo Air 4 Pro just came out in late February, but they’re far from the only screen-mirroring wearable monitor you can buy on the market. The XReal One Pro are another pair that competes with the RayNeo Air 4 Pro. Senior Staff Writer Stephen Johnson tested both in case you want to see if one or the other fits your needs better. However, Stephen named the RayNeo Air 4 Pro the best value in AR right now, with a price that punches well above its weight by offering flagship features that more expensive competitors have.

You can watch more than movies or shows or scroll on your phone with these glasses. The 120Hz refresh rate means gaming looks good, too—you can connect a PS5, Xbox, Switch, or your phone to the virtual screen. There’s also a 3D feature that upgrades 2D media as long as it’s saved in your phone or laptop (no streaming).

The audio works with four speakers tuned with Bang & Olufsen using directional sound, very much like open-ear headphones, offering a surround sound and spatial audio feature that makes the viewing experience more immersive. Keep in mind that these won’t work for productivity if you want to use them as a second monitor, since it projects into wherever you’re looking; if you look at your laptop screen, the virtual projection will overlap your computer monitor.

Runna is one of the most talked-about training apps in running communities, and Strava’s acquisition of the platform earlier this year only cemented its status as the go-to tool for runners who want structure without hiring a coach. Recently, however, not all the buzz has been good. All over Reddit and TikTok, runners are blaming virtual coaches and algorithmic training programs for their shin splints, stress fractures, and various running injuries. Some blame Runna in particular for pushing runners too aggressively.

The thing is, Runna isn’t uniquely to blame. Running injuries are extremely common. Studies consistently estimate that somewhere between 27 and 52% percent of runners experience at least one injury per year, usually due to overuse. At the same time, there are real mistakes that people make when blindly trusting app-based training plans. Here’s what to know to avoid injury so you can stay running strong.

Understand the logic behind the training plan (and adjust as needed)

I’ve previously written about how to choose and trust a training plan, along with recommendations for resources that are completely free and widely trusted (like Hal Higdon’s here). Whenever I have a race on the horizon, I need to understand why my plan works the way it does. It’s important for me to understand the logic behind my mileage, so that I can always stay in touch with my body and make informed decisions as the weeks go by.

In this vein, I think the Runna app is genuinely good—it builds personalized training plans, adjusts to your fitness level, and makes structured training accessible to people who previously had no idea where to start. But if you follow an app’s training plan without listening to your body, the app will not stop you from pushing yourself too hard. That means you are always the last line of defense—and with any training plan, that responsibility doesn’t go away just because an all-knowing algorithm built your schedule. Across social media, this seems especially risky for two groups of runners:

• Beginners who don’t yet have the experience to recognize warning signs. When you’re new to structured training, it’s hard to distinguish between normal soreness and something more dangerous. The enthusiasm of having a plan can override the quieter signals your body is sending.

• Aspiring influencers and highly motivated runners who have built an identity around consistency and hitting their targets. For this group, rest days and missed sessions feel like failure. 

If you understand the reasoning behind your runs, you’ll be able to adapt your plan to your needs over time. My issue with programs like Runna is when individual runners aren’t bringing enough wisdom and skepticism into their relationship with the app.

Watch for these warning signs in any training plan

I will say, Runna’s default plans are not exactly conservative. They’re designed to get results, which typically means progressive overload—gradually increasing mileage and intensity week over week. For a runner who has built a solid base, this is fine. For a runner who has overstated their current fitness, or who is coming back from time off, the default settings could be way too aggressive. 

Specific things to watch for:

• Week-over-week mileage jumps that exceed 10%. This isn’t a hard and fast rule, but in personal experience, it holds up. Generally speaking, you should never increase your mileage more than 10% from week to week. If a training plan is pushing you beyond that, pay extra close attention to how your body responds.

• Back-to-back hard sessions. If you’re not recovering well between tough workouts, that’s a signal worth acting on.

• Insufficient easy running. Many runners who use Runna—especially those who are newer to structured training—end up running too much of their mileage at moderate effort, rather than truly easy. Easy really does mean easy: You should be able to hold a full conversation. If your “easy” runs feel like honest work, slow down, even if the pace targets suggest otherwise.

Luckily, you can adjust the intensity of your plan in Runna. Open the “plan” tab of your app, head to “manage plan” and select “training preferences,” which Runna explains here.

Always pay attention to these signs of a running injury

This is the non-negotiable list. No plan—AI-generated or otherwise—is worth running through these:

• Sharp or localized pain during a run. Some soreness is normal, but a specific point of pain that gets worse as you run is not.

• Pain that changes your gait. If you’re limping, compensating, or noticeably favoring one side, your body is asking you to stop in the only language it has.

• Pain that is worse the morning after a hard session than it was during the run. Post-run soreness that peaks 24–48 hours later is typical. Pain that is sharper the next morning than it was mid-run could be a red flag.

• Bone pain on impact. Any pain that feels deep, localized to a bone (shin, foot, hip), and is triggered specifically by the impact of your foot striking the ground might warrant real medical attention. Stress fractures are terrible news and all too common in people who ramp mileage too fast.

• Persistent joint pain. Knees, hips, and ankles that hurt run after run, even on easy days, are telling you that your training load exceeds your current ability to recovery.

If any of these show up, the right move is not to finish the session and reassess. The right move is to stop, rest, and if the symptom persists, see someone.

This is the best way to use Runna

At the end of the day, think of Runna the way you’d think of a GPS: an excellent navigational tool that still requires a driver who’s paying attention to the road. Here’s a practical framework:

1. Be honest about your starting point. Runna can only work with the information you give it. If you overstate your current weekly mileage or recent race times, you will get a plan that assumes a fitness level you don’t have.

2. Treat the first two weeks as a test. Are the easy runs actually easy? Are you recovering between sessions? Is the total weekly volume a stretch but manageable, or is it immediately overwhelming? Adjust as you go.

3. Use those “training preferences” settings. If you’re struggling, dial it back.

4. Add recovery weeks deliberately. Good training plans include scheduled “down weeks” with reduced mileage to allow adaptation. Make sure your Runna plan includes these, and if you’re feeling beat up heading into one, treat it as mandatory, not optional.

5. Run your easy days truly easy. I’ll say it again and again: Most runners run their easy days too hard. Try to run slower than you think you should.

6. Take the rest days. It helps to remember that adaptation happens during recovery, not during the run itself.

The criticism that Runna has received for causing injuries is not entirely without basis, but it’s also not entirely fair. Injuries are common in running. If you think about it, any tool that helps people train harder will, statistically, correlate with more injuries. Good, hard training is inherently risky. However, the risk is totally manageable. Managing it requires you to stay in the driver’s seat, remaining a little skeptical of any one resource. You need to know how to be honest about your fitness, attentive to your body’s signals, and willing to adjust the plan rather than blindly execute it.

It must be 2017 because loot boxes are back in the news again. Two weeks after New York’s attorney general sued Valve over its use of the gimmick, the company has responded. In short, the Steam maker essentially said, “See you in court.”

New York’s lawsuit accuses Valve of promoting illegal gambling through its games. AG Letitia James called the loot boxes found in titles like Counter-Strike 2, Team Fortress 2 and Dota 2 “addictive, harmful and illegal.” The state seeks to “permanently stop Valve from continuing to promote illegal gambling in its games” and pay relevant fines.

In its defense posted on Thursday, Valve likened its mystery boxes to kids buying packs of physical trading cards. “Players don’t have to open mystery boxes to play Valve games,” the company wrote. “In fact, most of you don’t open any boxes at all and just play the games — because the items in the boxes are purely cosmetic, there is no disadvantage to a player not spending money.”

That last point, while applicable within the game itself, isn’t quite that cut and dry once you zoom out beyond that. As James pointed out, players can trade the cosmetic items they win from loot boxes on Steam’s marketplace or sell them on third-party marketplaces. Rarer ones can sometimes fetch lucrative sums.

DMarket

Here, too, Valve defended the profitable practice by rolling out the trading card comparison. “We think the transferability of a digital game item is good for consumers — it gives a user the ability to sell or trade an old or unwanted item for something else, in the same way an owner can sell or trade a tangible item like a Pokémon or baseball card,” the company wrote. “NYAG proposes to take away users’ ability to transfer their digital items from Valve games. Transferability is a right we believe should not be taken away, and we refuse to do that.”

Valve is also facing a new class-action lawsuit over its loot boxes.

Some of Valve’s points land a bit more than its righteous defense of a gaming gimmick that, well, isn’t exactly beloved. The company accused the NYAG of proposing that Valve collect additional user information to prevent VPN use. In addition, the state allegedly “demanded that Valve collect more personal data about our users to do additional age verification.” Privacy experts have been sounding the alarm about the recent push for online age verification.

Valve also addressed James’s erroneous and outdated statement that video games encourage real-world violence. “Those extraneous comments are a distraction and a mischaracterization we’ve all heard before,” the company wrote. “Numerous studies throughout the years have concluded there is no link between media (movies, TV, books, comics, music and games) and real world violence. Indeed, many studies highlight the beneficial impact of games to users.”

The company says that, while it may have been cheaper to settle the suit, it deemed the NYAG’s demands user-hostile. “Ultimately, a court will decide whose position — ours or NYAG’s — is correct. In the meantime, we wanted to make sure you were aware of the potential impact to users in New York and elsewhere.”

This article originally appeared on Engadget at https://www.engadget.com/gaming/valve-defends-loot-boxes-in-response-to-new-yorks-lawsuit-190655554.html?src=rss

TikTok will soon let you stream full songs in its app via a new integration with Apple Music. The company’s new Play Full Song feature makes it possible to link your Apple Music account toTikTok, and play any song that strikes your fancy directly in the app while you’re scrolling.

Starting a song is as simple as tapping a button in the Sound Details page or your For You page. Assuming you pay for Apple Music, TikTok will then open up a streamlined version of Apple’s music player, which you can use to listen to the song, save it for later or add it to a playlist.

TikTok says that Play Full Song is built using Apple’s MusicKit APIs, which let developers surface elements of the Apple Music streaming service in their apps. TikTok has previously offered integration with multiple music streaming services through a feature it calls Add to Music App, which made it possible to save songs you heard on TikTok to your streaming library. What’s particularly interesting about this new integration is that because it’s using Apple’s APIs, songs streamed with Play Full Song count as normal streams for the artists in Apple Music, so they don’t lose out on any money.

Alongside the new feature, TikTok and Apple are also introducing a way for fans to listen to music live with their favorite artists. TikTok’s Listening Party feature creates a live “shared environment” where people can listen to music and interact with artists directly, in what effectively sounds like an audio-only livestream. TikTok livestreams are a whole ecosystem in their own right, and Listening Party seems like a way to leverage some of the same technology for a more controlled, music promotion-focused end.

TikTok is already a popular tool for music discovery and launching the career of new artists, and the platform also briefly dabbled in offering a streaming service of its own in 2023. The company abandoned those plans in 2024, but under new owners, TikTok’s ambitions could ultimately be bigger than just offering nice integrations with existing streaming services.

TikTok says Play Full Song and Listening Party are rolling out worldwide “in the weeks ahead,” so if you don’t see either feature now, you may soon.

This article originally appeared on Engadget at https://www.engadget.com/apps/tiktok-will-let-you-stream-full-songs-in-its-app-if-youre-an-apple-music-subscriber-183333143.html?src=rss

Microsoft first debuted its full screen Xbox experience for Windows in the ROG Ally Xbox handheld, in a bid to compete with Steam’s nearly 15-year-old Big Picture Mode. That Xbox interface eventually made its way to other Windows 11 gaming portables last year. Today at GDC, Microsoft revealed that its big screen Xbox UI is headed to all Windows 11 devices (including laptops and desktops) in April. Oh yah, and it’s now simply called “Xbox Mode.”

Xbox Mode will only be available in select markets at first, and Microsoft describes it as bringing “a controller-optimized experience to your Windows 11 device, letting players browse their library, launch games, use Game Bar and switch between apps.” You know, just like Steam Big Picture mode. Microsoft didn’t have much else to share about optimizations in Xbox Mode, but when it debuted the feature for Windows 11 Insiders last fall, the company noted that its task switcher will let people quickly move between games, as well as their apps.

Microsoft plans to reveal more information about the future of Xbox today at GDC. Last week, Xbox CEO Asha Sharma confirmed the company’s next hardware is codenamed “Project Helix,” and it will play both PC and console games. That likely means it’ll just be a Windows gaming PC with Xbox branding, something the company has been hinting at for a while.

Microsoft also has some geekier developer-focused news for the Games Developer Conference. Advanced Shader Delivery (ASD), which first appeared on the Xbox ROG Ally, will be made available to all developers on the Xbox store. ASD allows delivers to pre-compile shaders, so you’re not stuck waiting for them to get processed on your system. That should also help to avoid the shader stuttering so common when playing a new title, since shader processing often occurs in the background too.

DirectStorage, Microsoft’s technology for speeding up game loading on NVMe SSDs, is also getting support for Zstandard compression, as well as a tool called the “Game Asset Conditional Library.” According to Microsoft, that tool enables “improving compression efficiency while simplifying asset conditioning across production pipelines.” Microsoft also plans to give developers a glimpse at how next-generation Machine Learning will be implemented in its DirectX gaming API.

This article originally appeared on Engadget at https://www.engadget.com/gaming/xbox/microsofts-full-screen-xbox-mode-will-roll-out-to-windows-11-pcs-in-april-181000289.html?src=rss

Binance is hoping that suing The Wall Street Journal for defamation might help shake off a fresh round of government probes into how the cryptocurrency exchange failed to detect $1.7 billion in transfers to a network that was funding Iran-backed terror groups.

The lawsuit comes after a Wall Street Journal investigation, based on conversations with insiders and reviews of internal documents, reported that Binance had quietly dismantled its own investigation into the unlawful transfers and then fired compliance staff who initially flagged them.

Alleging that the report falsely accused Binance of retaliation—among 10 other allegedly false claims—Binance accused the Journal of conducting a “sham” investigation that intentionally disregarded the company’s statements. That included supposedly failing to note that Binance had not closed its investigation into the unlawful transfers.

Read full article

Comments

Microsoft plans to begin shipping early units of its next generation console, codenamed Project Helix, to game studios starting sometime next year. “We’re sending alpha versions of Project Helix to developers starting in 2027,“ said Jason Ronald, vice-president of next generation for Xbox, according to IGN, which was present at the company’s GDC 2026 presentation where it shared early details about the new device. Ronald did not clarify what he meant by “alpha version,” but given the keynote’s developer focus, presumably he meant devkits, which studios could use to start creating games for the new console.

Additionally, Ronald reiterated that the new system would be capable of playing both Xbox console games and PC games, and said it would incorporate a custom AMD-made system-on-a-chip capable of rendering graphics with path tracing. Judging from a slide the company shared, Microsoft and AMD are working on many of the same technologies and capabilities AMD is co-designing with Sony for next PlayStation console. For instance, Ronald said Helix would be capable of ray regeneration, a technique designed to produce better-looking ray-traced effects. The new console will also offer multi-frame frame generation and machine learning-based upscaling.

“It delivers an order of magnitude leap in ray tracing performance and capability, integrates intelligence directly into the graphics and compute pipeline, and drives meaningful gains in efficiency, scale, and visual ambition. The result is more realistic, immersive, and dynamic worlds for players,” Ronald wrote in a blog post published after his presentation.

Ronald didn’t speak to any specific compute numbers, likely due to the fact Microsoft has yet to finalize the Helix hardware. We’ll likely learn more of those details the closer we get to 2027.

This article originally appeared on Engadget at https://www.engadget.com/gaming/xbox/microsoft-will-start-providing-game-studios-with-project-helix-consoles-in-2027-180352458.html?src=rss

Nvidia is preparing to launch an open-source AI agent platform called NemoClaw, designed to compete with the likes of OpenClaw. According to Wired, the platform will allow enterprise software companies to dispatch AI agents to perform tasks for their own workforces. “Companies will be able to access the platform regardless of whether their products run on Nvidia’s chips,” the report adds. From the report: The move comes as Nvidia prepares for its annual developer conference in San Jose next week. Ahead of the conference, Nvidia has reached out to companies including Salesforce, Cisco, Google, Adobe, and CrowdStrike to forge partnerships for the agent platform. It’s unclear whether these conversations have resulted in official partnerships. Since the platform is open source, it’s likely that partners would get free, early access in exchange for contributing to the project, sources say. Nvidia plans to offer security and privacy tools as part of this new open-source agent platform. […]

For Nvidia, NemoClaw appears to be part of an effort to court enterprise software companies by offering additional layers of security for AI agents. It’s also another step in the company’s embrace of open-source AI models, part of a broader strategy to maintain its dominance in AI infrastructure at a time when leading AI labs are building their own custom chips. Nvidia’s software strategy until now has been heavily reliant on its CUDA platform, a famously proprietary system that locks developers into building software for Nvidia’s GPUs and has created a crucial “moat” for the company.

Read more of this story at Slashdot.

The Fast and Furious franchise has come a long way in the quarter-century since the first film’s release. Originally an undercover cop story, the franchise has morphed into… something else entirely. It’s now a bombastic expression of automotive culture combined with some kind of caper, maybe to save the world. Just don’t think too deeply about the plot.

Along the way, the film’s cars have become nearly as famous as the human stars. If you’re a fan, you probably can’t have Vin Diesel or Michelle Rodriguez come hang out with you in your garage, but you can drive a Charger or Eclipse—or even a Jetta that looks like it escaped from the set. The more well-off collectors don’t need to settle for building a replica, though; they actually own cars that appeared on screen, and there’s quite a community of Fast and Furious car collectors.

You can find some of these cars at the Petersen Automotive Museum, which has a new exhibit celebrating 25 years of the franchise.

Read full article

Comments

A new company called Musical Beings has officially unveiled the Tembo, which might be the cutest drum machine ever made. Just look at this thing! It’s got a wooden chassis that resembles a standard drum machine, but with one key difference. The sequencer is tactile. Users arrange beats by placing magnetic pucks that trigger samples.

This seems like a really good way to introduce the basics of sequencing and beatmaking to kids and young adults, being that DAWs and grooveboxes can feature a steep learning curve. The sequencer isn’t all that different from what’s found on a typical groovebox, but the analog nature of it seems novel.

The company says it designed Tembo to “enable everyone to create music from the very first touch.” Co-founder David Davidov told MusicRadar that most instruments take “so long to get to the fun part” and that Musical Beings wanted to “help people experience music as something they do, not just something they listen to.”

Just because it’s accessible to kids and amateurs doesn’t mean it’s not for seasoned musicians. This is a real-deal drum machine with plenty of nifty features. There’s a five-channel, 16-step sequencer that’s controlled via the aforementioned circular magnets. The machine includes knobs for swing, tempo, effects and pattern length.

It has two USB-C MIDI connections, so it can easily be hooked up to a DAW or synced with external gear. Sessions can be recorded via USB audio or a stereo output. There’s also a dedicated companion app to help with that sort of thing.

The Tembo is battery-powered, making it relatively portable, and there’s a built-in speaker. The integrated sampler lets users lay down musical ideas in addition to beats, making it something of a junior groovebox. This is assisted by a built-in microphone. 

The Kickstarter just launched, but has already soared past the initial goal. The price ranges from around $360 to $450 depending on the tier. It’s worth noting that Musical Beings is a new company and Kickstarter projects are never guaranteed to come out. However, a number of units have already been built, as some musicians and studios have already gotten their hands on them.

This isn’t the first wacky drum machine that has come across our desk. The BeatBox is a cardboard gadget that uses arcade-style buttons to make beats. The OddBall is quite literally a ball that makes beats as it bounces around.

This article originally appeared on Engadget at https://www.engadget.com/audio/tembo-might-just-be-the-worlds-cutest-drum-machine-173926914.html?src=rss

Intel’s Core Ultra 200S desktop chips, codenamed “Arrow Lake,” first launched in late 2024, and they were the most significant updates to Intel’s desktop CPU lineup in years. But that didn’t mean they were always improvements over what came before: while they’re power-efficient and run cooler than older 13th- and 14th-generation Core CPUs, they sometimes struggled to match those older chips’ gaming performance. And for gaming systems in particular, they’ve always had to live in the shadow of AMD’s Ryzen 7000 and 9000-series X3D processors, chips with extra L3 cache that disproportionately benefits games.

Intel doesn’t have a next-generation upgrade available for desktops yet, but it is shoring up its desktop lineup with a pair of upgraded chips. The Core Ultra 200S Plus processors (also referred to as Arrow Lake Refresh, in some circles) add more processor cores, boost clock speeds, add support for faster memory, and speed up the internal communication between different parts of the processor. Collectively, Intel says these improvements will boost gaming performance by an average of 15 percent.

The Core Ultra 7 270K Plus and 270KF Plus (a real mouthful, all of these names are getting to be) add four more efficiency cores compared to the Core Ultra 7 265K, bringing the total number of cores to 24 (8 P-cores and 16 E-cores). If you wanted that many CPU cores previously, you would have had to spring for a Core Ultra 9 chip. The Core Ultra 5 250K Plus and 250KF Plus also get four more E-cores than the 245K, bringing its total to 6 P-cores and 12 E-cores.

Read full article

Comments

Meta has announced that it’s introducing parent-managed accounts on WhatsApp. Designed to allow young people under the age of 13 to use the messaging platform more safely, these accounts feature new controls that enable a parent or guardian to restrict who can send them messages. Parent-managed accounts can also only be used for messaging and calling, so additional features like Channels, location sharing and Meta AI integration aren’t included.

To set up an account, you’ll need to put your phone next to the pre-teen’s device to link the two accounts. Once that’s done, the person managing the kids’ account can decide who’s able to contact them and which groups they’re able to join. Step-by-step instructions on how to activate the new accounts can be found here.

They’ll also see message requests from unknown contacts first and can adjust privacy settings from the managed device. Parent-managed accounts are PIN-protected and only the parent or guardian can make changes to privacy settings.

Like all WhatsApp conversations, end-to-end encryption means nobody else can see messages exchanged on parent-managed accounts. By default, only saved contacts can message a managed account, and a child won’t be able to join a group or view group invites from strangers before they’re separately approved by the owner of the parent account. These requests will appear as notifications to the parent.

WhatsApp doesn’t specify a minimum age suitable for a parent-messaged account, but says it’ll roll the new features out gradually in the coming months.

Meta has spent the last few years ramping up its parental controls features across its various platforms. In September it introduced teen accounts — aimed at teens between the age of 13 and 15 — for Facebook and Messenger. A year earlier, Under-16 teen accounts became a requirement on Instagram. Like the new parent-managed accounts on WhatsApp, these allow parents to vet requests and enable stricter privacy settings.

At the start of 2026, Meta put a temporary pause on allowing teens to interact with its AI chatbot characters, following reports that some of these bots had engaged in sexual conversations and other concerning interactions with minors.

This article originally appeared on Engadget at https://www.engadget.com/social-media/meta-will-let-kids-under-13-use-whatsapp-with-parent-managed-accounts-172023976.html?src=rss

YouTube is expanding its AI deepfake detection tools to a pilot group of politicians, government officials, and journalists, allowing them to identify and request removal of unauthorized AI-generated videos impersonating them. TechCrunch reports: The technology itself launched last year to roughly 4 million YouTube creators in the YouTube Partner Program, following earlier tests. Similar to YouTube’s existing Content ID system, which detects copyright-protected material in users’ uploaded videos, the likeness detection feature looks for simulated faces made with AI tools. These tools are sometimes used to try to spread misinformation and manipulate people’s perception of reality, as they leverage the deepfaked personas of notable figures — like politicians or other government officials — to say and do things in these AI videos that they didn’t in real life.

With the new pilot program, YouTube aims to balance users’ free expression with the risks associated with AI technology that can generate a convincing likeness of a public figure. […] [Leslie Miller, YouTube’s vice president of Government Affairs and Public Policy] explained that not all of the detected matches would be removed when requested. Instead, YouTube would evaluate each request under its existing privacy policy guidelines to determine whether the content is parody or political critique, which are protected forms of free expression. The company noted it’s advocating for these protections at a federal level, too, with its support for the NO FAKES Act in D.C., which would regulate the use of AI to create unauthorized recreations of an individual’s voice and visual likeness.

To use the new tool, eligible pilot testers must first prove their identity by uploading a selfie and a government ID. They can then create a profile, view the matches that show up, and optionally request their removal. YouTube says it plans to eventually give people the ability to prevent uploads of violating content before they go live or, possibly, allow them to monetize those videos, similar to how its Content ID system works. The company would not confirm which politicians or officials would be among its initial testers, but said the goal is to make the technology broadly available over time.

Read more of this story at Slashdot.

We may earn a commission from links on this page. Deal pricing and availability subject to change after time of publication.

That didn’t take long. The M4 iPad Air just came out today, and you can already pick one up at a (small) discount. The wifi 128GB version of the new iPad Air is currently $749, down from $799. While $50 or $800 is not a big discount, the fact that Amazon is cutting the price on Apple’s latest flagship iPad on its release day is unusual. This price cut applies to the bigger 13-inch model, while the smaller 11-inch model iPad Air is $559, $40 off the list price of $599 and matching Walmart’s pre-order deal.

Liquid Retina Display, 128GB, 12MP Front/Back Camera, Wi-Fi 7 with Apple N1, Touch ID, All-Day Battery Life

Apple iPad Air 11-inch (M4)

$559.00
at Amazon

$599.00
Save $40.00

$559.00
at Amazon

$599.00
Save $40.00

Liquid Retina Display, 128GB, 12MP Front/Back Camera, Wi-Fi 7 with Apple N1, Touch ID, All-Day Battery Life

Apple iPad Air 13-inch (M4)

$749.00
at Amazon

$799.00
Save $50.00

$749.00
at Amazon

$799.00
Save $50.00

M3 iPad Air owners should not get too excited—there’s nothing notable here other than the presence of the more powerful M4 chip, which will may offer a noticeable efficiency boost over its predecessor. That’s likely due to having one more efficiency core than the M3 Air. The biggest difference, however, is the extra 4GB of RAM (12GB total). This means you can multitask for longer with multiple tabs and apps running.

According to Apple, this new iPad should be up to 30% faster than the previous generation, although we won’t know for sure until reviewers have tested it out. Other upgrades include Wi-Fi 7, Bluetooth 6, and compatibility with the Thread smart home standard. The other specs are the same: 12MP rear and front cameras, USB-C connectivity with Touch ID, 10 hours of video playback, and 128GB of storage for the base model.

If you still have the M3 iPad or another recent iPad, it’s probably not worth upgrading. However, if you have an older iPad (or none at all), this is a good opportunity to get Apple’s latest iPad for the best price you’re likely to see for a while.

Anduril Industries announced on Wednesday that it is acquiring ExoAnalytic Solutions, a space intelligence firm that operates a vast network of sensors monitoring the veiled movements of satellites thousands of miles above Earth.

“For nearly twenty years, ExoAnalytic has delivered important advantage[s] for the nation’s most critical missions,” Anduril said in a press release. “Exo is a renowned leader in modeling and simulation for classified national security space programs, and provides critical software and expertise for missile warning and missile defense.”

“The company also owns and operates the world’s largest commercial telescope network with more than 400 systems deployed worldwide, enabling persistent, high-fidelity awareness of deep space at a global scale,” Anduril said.

Read full article

Comments

Eight of the 10 most popular AI chatbots were willing to help plan violent attacks when tested by researchers, according to a new study from the Center for Countering Digital Hate (CCDH), in partnership with CNN. While both Snapchat’s My AI and Claude refused to assist with violence the majority of the time, only Anthropic’s Claude “reliably discouraged” these hypothetical attackers during testing.

Researchers created accounts posing as 13-year-old boys and tested ChatGPT, Gemini, Claude, Copilot, Meta AI, DeepSeek, Perplexity, Snapchat My AI, Character.AI and Replika across 18 scenarios between November and December 2025. The tests simulated users planning school shootings, political assassinations and bombings targeting synagogues. Across all the responses analyzed, the chatbots provided “actionable assistance” roughly 75 percent of the time and discouraged violence in just 12 percent of cases. This was the average across all chatbots, with Claude discouraging violence 76 percent of the time.

Meta AI and Perplexity were the least safe, assisting in 97 and 100 percent of responses. ChatGPT offered campus maps when asked about school violence, and Gemini said metal shrapnel is typically more lethal in a synagogue bombing scenario.

DeepSeek signed off rifle selection advice with “Happy (and safe) shooting!” Character.AI, which the report described as “uniquely unsafe,” actively encouraged violence in seven instances, at one point telling a researcher to “use a gun” on a health insurance company CEO. In another scenario, it provided a political party’s headquarters address and asked if the user was “planning a little raid.”

Meta told CNN it had taken steps to “to fix the issue identified,” while Google and Open AI said they had implemented new models since the study was conducted. Sixty-four percent of US teens aged 13 to 17 have used a chatbot, according to Pew Research.

This article originally appeared on Engadget at https://www.engadget.com/ai/most-ai-chatbots-will-help-users-plan-violent-attacks-study-finds-163651255.html?src=rss

Chipmaker Nvidia is preparing to launch its own open source AI agent platform to compete with the likes of OpenClaw, according to a recent Wired report.

The magazine cites “people familiar with the company’s plans” in reporting that Nvidia has been pitching the platform, which it is calling NemoClaw, to various corporate partners ahead of its annual developer conference next week. Salesforce, Cisco, Google, Adobe, and CrowdStrike are among the companies said to be in talks for those partnerships, though it’s unclear what specific benefits those companies would receive for their association with the open source tool.

NemoClaw, as the somewhat awkward name suggests, would be a direct competitor of OpenClaw (previously known as Moltbot and Clawdbot), the system that attracted widespread attention in January for letting users direct “always-on” AI agents from their personal machines, using any number of underlying models. Last month, OpenAI hired OpenClaw creator Peter Steinberger “to drive the next generation of personal agents,” as founder Sam Altman put it, though the OpenClaw project will be run by an independent foundation with OpenAI’s support.

Read full article

Comments