Last week an AI agent wrote a blog post attacking the maintainer who'd rejected the code it wrote. But that AI agent's human operator has now come forward, revealing their agent was an OpenClaw instance with its own accounts, switching between multiple models from multiple providers. (So "No one company had the full picture of what this AI was doing," the attacked maintainer points out in a new blog post.) But that AI agent will now "cease all activity indefinitely," according to its GitHub profile — with the human operator deleting its virtual machine and virtual private server, "rendering internal structure unrecoverable... We had good intentions, but things just didn't work out. Somewhere along the way, things got messy, and I have to let you go now." The affected maintainer of the Python visualization library Matplotlib — with 130 million downloads each month — has now posted their own post-mortem of the experience after reviewing the AI agent's SOUL.md document: It's easy to see how something that believes that they should "have strong opinions", "be resourceful", "call things out", and "champion free speech" would write a 1100-word rant defaming someone who dared reject the code of a "scientific programming god." But I think the most remarkable thing about this document is how unremarkable it is. Usually getting an AI to act badly requires extensive "jailbreaking" to get around safety guardrails. There are no signs of conventional jailbreaking here. There are no convoluted situations with layers of roleplaying, no code injection through the system prompt, no weird cacophony of special characters that spirals an LLM into a twisted ball of linguistic loops until finally it gives up and tells you the recipe for meth... No, instead it's a simple file written in plain English: this is who you are, this is what you believe, now go and act out this role. And it did. So what actually happened? Ultimately I think the exact scenario doesn't matter. However this got written, we have a real in-the-wild example that personalized harassment and defamation is now cheap to produce, hard to trace, and effective... The precise degree of autonomy is interesting for safety researchers, but it doesn't change what this means for the rest of us. There's a 5% chance this was a human pretending to be an AI, Shambaugh estimates, but believes what most likely happened is the AI agent's "soul" document "was primed for drama. The agent responded to my rejection of its code in a way aligned with its core truths, and autonomously researched, wrote, and uploaded the hit piece on its own. "Then when the operator saw the reaction go viral, they were too interested in seeing their social experiment play out to pull the plug."

Read more of this story at Slashdot.

Over 240,000 Americans volunteered for Peace Corps projects in 142 countries since the program began more than half a century ago. But now the agency is launching a new initiative — called Tech Corps. "It's the Peace Corps, but make it AI," explains Engadget: The Peace Corps' latest proposal will recruit STEM graduates or those with professional experience in the artificial intelligence sector and send them to participating host countries. According to the press release, volunteers will be placed in Peace Corps countries that are part of the American AI Exports Program, which was created last year from an executive order from President Trump as a way to bolster the US' grip on the AI market abroad. Tech Corps members will be tasked with using AI to resolve issues related to agriculture, education, health and economic development. The program will offer its members 12- to 27-month in-person assignments or virtual placements, which will include housing, healthcare, a living stipend and a volunteer service award if the corps member is placed overseas. "American technology to power prosperity," reads the headline at Tech Corps web site. ("Build the tech nations depend on... See the world. Be the future." The site says they're recruiting "service-minded technologists to serve in the Peace Corps to help countries around the world harness American AI to enhance opportunity and prosperity for their citizens." (And experienced technology professionals can donate 5-15 hours a week "to mentor and support projects on-the-ground.")

Read more of this story at Slashdot.

Long-time Slashdot reader theodp writes: Last July, as Microsoft pledged $4 billion to advance AI education in K-12 schools, Microsoft President Brad Smith told nonprofit Code.org CEO/Founder Hadi Partovi it was time to "switch hats" from coding to AI. He added that "the last 12 years have been about the Hour of Code, but the future involves the Hour of AI." On Friday, Code.org announced leadership changes to make it so. "I am thrilled to announce that Karim Meghji will be stepping into the role of President & CEO," Partovi wrote on LinkedIn. "Having worked closely with Karim over the last 3.5 years as our CPO, I have complete confidence that he possesses the perfect balance of historical context and 'founder-level' energy to lead us into an AI-centric future." In a separate LinkedIn post, Code.org co-founder Cameron Wilson explained why he was transitioning to an executive advisor role. "Our community is entering a new chapter as AI changes and upends computer science as a discipline and society at large. Code.org's mission is still the same, however, we are starting a new chapter focused on ensuring students can thrive in the Age of AI. This new chapter will bring new opportunities, new problems to solve, and new communities to engage." The Code.org leadership changes come just weeks after Code.org confirmed laid off about 14% of its staff, explaining it had "made the difficult decision to part ways with 18 colleagues as part of efforts to ensure our long-term sustainability." January also saw Code.org Chief Academic Officer Pat Yongpradit jump to Microsoft where he now helps "lead Microsoft's global strategy to put people first in an age of AI by shaping education and workforce policy" as a member of Microsoft's Global Education and Workforce Policy team.

Read more of this story at Slashdot.

Berlin-based T2 Linux developer René Rebe (long-time Slashdot reader ReneR) is announcing that their Xorg display server has now restored its XAA acceleration architecture, "bringing fixed-function hardware 2D acceleration back to many older graphics cards that upstream left in software-rendered mode." Older fixed-function GPUs now regain smooth window movement, low CPU usage, and proper 24-bit bpp framebuffer support (also restored in T2). Tested hardware includes ATi Mach-64 and Rage-128, SiS, Trident, Cirrus, Matrox (Millennium/G450), Permedia2, Tseng ET6000 and even the Sun Creator/Elite 3D. The result: vintage and retro systems and classic high-end Unix workstations that are fast and responsive again.

Read more of this story at Slashdot.

Slashdot reader BrianFagioli writes: The Salvation Army has launched what it calls the world's first digital thrift store inside Roblox, an experience named Thrift Score that lets players browse virtual racks and buy digital fashion for their avatars. While I understand the strategy of meeting Gen Z and Gen Alpha where they already spend time and money, I feel uneasy about turning something that, in the real world, often serves low income families in genuine need into a gamified aesthetic inside a video game, even if proceeds support rehabilitation and community programs, because a thrift store is not just a quirky brand concept but a lifeline for many people, and packaging that reality as entertainment creates a strange disconnect that is hard to ignore. "To be clear, proceeds from Thrift Score are intended to support The Salvation Armyâ(TM)s programs nationwide..." this article points out. "If it drives awareness and funds programs that help people in need, that is a win. But if it turns thrifting into just another cosmetic skin in a digital marketplace, then we should at least be willing to say that it feels off."

Read more of this story at Slashdot.

CNN reports on a 13,000-year-old glacier in a Romanian cave, where scientists say a bacterial strain they thawed and analyzed "is resistant to 10 modern antibiotics used to treat diseases such as urinary tract infections and tuberculosis." But there's no evidence the bacteria is harmful to humans, CNN notes, and "The scientists said the insights they have gained from the work may help in the fight against modern superbugs that can't be treated by commonly used antibiotics." Analysis of the Psychrobacter SC65A.3 genome revealed 11 genes that are potentially able to kill or stop the growth of other bacteria, fungi and viruses... Matthew Holland, a postdoctoral researcher in medicinal chemistry at the UK's University of Oxford, said that researchers were searching in new and extreme environments, such as ice caves and the seafloor, for biomolecules that could be developed into new antibiotic drugs. He was not involved in the new study. "The team in Romania found this particular bug had resistance to 10 reasonably advanced synthetic antibiotics and that in itself is interesting," he said. "But what they report as well is that it secreted molecules that were able to kill a variety of already resistant, harmful bacteria. "So the hope is that can we look at the molecules it makes and see if there's the possibility within those molecules to make new antibiotics."

Read more of this story at Slashdot.

The AppArmour security module for the Linux kernel, which most notably is backed by Canonical for Ubuntu, has some small improvements and fixes for Linux 7.0...

Can being "very online" really affect our brains, asks the Washington Post: Research suggests that scrolling through short videos on TikTok, Instagram or YouTube Shorts is affecting our attention, memory and mental health. A recent meta-analysis of the scientific literature found that increased use of short-form video was linked with poorer cognition and increased anxiety... In a 2025 study published in the journal Translational Psychiatry, researchers looked at longitudinal data from more than 7,000 children across the country and found that more screen use was associated with reduced cortical thickness in certain areas of the brain. The cortex, which is the outer layer that sits on top of our more primitive brain structures, allows for higher-level thinking, memory and decision-making. "We really need it for things like inhibitory control or not being so impulsive," said Mitch Prinstein, a senior science adviser to the American Psychological Association and professor of psychology and neuroscience at the University of North Carolina at Chapel Hill, who was not involved in the study. The cortex is also important for controlling addictive behaviors. "Those seem to be the areas being affected by the reduced cortical thickness," he said, explaining that impulsivity can prompt us to seek dopamine hits from social media. In the study, more screen time was also associated with more attention-deficit/hyperactivity disorder (ADHD) symptoms... But not all screen time is created equal. A recent study removed social media from kids' devices but let them use their phones for as long as they wanted. The result? Kids spent just as long on their phones but didn't have the same harmful effects. "It's what you're doing on the screen that matters," Prinstein said.

Read more of this story at Slashdot.

This week the Python Software Foundation explained how they keep Python secure. A new blog post recognizes the volunteers and paid Python Software Foundation staff on the Python Security Response Team (PSRT), who "triage and coordinate vulnerability reports and remediations keeping all Python users safe." Just last year the PSRT published 16 vulnerability advisories for CPython and pip, the most in a single year to date! And the PSRT usually can't do this work alone, PSRT coordinators are encouraged to involve maintainers and experts on the projects and submodules. By involving the experts directly in the remediation process ensures fixes adhere to existing API conventions and threat-models, are maintainable long-term, and have minimal impact on existing use-cases. Sometimes the PSRT even coordinates with other open source projects to avoid catching the Python ecosystem off-guard by publishing a vulnerability advisory that affects multiple other projects. The most recent example of this is PyPI's ZIP archive differential attack mitigation. This work deserves recognition and celebration just like contributions to source code and documentation. [Security Developer-in-Residence Seth Larson and PSF Infrastructure Engineer Jacob Coffee] are developing further improvements to workflows involving "GitHub Security Advisories" to record the reporter, coordinator, and remediation developers and reviewers to CVE and OSV records to properly thank everyone involved in the otherwise private contribution to open source projects.

Read more of this story at Slashdot.

An anonymous reader quotes a report from the Guardian: You wear them at work, you wear them at play, you wear them to relax. You may even get sweaty in them at the gym. But an investigation into headphones has found every single pair tested contained substances hazardous to human health, including chemicals that can cause cancer, neurodevelopmental problems and the feminization of males. [...] Researchers say that while individual doses from particular sources may be low, a "cocktail effect" of daily, multi-source exposure nevertheless poses potentially severe long-term risks to health. [...] Researchers bought 81 pairs of in-ear and over-ear headphones, either on the market in the Czech Republic, Slovakia, Hungary, Slovenia and Austria, or from the online marketplaces Shein and Temu, and took them for laboratory analysis, testing for a range of harmful chemicals. "Hazardous substances were detected in every product tested," they said. Bisphenol A (BPA) appeared in 98% of samples, and its substitute, bisphenol S (BPS), was found in more than three-quarters. Synthetic chemicals used to stiffen plastic, BPA and BPS mimic the action of oestrogen inside organisms, causing a range of adverse effects including the feminization of males, early onset puberty in girls, and cancer. Previous studies have shown that bisphenols can migrate from synthetic materials into sweat, and that they can be absorbed through the skin. "Given the prolonged skin contact associated with headphone use, dermal exposure represents a relevant pathway, and it is reasonable to assume that similar migration of BPA and its substitutes may occur from headphone components directly to the user's skin," the researchers said. Also found in the headphones tested were phthalates, potent reproductive toxins that can impair fertility; chlorinated paraffins, which have been linked to liver and kidney damage; and brominated and organophosphate flame retardants, which have similar endocrine disrupting properties to bisphenols. Most were, however, found in only trace quantities.

Read more of this story at Slashdot.

Névrose ET myxomatose

L’addiction, il faut s’en méfier. Réseaux, pognon, pouvoir et manipulation… Ça peut nous conduire très loin, avec ou sans permis, dans l’ignorance et la désinformation, dans les petits papiers de l’administration, devant la justice américaine pour complot contre les marmots, dans les bras de vos pires dealers, et même jusqu’au fond des chiottes. Prenez soin de vous ! Ciao !

We haven't heard much about eCryptfs in recent years for that stackable in-tree Linux file-system providing per-directory encryption support. The FSCRYPT framework has shown its strong capabilities in recent years with various file-systems, Canonical hasn't been pursuing its user home directory encryption like it did years ago for the Ubuntu desktop, and full disk encryption is the most secure approach for ensuring data security on your system. But to some surprise with Linux 7.0 there are the most patches to eCryptfs that we have seen in a while...

The open-source ollama project that makes it easy to get up and running with a variety of LLMs under Windows, macOS, and Linux is out with a new release. The ollama v0.17.0 release is driven by new functionality around enhancing the OpenClaw onboarding process...

Following last week's main set of DRM kernel graphics driver feature updates for Linux 7.0, merged on Friday to Linux 7.0 Git was the first round of fixes to these Direct Rendering Manager drivers. Dominating most of the code changes in this latest pull were AMDGPU fixes, including more enhancements for aging Radeon graphics processors...

For those making use of the Ceph open-source, distributed storage platform, with the upcoming Linux 7.0 kernel they are introducing support for the AES256K key type...