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
Sony has announced its latest additions to the PlayStation Plus Game Catalog. While they might not top last month's introduction of Marvel's Spider-Man 2, Extra and Premium subscribers are still getting access to some notable games.
In March, the new additions include Warhammer 40,000: Space Marine 2 (PS5), one of Engadget's picks for the Best Games of 2024, and Persona 5 Royal (PS5 and PS4). The PS4 version of Persona 5 Royal – Ultimate Edition, the complete version of one of Atlus' most popular RPGs, will also be available alongside a collection of other fun additions.
Warhammer 40,000: Space Marine 2 is a third-person action game reminiscent of Gears of War, but set in Games Workshop's elaborate Warhammer 40,000 universe. If you're a fan of the larger franchise, Space Marine 2 is well worth a look, especially with a third entry on the way, but the game is also a solid option if you're just looking for a new co-op game to try with friends.
Persona 5 Royal is a known-quantity among fans of Atlus' social simulation/RPG series, and the way it both expands on and streamlines the original Persona 5 also makes it a suitable entry point to the series as a whole. If seeing Japanese teenagers deal with their personal problems while embarking on Inception-style missions into the hearts of the corrupt adults of Tokyo sounds interesting, you'll love Persona 5 Royal.
Alongside those standouts, PS Plus Game Catalog is also getting:
EA Sports Madden NFL 26 (PS5)
Blasphemous 2 (PS5 and PS4)
Metal Eden (PS5)
Lord of the Rings: Return to Moria (PS5)
Astroneer (PS5 and PS4)
And as promised last month, the PS Plus Classics Catalog is expanding to include Tekken: Dark Resurrection, a revamped and rebalanced version of Tekken 5 that includes expanded character customization options and new stages that weren't in the original 2004 fighting game. This one’s only for PS Plus Premium subscribers on PS4 and PS5.
All of these will be available on the PS Plus Game Catalog on March 17.
This article originally appeared on Engadget at https://www.engadget.com/gaming/playstation/the-ps-plus-games-catalog-is-getting-space-marine-2-and-persona-5-royal-in-march-162716284.html?src=rss
Fortnite's original Save the World game mode will be free to play beginning on April 16. This is not the battle royale mode that went on to take over the world, but rather a PvE co-op campaign that focuses on crafting and survival. It currently costs $9 but that ends soon.
As a matter of fact, Epic Games will stop selling the paid version of the content tonight, March 11, at 8PM ET. If you have $9 burning a hole in your pocket to spend on something that'll be free next month, go for it. Current players will still be able to access the content, so there are no worries there.
Once the free version goes live, pre-existing players will get a bunch of perks as a thank you of sorts. This includes vouchers for V-bucks, gold bars and end-game resources that are used to boost player and item levels.
Save the World will still be a great way to gather V-bucks, for those worried the freemium shift will change things. Epic says players will earn this currency "through Daily Quests, Mission Alerts, Storm Shield Defense Missions and existing Challenges."
Pre-registration for the free mode is already live and there's a welcome bit of news here. For the first time, Save the World will be available on Switch 2. It'll remain unavailable on the original Switch and smartphones. The mode is also playable on PC, PlayStation and Xbox.
Epic promises those who pre-register will get perks when the free version launches, but says the specifics will vary depending on how many people sigh up by April 16. Save the World originally came out in 2017 and was a modest success that was immediately dwarfed by the PUBG-inspired battle royale mode.
This article originally appeared on Engadget at https://www.engadget.com/gaming/fortnites-original-save-the-world-mode-will-be-free-to-play-starting-on-april-16-160757693.html?src=rss
When you think of an AI-forward PC, you might think of something like NVIDIA's $3,999 DGX Spark — a computer with enough computing power to run complex large language models locally. That's not what Rabbit is trying to build with Project Cyberdeck. Instead, the company's goal is to produce a device tailored for vibe coding, and Engadget was given an exclusive first look at the upcoming PC.
Rabbit began working on Project Cyberdeck after the company's CEO, Jesse Lyu, saw how much his software engineers were using Claude Code. Lyu thought a small form factor PC, like the netbooks that were popular in the late aughts, with a command line interface would be ideal for on-the-go vibe coding, but when he went online to look for something that fit the bill, he was disappointed.
"They all come with shitty rubber dome keyboards," Lyu says of low-cost PCs like the latest Chromebooks, which use flexible silicone sheets under their keys to save on space and cost. "They're not something you would enjoy typing on for an extended period of time." So Rabbit decided to build its own device. For inspiration, Lyu and company looked to an unlikely source: the Sony Vaio P.
The Cyberdeck takes inspiration from the Sony Vaio P.
Sony
Sony's netbook was only briefly available from the start of 2009 to about the end of 2010. At the time, the 8-inch Vaio P was the world's lightest netbook, weighing just 1.4 pounds, but it had a host of issues. It was also expensive, costing considerably more than other Intel Atom notebooks of the time. In 2009, the most affordable Vaio P would set you back $900 (about $1,365 adjusted for inflation). With Project Cyberdeck, Rabbit is aiming for a device that costs about $500, and hopefully avoids a similar fate.
I saw a few early renders of Project Cyberdeck, which Rabbit isn't ready to share publicly yet. Imagine a cross between the Rabbit R1, Vaio P and the original Nintendo DS. It looks cute. All the renders had four USB-C ports to allow users to connect the device to external monitors and peripherals, though the actual IO specs are as-yet undecided.
The company is in the process of sourcing components and working towards a final design, so details can — and will — change. I saw some of the parts Lyu has been testing in his office, but no final prototype as such.
For one, Rabbit still needs to decide on a chipset. The company is aiming for a performance benchmark relative to the Raspberry Pi 5, which has a Broadcom BCM2712 quad-core Arm Cortex A76 processor clocked at 2.4GHz. With 16GB of RAM, the Raspberry Pi 5 can run two external monitors, a capability Rabbit hopes to match with the Cyberdeck. The idea here is to make a device that's powerful enough it won't feel slow when it's communicating with Anthropic and OpenAI's servers, but affordable enough to make it a no-brainer purchase for developers.
The company confirmed Project Cyberdeck will run Linux. Rabbit will allow users to modify the operating system and install any third-party tools they want. Additionally, all the software features the company has developed for RabbitOS will be available through command-line prompts.
Two parts of the device Lyu hopes are major differentiating factors are the keyboard and screen. Lyu appears set on shipping a computer with a 40 percent keyboard that has low-profile mechanical switches and a fully hot swappable PCB, so users can tweak the typing feel to their liking. Lyu also had a sample 7-inch OLED screen on his desk when I spoke to him. That specific panel offers touch input, a 165Hz refresh rate and 815 nits of brightness. While it might not be the one Rabbit settles on, OLED is the goal, because of what it would mean for battery life.
For the uninitiated, OLED panels produce black values by turning off individual diodes, and since each diode is self-emitting, there's no need for a power-hungry backlight. Like every smartphone manufacturer, Rabbit is taking advantage of this by planning to offer a dark mode interface from day one.
One aspect of the Cyberdeck's design Lyu can't definitively speak to is how much RAM it will feature. The entire industry is dealing with datacenter demand for high-bandwidth memory that has sent the price of computers, smartphones and other electronics soaring. Lyu believes Rabbit won't be forced to delay the Cyberdeck out of 2026, but he also didn't rule out the possibility either. If things change for the better, he's confident Rabbit would be able to take advantage, since it took the company about 93 days to ship the first R1 device after it began working on the design.
Separately, I wonder if people will want to carry around a second device solely for their coding needs? You don't need a dedicated machine to access Claude Code or OpenAI Cursor. Even companies like Apple have begun integrating vibe coding services into their development environments. Rabbit could be on track for a repeat of the R1, but with so many details of the Cyberdeck left undecided, for now, it's too early to know for sure. The company will get to make its case when it shares more details in the coming weeks and months.
This article originally appeared on Engadget at https://www.engadget.com/ai/rabbits-cyberdeck-is-a-modern-take-on-a-netbook-151907273.html?src=rss
Phil Spencer leaving his long-held role at Xbox might have made all the headlines last month, but a few years ago the big story was company veteran Larry "Major Nelson" Hryb’s departure from Microsoft. Hryb recently seemingly hinted at a return to the company at which he spent more than two decades, but he’s now landed at Commodore instead.
The ex-Xbox icon joins the recently revived 80s computer brand as a Community Development Advisor, where his job will be to "help support and expand the global community," Commodore said in a press release. Hryb, who was the public face of Xbox during the brand’s heyday, is now tasked with helping to modernize Commodore by introducing it to "a new generation of creators, developers, and enthusiasts."
"I've always believed the best thing a company can do is partner with its community – and with a passionate fan base carrying the torch for 31 years, Commodore’s situation is truly unique when it comes to community engagement," said Hryb. "The community didn't wait around – instead they built something remarkable. Players, hardware hobbyists, developers, content creators, and publishers are all a part of the Commodore community, and now we get to build what’s next together."
Hryb’s most recent role was at game engine maker Unity, where he served as Director of Community and Advocacy for less than two years before being laid off in January. As for Commodore, the company might be entering a new era, but its comeback product launch is a firmly nostalgic play, with the recently released Commodore 64 Ultimate being an authentic recreation of its most famous 8-bit computer.
This article originally appeared on Engadget at https://www.engadget.com/gaming/larry-major-nelson-hryb-joins-commodore-to-help-build-its-community-145908119.html?src=rss
Looking Glass has been doggedly committed to making holographic displays the next big thing since 2019, and with its new Musubi digital photo frame, it might finally be offering its tech at a price that's hard to deny. Musubi is scheduled to start shipping in June, and unlike the company's previous, more developer-focused kits, the company's new display only costs $149.
Musubi is a 7-inch frame with a glass border and white matte that acts as the home for whatever content you convert and upload to it. Looking Glass says the Musubi can store up to 1,000 images or 30-second video clips, and is able to display your content for three hours on a single charge, or indefinitely if you plug it in with an included wall adapter. You'll have to convert your photos and videos into holographic files using Looking Glass' free desktop app in order to display them, but once they're converted, all you need to do is transfer them over USB-C to start showing them off on Musubi.
Musubi can also cycle through multiple holographic images.
Looking Glass
Looking Glass has offered multiple versions of this concept before — including the compact, $300 Looking Glass Go from 2023 — but Musubi is supposed to be the best representation of the company's current display stack. The frame uses the Hololuminescent Display (HLD) technology Looking Glass announced in 2025, which "combines 2D display layers with a 3D holographic volume" to show off holograms that are viewable by multiple people at the same time, without the need for eye-tracking or glasses. It's hard to get a sense for the whole Musubi experience from the company's YouTube video alone, but the results seem novel, if a bit limited.
You can pre-order Musubi starting today through Looking Glass' Kickstarter campaign. For the first 24 hours of the company's Kickstarter, the frame will be available for $99. Afterwards, Musubi will sell for $149. Anything on Kickstarter should be treated with a certain amount of caution, but Looking Glass' past campaigns and the company's commitment to start shipping Musubi in June does suggest it’s confident the frame will be released without issues.
This article originally appeared on Engadget at https://www.engadget.com/ar-vr/looking-glass-musubi-showcases-its-holographic-display-in-a-consumer-friendly-package-130000304.html?src=rss
Anthropic has launched a new research initiative called Anthropic Institute and has revealed that its Public Policy team is opening its first office in Washington, DC this spring. The company has made the announcement just a couple of days after it sued the US government to challenge the supply chain risk designation it received from the Defense Department. As Axiosnotes, Anthropic is tripling its Public Policy team at a time when AI companies are establishing a presence in Washington, so that they can influence future policies around artificial intelligence. In Anthropic’s case, it might have to find a way to be re-accepted by the US government first after President Trump ordered federal agencies to stop using its technology.
Sarah Heck, who joined the company as Head of External Affairs, will take over from co-founder Jack Clark as Head of Policy. Meanwhile, Clark has taken the role as Head of Public Benefit and will lead the Anthropic Institute. The company explains that the institute’s role is to “tell the world” what it learns about the challenges that arise as AI firms develop more advanced AI systems. Examples include how powerful AI technologies will reshape jobs and economies and what kinds of threats they’ll magnify or introduce.
The institute will bring together and expand Anthropic’s current research teams: The Frontier Red Team that stress-tests AI systems, the Societal Impacts team that looks at how AI is used in the real world, and the Economic Research team that tracks AI’s impact on jobs and the larger economy. Anthropic has hired Matt Botvinick, a former Senior Director of Research at Google DeepMind, and Zoë Hitzig, who studied AI’s social and economic impacts at OpenAI, to be founding members of the Institute.
This article originally appeared on Engadget at https://www.engadget.com/ai/anthropic-is-opening-an-office-in-dc-while-battling-pentagon-in-court-115700127.html?src=rss
Meta announced new features today aimed at cracking down on scams perpetrated via its platforms. First, Meta is launching AI tools for identifying impersonator of brands and celebrities, as well as for detecting deceptive links, which should help it to quickly take down frauds. Second, it is adding new alerts to caution against interacting with a potentially fraudulent account. Facebook will roll out alerts for suspicious friend requests, WhatsApp is getting warnings for device linking requests, and Messenger will also issue warnings if an account seems suspect.
Finally, Meta is also continuing to expand its processes for advertiser verification. The company said it aims to have verified advertisers account for 90 percent of its ads revenue by the end of the year, up from the current share of 70 percent. Last year, Meta estimated that marketing for scams and banned products could have been responsible for 10 percent of its 2024 revenue.
The social media company has been ramping up its actions against scams, particularly those known as celeb bait. Last month, it sued three entities from Brazil and China that were behind scams that leveraged images and deepfakes of popular people to promote dubious products and investment schemes. Meta said today that over the course of 2025, it removed 159 million scam ads as well as 10.9 million Facebook and Instagram accounts tied to criminal scam centers.
This article originally appeared on Engadget at https://www.engadget.com/social-media/meta-rolls-out-new-features-for-scam-protection-110000173.html?src=rss
At the start of the year, Google brought a host of new Gemini-powered features, including built-in Nano Banana image generation, to Chrome. After debuting in the United States, those features are now making their way to Chrome users in Canada, India and New Zealand, with support for 50 additional in tow. Among the new languages Gemini in Chrome can now converse in are French, Gujarati, Hindi and Spanish.
To try out Gemini in Chrome, tap the sparkle icon at the top right of the interface. This will open the sidebar interface Google introduced in January. From there, you can chat with the company's Gemini chatbot without the need to switch tabs. From the sidebar, you can also access Google's in-house image generator. Additionally, Gemini in Chrome offers integrations with Gmail, Maps, Calendar, YouTube and other Google apps. If you live outside Canada, India or New Zealand, Google says it will make Gemini in Chrome available in more countries and languages throughout the rest of 2026. Oh, and if don’t want to use Gemini in Chrome, you can right click on the sparkle icon and select unpin to never see it again.
This article originally appeared on Engadget at https://www.engadget.com/ai/google-starts-rolling-out-gemini-in-chrome-to-users-in-canada-india-and-new-zealand-023000528.html?src=rss
The inspector general's office of the Social Security Administration is investigating allegations of a security breach by a member of the so-called Department of Government Efficiency operation spearheaded by Elon Musk. A whistleblower has claimed that a former software engineer from DOGE said he possessed two databases from the SSA, "Numident" and the "Master Death File." The person reportedly asked for help transferring the databases from a thumb drive "to his personal computer so that he could ‘sanitize’ the data before using it at [the company]," an unnamed government contractor where he is currently employed. Those databases include personal information about more than 500 million living and deceased Americans.
The Washington Post reported that the whistleblower complaint was filed with the inspector general in January. "When The Post contacted the agency and the company in January, both said they had not heard of the complaint. Both said they subsequently looked into the allegations and did not find evidence to confirm the claims," the publication said. It is unclear why the complaint is now being investigated and neither party offered comment this week for ThePost's article. The SSA watchdog informed both members of Congress and the Government Accountability Office of its investigation.
These allegations follow a different whistleblower complaint filed last August about DOGE access and mishandling of data from the SSA. Charles Borges, former chief data officer at the agency, claimed that a SSA database was stored in an unsecured cloud environment. "This is absolutely the worst-case scenario," Borges told The Post of the latest claims. "There could be one or a million copies of it, and we will never know now."
This article originally appeared on Engadget at https://www.engadget.com/cybersecurity/social-security-watchdog-investigating-claims-that-doge-engineer-copied-its-databases-212722061.html?src=rss
