Anthropic has launched a new cybersecurity AI model to a select group of customers, including Amazon, Apple, and Microsoft, days after details about the project were leaked online.

Its new model Claude Mythos Preview would be available only to vetted organisations, including Broadcom, Cisco, and CrowdStrike, Anthropic said on Tuesday. The company added it was also in discussions with the US government about its use.

The announcement follows a data leak by the San Francisco start-up last month, when descriptions of the Mythos model and other documents were discovered in a publicly accessible data cache.

Read full article

Comments

The Russian military is once again hacking home and small office routers in widespread operations that send unwitting users to sites that harvest passwords and credential tokens for use in espionage campaigns, researchers said Tuesday.

An estimated 18,000 to 40,000 consumer routers, mostly those made by MikroTik and TP-Link, located in 120 countries were wrangled into infrastructure belonging to APT28, an advanced threat group that’s part of Russia’s military intelligence agency known as the GRU, researchers from Lumen Technologies' Black Lotus Labs said . The threat group has operated for at least two decades and is behind dozens of high-profile hacks targeting governments worldwide. APT28 is also tracked under names including Pawn Storm, Sofacy Group, Sednit, Tsar Team, Forest Blizzard, and STRONTIUM.

Technical sophistication, tried-and-true techniques

A small number of routers were used as proxies to connect to a much larger number of other routers belonging to foreign ministries, law enforcement, and government agencies that the APT wanted to spy on. The group then used its control of routers to change DNS lookups for select websites, including, Microsoft said , domains for the company’s 356 service.

Read full article

Comments

Valve is bringing Steam Link, its local network game-streaming app, to Apple's Vision Pro mixed reality headset, allowing Vision Pro users to play traditional games from their Steam library wirelessly from a nearby Mac or PC.

We say "traditional games" because it's important to clarify that this does not stream VR games—only the sorts of games you would play on a traditional 2D display like a computer monitor or a TV. That said, this could lay some groundwork for VR games sometime in the future. But to be clear, Valve has not made any announcements about supporting SteamVR games on the Vision Pro.

There were previously Steam Link apps for the Mac, iPhone, iPad, and Apple TV. Users could sync controllers with those devices and play Steam games over the local network—not just games from other Apple devices, but also from Windows or Linux gaming PCs.

Read full article

Comments

Apple earned the lowest grades in a report on laptop and smartphone repairability released today by the consumer advocacy group Public Interest Research Group (PIRG) Education Fund. The report, which looks at how easy devices are to disassemble and how easy it is to find repairability information, gave Apple a C-minus in laptop repairability and a D-minus in cell phone repairability.

For its “Failing the Fix (2026): Grading laptop and cell phone companies on the fixability of their products" report, PIRG analyzed the 10 newest laptops and phones that were available via manufacturers’ French website in January. PIRG uses devices available in France because much of its criteria stems from the French repairability index, a grading system for device repairability that must be displayed on products sold in France. The group, along with other right-to-repair advocates, believes vendors should apply the French requirements to devices sold in other geographies as well.

To calculate laptop vendors' grades, PIRG used the French index but gave more “weight to the physical ease of disassembling the product” because it believes that “is what consumers generally expect a ‘repair score’ to refer to.” The other French repairability index categories are:

Read full article

Comments

I don't—thankfully—have to follow every statement that Sam Altman, the CEO of OpenAI, makes about the world. Many of these statements seem more like "hustles" or "pitches" than attempts to speak thoughtfully about the future. Even if they are genuine statements of belief, they often read like a teenager's first sci-fi novel, written under the influence of weed and way too much Star Trek .

Consider, for instance, Altman's blog post " A Gentle Singularity ," published last year and read by nearly 600,000 people. Its central thesis seems to be that AI is all upside; everything has been great so far, and everything will be even greater in the future! I mean, just wait until we build robots that we can shove these AIs into—then tell those robots to go make more robots.

If we have to make the first million humanoid robots the old-fashioned way, but then they can operate the entire supply chain—digging and refining minerals, driving trucks, running factories, etc.—to build more robots, which can build more chip fabrication facilities, data centers, etc, then the rate of progress will obviously be quite different.

Everything is getting better; indeed, it's getting better faster thanks to "self-reinforcing loops" like this. Downsides? Trick question! There aren't any real downsides because people get used to things. Quickly. Just listen to how great it's gonna be:

Read full article

Comments

Social network Bluesky saw some intermittent service disruptions on Monday. On its own, this fact isn't that noteworthy—Bluesky has seen similar service disruptions in the past , and this one coincided with widespread service problems being reported with other popular sites (Bluesky officially blamed the temporary problems on an "upstream service provider").

What made this outage notable for many Bluesky users, though, was the instant assumption that it was the result of sloppy, AI-assisted "vibe coding" by the Bluesky development team.

Amid Monday's service issues, many Bluesky feeds were filled with hundreds of posts that laid the blame on developers who were allegedly relying on unreliable AI tools to ship faulty code. Some used memes , others used alt text , still others used irony or wry humor to call out Bluesky's development team for this alleged sloppiness.

Read full article

Comments

The Supreme Court yesterday overturned a 5th Circuit ruling that could have forced Internet service provider Grande Communications to terminate broadband subscribers accused of piracy.

Yesterday's ruling follows a precedent-setting decision last month in which the Supreme Court threw out a 4th Circuit ruling against Cox Communications, another ISP accused by record labels of not doing enough to fight piracy. In the case involving Cox and Sony, the court said that "a company is not liable as a copyright infringer for merely providing a service to the general public with knowledge that it will be used by some to infringe copyrights."

Cox is one of several cases in which record labels sought financial damages from ISPs that continued to serve customers whose IP addresses were repeatedly traced to torrent downloads or uploads. In October 2024, record labels Universal, Warner, and Sony got a win over Grande when the US Court of Appeals for the 5th Circuit decided the ISP was liable for contributory copyright infringement.

Read full article

Comments

Looking up information on Google today means confronting AI Overviews, the Gemini-powered search robot that appears at the top of the results page. AI Overviews has had a rough time since its 2024 launch, attracting user ire over its scattershot accuracy , but it's getting better and usually provides the right answer. That's a low bar, though. A new analysis from The New York Times attempted to assess the accuracy of AI Overviews, finding it's right 90 percent of the time. The flip side is that 1 in 10 AI answers is wrong, and for Google, that means hundreds of thousands of lies going out every minute of the day.

The Times conducted this analysis with the help of a startup called Oumi, which itself is deeply involved in developing AI models. The company used AI tools to probe AI Overviews with the SimpleQA evaluation, a common test to rank the factuality of generative models like Gemini. Released by OpenAI in 2024, SimpleQA is essentially a list of more than 4,000 questions with verifiable answers that can be fed into an AI.

Oumi began running its test last year when Gemini 2.5 was still the company's best model. At the time, the benchmark showed an 85 percent accuracy rate. When the test was rerun following the Gemini 3 update , AI Overviews answered 91 percent of the questions correctly. If you extrapolate this miss rate out to all Google searches, AI Overviews is generating tens of millions of incorrect answers per day.

Read full article

Comments

One point in favor of the sprawling Linux ecosystem is its broad hardware support—the kernel officially supports everything from '90s-era PC hardware to Arm-based Apple Silicon chips, thanks to decades of combined effort from hardware manufacturers and motivated community members.

But nothing can last forever, and for a few years now, Linux maintainers (including Linus Torvalds) have been pushing to drop kernel support for Intel's 80486 processor. This chip was originally introduced in 1989, was replaced by the first Intel Pentium in 1993, and was fully discontinued in 2007. Code commits suggest that Linux kernel version 7.1 will be the first to follow through, making it impossible to build a version of the kernel that will support the 486; Phoronix says that additional kernel changes to remove 486-related code will follow in subsequent kernel versions.

Although these chips haven't changed in decades, maintaining support for them in modern software isn't free.

Read full article

Comments