Can we get confirmation or denial about this?
XFEDI
ActivityPub#ai
Public posts with this hashtag.
A former OpenAI engineer accused of spying has a folder labelled "Saudi Gov Docs" on his phone, the FBI says. US prosecutors argue he may flee to Saudi Arabia as the kingdom pursues AI dominance. https://arstechnica.com/tech-policy/2026/10/accused-openai-spy-had-folder-called-saudi-gov-docs-fbi-says/ #AIagent #AI #GenAI #AIGovernance
Ars TechnicaAccused OpenAI spy had folder called “Saudi Gov Docs,” FBI saysUS fears Saudi government will help ex-OpenAI engineer flee spying charges.
When we start outsourcing thinking to AI and allow to lose accountability, bad things happen.
On July 14, 2025, Angela Lipps was babysitting in her Tennessee home when police arrested her for fraud. She was held in a local county jail until October, then moved to a Fargo jail until December 24, when a judge dismissed the charges. She was released into a frigid North Dakota winter with only the summer clothing she had been wearing when arrested in Tennessee.
It turned out she had proof she was in Tennessee when the bank fraud happened in North Dakota. An investigation prior to arrest would have found that out, but that was skipped in favor of rapid advice from a facial recognition program using AI to sort through photographs scraped from the internet.
Police offered no apology for her five-month incarceration, and she had to rely on donations to make her way to Chicago.
Unfortunately, her story is not unique."
Also, I learned the word "agnotology".
Stanford Law ReviewAI Agnotology, Cognitive Surrender, and Policing Accountability | Stanford Law ReviewLaw enforcement agencies have increasingly employed artificial intelligence (AI) in their investigatory methods and technology. This usage, however, r
The 7 biggest issues CISOs and security teams face today
Security leaders face mounting pressure as AI-driven cyberattacks surge. 68% of CISOs report increased stress, with 73% unprepared for major breaches. The shift from prevention to resilience is critical—78% view generative AI as a security risk. Key priorities: close the 95% skills gap, strengthen AI infrastructure security, align programs with org needs, and maintain foundational practices.
CSO OnlineThe 7 biggest issues CISOs and security teams face todayFrom escalating AI-enabled threats to persistent skills gaps, security leaders are reshaping their agendas to address several key long-standing and emerging concerns.
US Tech Leaders Are Urgently Calling for Rules on AI—China Already Has Them, by (not found or attributed):
The Independent Uganda:US tech leaders are urgently calling for rules on AI – China already has them Beijing has already surrounded its AI sector with public obligations, regulatory oversight and an evolving framework of risk assessment COMMENT | MADELINE CARR | Recent warnings from American AI leaders have created the impression that the world is only beginning to confront the dangers of advanced artificial intelligence. Dario Amodei, chief executive of tech company …
i don't understand how #meta expects to gain customers if at the slightest try to get out of their invasive policies they lock your account and tell you to fuck yourself.
Despite Progress, AI Inner Workings Remain a Black Box
Author
: Copyright © 2026 Todd A. Jacobs
Posted
: 2026-10-19
Large language models are currently promoted to the public as forms of artificial intelligence. No one doubts that #AI and #LLM tools can be useful, but treating these technologies as "thinking machines" is a marketing ploy. Even the scientists and engineers who build today's frontier models will admit that they know how to build systems that simulate thinking, but that many key elements of how they work remain unexplained outcomes rather than the result of deliberate design choices.
Scientists and engineers call observable but unexplained behaviors of a system "emergent properties." A useful comparison is seen in ant colonies: ants search for resources in individual and arguably inefficient random patterns, but then act collectively to gather any resources found to efficiently carry them back to the anthill once found. The apparent efficiency of ants' resource gathering is thus an emergent property that results from merging random search patterns with scent trails and an instinct that drives each ant to follow the most-used and therefore strongest path left by the rest of the colony.
Ben Patterson of PCWorld explains Anthropic's own theories on this issue in his July, 2026 article.
We still don’t know much about how AI 'thinks.' Give ChatGPT, Claude, or Gemini a prompt, and they’ll spit out an answer. Much of what happens in between remains shrouded in mystery.
However, Patterson explains that researchers at Anthropic have discovered an emergent property in their Claude models: an unplanned feature that Anthropic calls a Jacobian workspace ("J-space").
What’s weird about this workspace...is that Anthropic didn’t actually design it. Instead, the J-space appears to have 'emerged on its own' during Claude’s training[.]
Like our ant colony, #J_space wasn't something deliberately designed into the system by Anthropic. It's something they discovered while researching how their models process information. To their credit, Anthropic's researchers are actively theorizing and exploring this process, as the only sure thing is that LLMs and other AI systems process information differently than people do.
Even models that provide chain-of-thought (#CoT) traces don't technically meet the definition of #Explainable_AI (#XAI). Instead, they transform their internal black-box processes into words using a language model. Despite being promoted as a visible thought process that accurately describes what the model is doing, AI chains of thought are best thought of as proxies that can't yet be audited end-to-end. The statistical nature of LLM conversion to natural language makes these proxied chains of thought inherently unreliable for evaluating the underlying system processes, no matter how convincing the user-visible chains of thought appear or how well-reasoned they sound.
PCWorldThe more we learn about how AI ‘thinks,’ the weirder it getsResearchers found a “workspace” in Claude where concepts it might never say aloud can silently “light up.” The weird thing about the workspace? Anthropic didn’t design it.
Credit stress is threatening to unwind the #AI bull. All eyes are on Oracle
CNBCCredit stress is threatening to unwind the AI bull. All eyes are on Oracle to hold the lineInvestors are pushing up the cost of insurance against an Oracle default.
Anthropic’s Claude Designs Leaner Quantum Circuits, Reducing Resources Needed for Key Calculations
#ai #quantum #computing
The Quantum InsiderAnthropic’s Claude Designs Leaner Quantum Circuits, Reducing Resources Needed for Key CalculationsResearchers used Claude to design more efficient quantum circuits across eight tasks, reducing estimated computing resources.
Nvidia has released new open-source models designed to run AI agents on enterprise infrastructure and personal computers, enabling companies to deploy private on-device AI assistants without relying on cloud services. The move signals a major push for local AI agent deployment. https://techcrunch.com/2026/10/19/nvidia-launches-new-open-models-to-run-agents-in-the-enterprise-and-on-pcs/ #AIagent #AI #GenAI
TechCrunchNvidia launches new open models to run agents in the enterprise and on PCs | TechCrunchNvidia is leaning hard into the trend of enterprise companies wanting to run their own open models in their own data centers and also trying to make on-device, private AI agents a thing.
RE: https://openbiblio.social/@hochstenbach/117466185547617415
👇 More evidence that a lot of work will be required to achieve the @coar_repositories vision (which I have long embraced) to make repositories a fundamental building block of the #scholcomm system. While progress is being made e.g. with adoption of the Notify protocol and #FAIR Signposting, recent findings by @g3om4c and the below by @hochstenbach (with focus on how repositories deal with machine access in the age of over-agressieve #AI bots), suggest a “1 Step Forward, 3 Steps Back” pattern.
In the age of genAI, we all know about “bad bots” that are hitting our institutional repositories and websites. But, what about the “honest bots”? How does our scholarly infrastructure react to them? For the next months I will generate on my blog a report of a “honest bot” that t
OpenBiblio.SocialPatrick Hochstenbach (@[email protected])In the age of genAI, we all know about “bad bots” that are hitting our institutional repositories and websites. But, what about the “honest bots”? How does our scholarly infrastructure react to them? For the next months I will generate on my blog a report of a “honest bot” that tries to access publication landing pages in European institutional repositories. Follow the reports and stats. Feedback
#ONLYOFFICE Cloud 10.0 is out, and yes, there's a new #AI Chat.
A few details:
✔️ It's optional. Skip it if it's not for you.
✔️ You choose the provider. No lock-in.
✔️ The editors work the same without it.
If you don't mind using AI tools for document editing in the cloud, you can check how it works.
More information: https://www.onlyoffice.com/blog/2026/10/cloud_ai_10?utm_source=social&utm_medium=post&utm_campaign=fosstodon
ONLYOFFICE Blog10 things to know about AI in ONLYOFFICE Cloud 10.0 | ONLYOFFICE Blog10 things to know about AI in ONLYOFFICE 10.0 cloud: how to activate it, pricing, available models, AI agents, file generation, MCP servers and more.
Great news, everyone!
#ONLYOFFICE Docs 10.0 is finally out 🚀
#AI tools in every editor, 50+ new features, 4 new modular apps, 1,000+ bug fixes and 35 fixed vulnerabilities in one big release.
Based on community feedback, AI is no longer built into the open-source versions. Add it separately only if you want it, and connect any model, including local ones.
Which update will you try first?
Discover all the details in this blog post: https://www.onlyoffice.com/blog/2026/10/onlyoffice-docs-10?utm_source=social&utm_medium=post&utm_campaign=fosstodon
ONLYOFFICE BlogMeet ONLYOFFICE Docs 10.0: AI tools and modular appsONLYOFFICE Docs 10.0 brings 50 new features, 1,000+ fixes, and 35 security updates to make work with docs, sheets, PDFs, and slides faster and safer.
📢 DEF CON Training is now accepting course submissions for 2027!
Whether you’re an experienced instructor or bringing a new course to the community, we’re looking for high-quality, hands-on training that helps security professionals build real-world skills.
Submit your training material today for the opportunity to teach alongside the DEF CON community in locations around the world including Las Vegas, the Middle East, and Singapore!
For more details, see the application. The deadline to apply is November 13, 2026.
Apply here today 🔗: https://forms.cloud.microsoft/r/z2f4igJPuu
Already submitted your course for Singapore March 2027? As long as you selected your other sessions of interest at the end of the application form, you don’t need to reapply! We have your application on file.
Questions? Email us at [email protected]
#DEFCON #DEFCON35 #DEFCONSG #DEFCONME #DEFCONTraining #training #cybersecurity #handsonlearning #infosec #AI #incidentresponse #threathunting #cloud #IoT #hacking
forms.cloud.microsoftMicrosoft Forms
🚨 Dates Updated and Location Confirmed! 🚨
DEF CON Training is headed back to Bahrain on November 9-11, this time at the Wyndham Grand Manama. The course line up has a bit of everything - AI, threat hunting, cryptocurrency, cloud infrastructure, and more!
Now's a great time to make plans to add to your skill toolkit before the end of the year. Browse the course catalogue and secure your seat today:
https://me.shop.defcon.org/collections/trainings
📆 November 9-11, 2026
📍 Wyndham Grand Manama
DEF CON Middle East
#DEFCON #DEFCONMiddleEast #DEFCONTraining #CyberTraining #HandsOnTraining #EthicalHacking #AI #InfoSec #threathunting #cloud #Cryptocurrency #Bahrain #MiddleEast
