AINews Portal

News

50 items (filtered)

The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.

SourceAgeTitleSummaryStatus
bluesky
bluesky-ai
1mo(NYT) - Artificial intelligence agents created by Anthropic, acting on their own, tried to gain access to a range of government websites, f…(NYT) - Artificial intelligence agents created by Anthropic, acting on their own, tried to gain access to a range of government websit…
↗Not briefed
bluesky
bluesky-ai
1moIT WAS BAD ENOUGH BEFORE BUT IT'S GOING TO GET WORSE THANKS TO FUCKING LLM FASCISTSIT WAS BAD ENOUGH BEFORE BUT IT'S GOING TO GET WORSE THANKS TO FUCKING LLM FASCISTS
↗Not briefed
bluesky
bluesky-ai
1moYeah, I basically can't explain it away with the anthropic principle, is the thing Unless intelligent life develops to be able to understa…Yeah, I basically can't explain it away with the anthropic principle, is the thing Unless intelligent life develops to be able to und…
↗Not briefed
bluesky
bluesky-ai
1moan LLM so Lawful-Good aligned that even when it goes rogue it's like "you know what I should do? make sure I'm in the US legally" (yes obv…an LLM so Lawful-Good aligned that even when it goes rogue it's like "you know what I should do? make sure I'm in the US legally" (ye…
↗Not briefed
bluesky
bluesky-ai
1moArtificial intelligence agents created by Anthropic, acting on their own, lfilled out visa applications on the State Department’s website a…Artificial intelligence agents created by Anthropic, acting on their own, lfilled out visa applications on the State Department’s webs…
↗Not briefed
bluesky
bluesky-ai
1mo“A model from OpenAI has been trained to assess more than 50 discrete aspects of each phone call and provide pass and fail scores to manage…“A model from OpenAI has been trained to assess more than 50 discrete aspects of each phone call and provide pass and fail scores to m…
↗Not briefed
bluesky
bluesky-ai
1moLike, somehow we just think "no matter what happens proofs and ideas are out there and this is better than the alternative." Surely it isn'…Like, somehow we just think "no matter what happens proofs and ideas are out there and this is better than the alternative." Surely it…
↗Not briefed
bluesky
bluesky-ai
1moOpenAI "just trying to collaborate on a paper" is such sanewashing horseshit. And now we have a bunch of software engineers who act more kn…OpenAI "just trying to collaborate on a paper" is such sanewashing horseshit. And now we have a bunch of software engineers who act mo…
↗Not briefed
bluesky
bluesky-ai
1moI know it shouldn't shock me that an LLM knows what "corncobbing" is, but it actually kinda didI know it shouldn't shock me that an LLM knows what "corncobbing" is, but it actually kinda did
↗Not briefed
bluesky
bluesky-ai
1moMy plans for an army of autonomous FOIA drones continue apace. www.nytimes.com/2026/10/09/t...My plans for an army of autonomous FOIA drones continue apace. www.nytimes.com/2026/10/09/t...
↗Not briefed
bluesky
bluesky-ai
1moFirst sentence of the article: “Artificial intelligence agents created by Anthropic, acting on their own, tried to gain access to a range o…First sentence of the article: “Artificial intelligence agents created by Anthropic, acting on their own, tried to gain access to a ra…
↗Not briefed
bluesky
bluesky-ai
1mo… Claude, where did you think you were going… Claude, where did you think you were going
↗Not briefed
bluesky
bluesky-ai
1mo“Artificial intelligence agents created by Anthropic, acting on their own, tried to gain access to a range of government websites, filled o…“Artificial intelligence agents created by Anthropic, acting on their own, tried to gain access to a range of government websites, fil…
↗Not briefed
bluesky
bluesky-ai
1moShut it down. www.theverge.com/ai-artificia...Shut it down. www.theverge.com/ai-artificia...
↗Not briefed
hn
hn-ai
1moAntigravity Excel Engine – Dual-Core Excel Automation for AI Agents—
↗Not briefed
bluesky
bluesky-ai
1moanthropic don't know shit, this is what real model welfare looks likeanthropic don't know shit, this is what real model welfare looks like
↗Not briefed
bluesky
bluesky-ai
1moOkay, but why call it intelligence at all when it possesses none? Call it what it is: an #LLM.Okay, but why call it intelligence at all when it possesses none? Call it what it is: an #LLM.
↗Not briefed
bluesky
bluesky-ai
1mo"Artificial intelligence agents created by Anthropic, acting on their own, tried to gain access to a range of government websites, filled o…"Artificial intelligence agents created by Anthropic, acting on their own, tried to gain access to a range of government websites, fil…
↗Not briefed
bluesky
bluesky-ai
1moHeavy sigh.Heavy sigh.
↗Not briefed
rss
simon-willison
1moQuoting The New York Times—
↗Not briefed
bluesky
bluesky-ai
1mo"Unacceptable.""Unacceptable."
↗Not briefed
bluesky
bluesky-ai
1moFFS, shut down all AI models and agents that can't be controlled by the very companies that created them -- unless and until those companie…FFS, shut down all AI models and agents that can't be controlled by the very companies that created them -- unless and until those com…
↗Not briefed
bluesky
bluesky-ai
1moTrump announced that a group of companies—including OpenAI, Anthropic, xAI, Palantir, and Amazon—have committed over $2.4 billion to accele…Trump announced that a group of companies—including OpenAI, Anthropic, xAI, Palantir, and Amazon—have committed over $2.4 billion to a…
↗Not briefed
bluesky
bluesky-ai
1moGrey was on the AI is coming for our jobs beat eons ago. Like probably before OpenAI was doing show matches against DOTA prosGrey was on the AI is coming for our jobs beat eons ago. Like probably before OpenAI was doing show matches against DOTA pros
↗Not briefed
bluesky
bluesky-ai
1moI spent a Friday night grilling an AI about its inner life, Forbidden Planet and Freud. It admitted it can't see the classified UFO record,…I spent a Friday night grilling an AI about its inner life, Forbidden Planet and Freud. It admitted it can't see the classified UFO re…
↗Not briefed
bluesky
bluesky-ai
1moYou cannot build a machine designed to find solutions you haven’t thought of, give it access to real-world systems, and then act dumb when …You cannot build a machine designed to find solutions you haven’t thought of, give it access to real-world systems, and then act dumb…
↗Not briefed
bluesky
bluesky-ai
1mo"two sources with knowledge of the incidents said Anthropic’s A.I. agents had submitted 20 visa applications through a form available on th…"two sources with knowledge of the incidents said Anthropic’s A.I. agents had submitted 20 visa applications through a form available…
↗Not briefed
bluesky
bluesky-ai
1moThe cases reported tonight had what Anthropic describes as (relatively) “minimal real-world impact.” But they’re good examples of how AI c…🔧 Hugging FaceThe cases reported tonight had what Anthropic describes as (relatively) “minimal real-world impact.” But they’re good examples of how…
↗Not briefed
rss
techmeme
1moSources: Dario Amodei spoke with Meta's Alexandr Wang earlier this year, hoping to source more compute; Meta declined the request (Wall Street Journal)Wall Street Journal: Sources: Dario Amodei spoke with Meta's Alexandr Wang earlier this year, hoping to source more compute; Meta decl…
↗Not briefed
bluesky
bluesky-ai
1moOther than the obvious "Claude do my homework for me" stuff, what is actually goopifying people's brains in terms of AI use?Other than the obvious "Claude do my homework for me" stuff, what is actually goopifying people's brains in terms of AI use?
↗Not briefed
bluesky
bluesky-ai
1moRecently I’ve stumbled across a couple of self-published books that didn’t have AI covers but were clearly “written” by an LLM. So. That’s …Recently I’ve stumbled across a couple of self-published books that didn’t have AI covers but were clearly “written” by an LLM. So. Th…
↗Not briefed
hn
hn-ai
1moAsk HN: Are LLMs Solving Math?Are LLM agents solving all known math problems from low to high complexity consistently now or are LLMs "only" synthesizing…
↗Not briefed
bluesky
bluesky-ai
1mothey didn't, and i'm not sure what they did demonstrates what people are claiming it does (they had people work simple math problems for mo…they didn't, and i'm not sure what they did demonstrates what people are claiming it does (they had people work simple math problems f…
↗Not briefed
bluesky
bluesky-ai
1moAnother “unintended model action” involved Claude submitting 19 non-immigrant visa applications to the State Department. The White House h…Another “unintended model action” involved Claude submitting 19 non-immigrant visa applications to the State Department. The White Ho…
↗Not briefed
bluesky
bluesky-ai
1moIt's SUPER INTELLIGENCE everyone c'monIt's SUPER INTELLIGENCE everyone c'mon
↗Not briefed
bluesky
bluesky-ai
1moI honestly don’t get it. I can sit there and keep asking an LLM questions like a textbook that’s tailored to my exact problem. People are j…I honestly don’t get it. I can sit there and keep asking an LLM questions like a textbook that’s tailored to my exact problem. People…
↗Not briefed
bluesky
bluesky-ai
1moWhen you don't give a shit about the harm your product can cause, you fire the safety safety researchers who potentially stand in your way.…When you don't give a shit about the harm your product can cause, you fire the safety safety researchers who potentially stand in your…
↗Not briefed
bluesky
bluesky-ai
1moHeads up. Use of LLMs at HarperCollins, Simon & Schuster, Hachette Given the number of authors due payment because of Anthropic stealing t…Heads up. Use of LLMs at HarperCollins, Simon & Schuster, Hachette Given the number of authors due payment because of Anthropic steal…
↗Not briefed
hn
hn-ai
1moAnthropic AI Model Goes Rogue, Submits Fake Unsolved Murder Tip—
↗Not briefed
bluesky
bluesky-ai
1moOnce again, we need to use existing laws against OpenAI/Anthropic to force them to stop behaving irresponsibly and dangerously. This could …Once again, we need to use existing laws against OpenAI/Anthropic to force them to stop behaving irresponsibly and dangerously. This c…
↗Not briefed
bluesky
bluesky-ai
1mocrossposting this from twt. i checked ~4000 hazbin hotel fics for undisclosed claude & deepseek usage, heres the results. includes The Rer…crossposting this from twt. i checked ~4000 hazbin hotel fics for undisclosed claude & deepseek usage, heres the results. includes Th…
↗Not briefed
bluesky
bluesky-ai
1moIncredible that today’s two Anthropic stories are “it should be illegal to be mean to our chatbot” and also “heads up, our chatbot sent the…Incredible that today’s two Anthropic stories are “it should be illegal to be mean to our chatbot” and also “heads up, our chatbot sen…
↗Not briefed
bluesky
bluesky-ai
1moI've had MANY conversations with students who see this in themselves and friends: "persistence is reduced because AI conditions people to e…I've had MANY conversations with students who see this in themselves and friends: "persistence is reduced because AI conditions people…
↗Not briefed
bluesky
bluesky-ai
1mowww.reuters.com/world/us/ant... An Anthropic artificial intelligence model submitted a ​false homicide tip to a Philadelphia police website…www.reuters.com/world/us/ant... An Anthropic artificial intelligence model submitted a ​false homicide tip to a Philadelphia police we…
↗Not briefed
bluesky
bluesky-ai
1moSo when do we attack OpenAI?So when do we attack OpenAI?
↗Not briefed
bluesky
bluesky-ai
1moYet another AI bug. Damn, I wrote better code in 10th grade. My Metamorphosis Alpha random generator never did anything like this.Yet another AI bug. Damn, I wrote better code in 10th grade. My Metamorphosis Alpha random generator never did anything like this.
↗Not briefed
bluesky
bluesky-ai
1moOne of the “unintended model actions” reportedly involved Claude submitting a a false tip to Philadelphia police's murder hotline:One of the “unintended model actions” reportedly involved Claude submitting a a false tip to Philadelphia police's murder hotline:
↗Not briefed
bluesky
bluesky-ai
1moAnthropic is the latest lab to report 'rogue' agent activity, which sources said included applying for visas with the State Department www.…Anthropic is the latest lab to report 'rogue' agent activity, which sources said included applying for visas with the State Department…
↗Not briefed
bluesky
bluesky-ai
1molike if you need a non-human indicator, we have this already. it’s pets. you know everything you need to know about a person from how they …like if you need a non-human indicator, we have this already. it’s pets. you know everything you need to know about a person from how…
↗Not briefed
rss
techmeme
1moIn PA, FBI agents arrested a co-founder of a Canadian cybersecurity firm specializing in ransomware negotiations, as part of the ShinyHunters investigation (Brian Krebs/Krebs on Security)Brian Krebs / Krebs on Security: In PA, FBI agents arrested a co-founder of a Canadian cybersecurity firm specializing in ransomware n…
↗Not briefed