The National Reconnaissance Office, the agency overseeing the US government’s fleet of spy satellites, has declassified a decades-old program used to eavesdrop on the Soviet Union’s military communication signals.

The program was codenamed Jumpseat, and its existence was already public knowledge through leaks and contemporary media reports. What’s new is the NRO’s description of the program’s purpose, development, and pictures of the satellites themselves.

In a statement, the NRO called Jumpseat “the United States’ first-generation, highly elliptical orbit (HEO) signals-collection satellite.”

Read full article

Comments

The National Center for Missing and Exploited Children said it received more than 1 million reports of AI-related child sexual abuse material (CSAM) in 2025. The “vast majority” of that content was reported by Amazon, which found the material in its training data, according to an investigation by Bloomberg. In addition, Amazon said only that it obtained the inappropriate content from external sources used to train its AI services and claimed it could not provide any further details about where the CSAM came from. 

“This is really an outlier,” Fallon McNulty, executive director of NCMEC’s CyberTipline, told Bloomberg. The CyberTipline is where many types of US-based companies are legally required to report suspected CSAM. “Having such a high volume come in throughout the year begs a lot of questions about where the data is coming from, and what safeguards have been put in place.” She added that aside from Amazon, the AI-related reports the organization received from other companies last year included actionable data that it could pass along to law enforcement for next steps. Since Amazon isn’t disclosing sources, McNulty said its reports have proved “inactionable.”

“We take a deliberately cautious approach to scanning foundation model training data, including data from the public web, to identify and remove known [child sexual abuse material] and protect our customers,” an Amazon representative said in a statement to Bloomberg. The spokesperson also said that Amazon aimed to over-report its figures to NCMEC in order to avoid missing any cases. The company said that it removed the suspected CSAM content before feeding training data into its AI models. 

Safety questions for minors have emerged as a critical concern for the artificial intelligence industry in recent months. CSAM has skyrocketed in NCMEC’s records; compared with the more than 1 million AI-related reports the organization received last year, the 2024 total was 67,000 reports while 2023 only saw 4,700 reports. 

In addition to issues such as abusive content being used to train models, AI chatbots have also been implicated in several dangerous or tragic cases involving young users. OpenAI and Character.AI have both been sued after teenagers planned their suicides with those companies’ platforms. Meta is also being sued for alleged failures to protect teen users from sexually explicit conversations with chatbots.

This article originally appeared on Engadget at https://www.engadget.com/ai/amazon-discovered-a-high-volume-of-csam-in-its-ai-training-data-but-isnt-saying-where-it-came-from-224749228.html?src=rss

Complaining about Windows 11 is a popular sport among tech enthusiasts on the Internet, whether you’re publicly switching to Linux, publishing guides about the dozens of things you need to do to make the OS less annoying, or getting upset because you were asked to sign in to an app after clicking a sign-in button.

Despite the negativity surrounding the current version of Windows, it remains the most widely used operating system on the world’s desktop and laptop computers, and people usually prefer to stick to what they’re used to. As a result, Windows 11 has just cleared a big milestone—Microsoft CEO Satya Nadella said on the company’s most recent earnings call (via The Verge) that Windows 11 now has over 1 billion users worldwide.

Windows 11 also reached that milestone just a few months quicker than Windows 10 did—1,576 days after its initial public launch on October 5, 2021. Windows 10 took 1,692 days to reach the same milestone, based on its July 29, 2015, general availability date and Microsoft’s announcement on March 16, 2020.

Read full article

Comments

NVIDIA has officially launched a native GeForce NOW client for Linux as a Flatpak, giving Linux gamers access to cloud-rendered RTX gaming. Phoronix reports: While confined to a Flatpak, for now NVIDIA is just “officially” supporting it on Ubuntu 24.04 LTS and later. Granted, thanks to Flatpak it should run on other non-Ubuntu distributions too but in terms of the official support and where they are qualifying their builds they are limiting it just to Ubuntu 24.04 LTS and later. […] At launch the Flatpak build is also just for x86_64 Linux with no AArch64 Linux builds or similar at this time.

Running GeForce NOW on Linux while games are rendered in NVIDIA’s cloud with Blackwell GPUs, you still need to be using a modern GPU with H.264 or H.265 Vulkan Video support NVIDIA isn’t yet supporting Vulkan Video AV1 with GeForce NOW on Linux but just H.264/H.265. If you are using NVIDIA graphics the NVIDIA R580 series or newer is recommended while using the X.Org session. If you are using Intel or AMD Radeon graphics, Mesa 24.2+ is recommended and using the Wayland session.

When you are up and running with GeForce NOW on Linux, you have access to over 4,500 games. The free tier of GeForce NOW provides standard access to the gaming servers and limited session caps for an introductory-level experience. It’s with the performance tier where you can enjoy RTX ray-tracing and 1440p @ 60 FPS performance and up to six hour sessions. With GeForce NOW’s Ultimate tier is where you are running on GeForce RTX 5080 GPU servers with support for up to 5K @ 120 FPS gaming or 1080p @ 360 FPS with up to eight hour gaming sessions in length.

Read more of this story at Slashdot.

Because printing money is easy for some companies, LEGO has announced an official Lord Of The Rings Sauron’s Helmet build set. The kit will be released March 1st for $70 and include 538 pieces, including a Sauron minfig holding the One Ring. Honestly, $70 is still expensive, but after all the $200 and $300 sets they’ve been releasing lately it almost seems like a steal. It isn’t though, just to be clear. When completed (provided you didn’t eat any crucial pieces), the set makes a 13″ x 5″ replica of Sauron’s helmet. And what a helmet it is! Has anybody ever cosplayed as sloppy Sauron and made a Sauron beer helmet? Because that’s a convention-stealing cosplay idea right there. If you use it though I get half the costume contest prize money.

At this point, we’ve all heard plenty of stories about AI chatbots leading users to harmful actions, harmful beliefs, or simply incorrect information. Despite the prevalence of these stories, though, it’s hard to know just how often users are being manipulated. Are these tales of AI harms anecdotal outliers or signs of a frighteningly common problem?

Anthropic took a stab at answer ingthat question this week, releasing a paper studying the potential for what it calls “disempowering patterns” across 1.5 million anonymized real-world conversations with its Claude AI model. While the results show that these kinds of manipulative patterns are relatively rare as a percentage of all AI conversations, they still represent a potentially large problem on an absolute basis.

A rare but growing problem

In the newly published paper “Who’s in Charge? Disempowerment Patterns in Real-World LLM Usage,” researchers from Anthropic and the University of Toronto try to quantify the potential for a specific set of “user disempowering” harms by identifying three primary ways that a chatbot can negatively impact a user’s thoughts or actions:

Read full article

Comments

Science fiction and science leaders alike have warned us that artificial intelligence may one day take over the world, but until those predictions come to pass, generative AI’s biggest impact on my life has been overloading my social media feeds with slop. It seems I can’t open TikTok, Instagram, or YouTube without running smack into bizarre and troubling AI concoctions featuring babies in danger and cats having affairs. It really is the wild west (or maybe Westworld) out there.

I think few among us really believe these videos are any good, and it’s pretty obvious they aren’t good for us, or for the world. Short-form video is already numbing enough, but this AI content is generally completely devoid of any meaning or substance. And yet, it’s everywhere. I haven’t spent too much time on YouTube Shorts recently, but in my limited experience, the feed has been chock full of AI, especially if I’m logged out of my personal account.

Still, if you’re a dedicated YouTube Shorts user (or a frequent YouTube user in general) you might have noticed something odd in recent days: There don’t seem to be quite as many AI videos on the platform right now. There are still a lot, don’t get me wrong, but it turns out YouTube has recently taken action to remove some of its AI content—the sloppiest of the slop.

YouTube’s war on AI slop

Android Police spotted the development on Wednesday, basing its findings on a November report from Kapwing, a company that develops an online video editor. Kapwing investigated AI slop across YouTube’s vast content library, noting the top 100 most-subscribed YouTube channels that publish this sort of AI content. In the two months since that report, Android Police noticed that 16 of those 100 channels are no longer with us.

That includes the most popular AI channel on YouTube, at least according to Kapwing. “CuentosFacianantes” had 5.95 million subscribers at the time of their initial report, and produced AI-generated shorts inspired by Dragon Ball. The channel had amassed roughly 1.28 billion views by the end of last year; despite launching in 2020, it had curated its library to begin Jan. 8, 2025, so those numbers were racked up pretty recently. The number two channel, “Imperio de Jesus” with 5.87 million subscribers, and the number seven channel “Super Cat League,” with 4.21 million subscribers, were also shut down.

According to Android Police, the 16 channels in question had a total of 35 million subscribers and over 4.7 billion views across their collective videos. Some of these channels are completely gone, while others simply have had their videos removed.

Why is YouTube removing AI slop?

YouTube CEO Neal Mohan published a post on Jan. 21 of this year describing the company’s vision for 2026. Towards the end of that letter, he acknowledges AI content, predicting that, “AI will be a boon to the creatives who are ready to lean in,” and comparing it to tools like Photoshop and CGI, adding “AI will remain a tool for expression, not a replacement.” However, Mohan was also critical of the technology, noting that it’s becoming more difficult to tell real videos from AI. He notes that YouTube is now removing “any harmful synthetic media that violates our Community Guidelines,” and is giving creators tools to help identify and block deepfakes.

More interestingly, the letter includes a section labeled “Managing AI slop,” which is the first time I’ve seen a company like YouTube use that expression. Mohan says that YouTube’s goal is to be a place where free expression thrives, but also a place “where people feel good spending their time.” To that point, he says, “To reduce the spread of low quality AI content, we’re actively building on our established systems that have been very successful in combatting spam and clickbait, and reducing the spread of low quality, repetitive content.”

Mohan doesn’t call out any accounts by name, nor does he acknowledge the accounts and content the company has already deleted, but it’s a clear line in the sand: YouTube is not against AI-generated content, but it will remove low-quality AI content it feels is, well, slop. That’s good news for anyone who uses YouTube (so, pretty much everyone), even if it’s far from a cure for the growing problem.

I’ve reached out to YouTube for comment on this story, and will update this piece if I hear back.

The preprint repository arXiv will require all submissions to be written in English or accompanied by a full English translation starting February 11, a policy change that explicitly permits the use of AI translators even as research suggests large language models remain inconsistent at the task.

Until now, authors only needed to submit an abstract in English. ArXiv hosts nearly 3 million preprints and receives more than 20,000 submissions monthly, though just 1% are in languages other than English.

Ralph Wijers, chair of arXiv’s editorial advisory council, advises authors to verify any AI-generated translations. “Our own experience is that AI translation is good but not good enough,” he says. A 2025 study from ByteDance Seed and Peking University ranked 20 LLMs on translation quality between Chinese and English; GPT-5-high scored nearly 77, just below the human expert benchmark of 80, but most models including GPT-4o, Claude 4, and Deepseek-V3 scored under 60.

Read more of this story at Slashdot.

Two Elon Musk companies are reportedly planning to merge. On Thursday, Reuters reported that SpaceX and xAI are holding merger talks ahead of a planned IPO. Part of their plan is to launch AI data centers into space (but unfortunately, only as far as Earth’s orbit).

Last week, it was reported that Musk planned to take SpaceX public despite having once said it wouldn’t happen until the company had a presence on Mars. Now, the IPO could happen as early as this year. Shares of xAI would reportedly be exchanged for shares in SpaceX under the merger. Reuters reports that two entities were set up in Nevada on January 21 to facilitate the deal.

If the idea of two Musk companies becoming one sounds familiar, that’s because it happened less than a year ago. In March 2025, xAI bought X, putting Grok (known for nonconsensual “nudifying” images) and X (infamous for being a far-right hellscape) together under one unholy roof.

The latest idea Musk is pitching is blasting AI data centers off into space. At last week’s gathering of the rich and powerful in Davos, Switzerland, he said, “The lowest cost place to put AI will be in space. And that will be true within two years, maybe three at the latest.” The idea is that data centers in orbit could harness solar power and reduce cooling costs. However, industry analysts and executives consider it a risky bet, questioning whether the savings would warrant the massive investment. If or when the AI bubble bursts, the plan could go down in flames — if not literally, then figuratively.

This article originally appeared on Engadget at https://www.engadget.com/big-tech/elon-musks-spacex-and-xai-are-reportedly-holding-merger-talks-211740150.html?src=rss

In just the kind of news I like to hear, this is a video documenting the good work of 12-year old Roman McKonn, who’s helped rehome over 4,800 shelter dogs through transporting, fostering, and making videos with the dogs. Great job, Roman. He’s also very well spoken for a 12-year old. Heck, he’s well spoken for an adult, because most adults I know can only communicate through text. You see them in real life and they’re all elbows when they can’t use emojis. If I could take care of all the dogs I would. I daydream about acres of land and dogs as far as the eye can see living their best lives. Absent from the dream? “Other people.” You’ve dreamt it too!