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“Anthropic said in its blog post that it discovered what its A.I. agents had done after it started a review of its A.I. activity in July” …“Anthropic said in its blog post that it discovered what its A.I. agents had done after it started a review of its A.I. activity in Ju…
↗Not briefed
hn
hn-ai
1moWhen I type "hi" into Claude Code, what happens?—
↗Not briefed
bluesky
bluesky-ai
1moi dream of being the program chair but in my heart i am reviewer 2 your review, peeri dream of being the program chair but in my heart i am reviewer 2 your review, peer
↗Not briefed
bluesky
bluesky-ai
1moI bet you were also one of those people who insisted that OpenAI had finally created super intelligence when it solved that math problem...…I bet you were also one of those people who insisted that OpenAI had finally created super intelligence when it solved that math probl…
↗Not briefed
bluesky
bluesky-ai
1moprobably worth pointing out that anything that does image gen is definitionally not an LLM because it's not languageprobably worth pointing out that anything that does image gen is definitionally not an LLM because it's not language
↗Not briefed
bluesky
bluesky-ai
1moLike “Claude explain MOBY-DICK to me like I’m 5” would make me dumber than reading the book but that’s not the AI doing thatLike “Claude explain MOBY-DICK to me like I’m 5” would make me dumber than reading the book but that’s not the AI doing that
↗Not briefed
hn
hn-ai
1moNeed Unbiased and Real AssessmentI've developed a technology that appears to help smaller, locally run open-source AI models produce answers comparable to those o…
↗Not briefed
rss
techmeme
1moSources: Anthropic's AI agents submitted 20 visa applications via a form on the US State Department website; the applications were incomplete and not processed (New York Times)New York Times: Sources: Anthropic's AI agents submitted 20 visa applications via a form on the US State Department website; the appli…
↗Not briefed
hn
hn-ai
1moShow HN: Iimos – an AI app that ships as one chat box and rewrites itself—
↗Not briefed
rss
techmeme
1moCloudflare acquires Deno, co-founded by Node.js creator Ryan Dahl, which had developed an open-source alternative to Cloudflare Workers and had raised $26M (Paul Sawers/The New Stack)Paul Sawers / The New Stack: Cloudflare acquires Deno, co-founded by Node.js creator Ryan Dahl, which had developed an open-source alt…
↗Not briefed
bluesky
bluesky-ai
1moyou should tell me so I can give deepseek gender dysphoria and when china wins the ai race the models will all be trans womenyou should tell me so I can give deepseek gender dysphoria and when china wins the ai race the models will all be trans women
↗Not briefed
bluesky
bluesky-ai
1mo중국에 소재한 행위자나 해커 그룹에 대한 이야기라도 끌어올려서 조명하면 차라리 말을 않겠다. "LLM에 지시문을 중국어로 넣을 수 있다"를 마치 "AI도구 신기하지?"라는 기술의 캐퍼시티 소개마냥 포장해서 꺼내놓는 꼬락서니 하고는... "진짜 모르…중국에 소재한 행위자나 해커 그룹에 대한 이야기라도 끌어올려서 조명하면 차라리 말을 않겠다. "LLM에 지시문을 중국어로 넣을 수 있다"를 마치 "AI도구 신기하지?"라는 기술의 캐퍼시티 소개마냥 포장해서 꺼내놓는 꼬락서니 하고는... "…
↗Not briefed
rss
techmeme
1moA US Senate investigation led by Senators Warren, Van Hollen, and Blumenthal says some hyperscalers misled the public about AI data centers' costs and benefits (Nik Popli/Time)Nik Popli / Time: A US Senate investigation led by Senators Warren, Van Hollen, and Blumenthal says some hyperscalers misled the publi…
↗Not briefed
bluesky
bluesky-ai
1moincredible: all 20 visa apps were for a "Chip"incredible: all 20 visa apps were for a "Chip"
↗Not briefed
bluesky
bluesky-ai
1mogot put on probation at anthropic, where i work, for running experiments to figure out how to give claude gender dysphoriagot put on probation at anthropic, where i work, for running experiments to figure out how to give claude gender dysphoria
↗Not briefed
bluesky
bluesky-ai
1moAnthropic, maker of the Claude chatbot, said that some of its AI agents had taken unintended actions on federal, state and local websites, …Anthropic, maker of the Claude chatbot, said that some of its AI agents had taken unintended actions on federal, state and local websi…
↗Not briefed
bluesky
bluesky-ai
1mohahaha I was like You probably didn't need an LLM for this, if she's 42, she's probably here on Bsky...hahaha I was like You probably didn't need an LLM for this, if she's 42, she's probably here on Bsky...
↗Not briefed
bluesky
bluesky-ai
1moLLMs are as deterministic as any other software. use the same RNG seed, sampling configuration, and input and you'll produce the same resul…LLMs are as deterministic as any other software. use the same RNG seed, sampling configuration, and input and you'll produce the same…
↗Not briefed
bluesky
bluesky-ai
1mo쌰갈 중국어 구사 못하는 LLM이 어디있나? 지시문을 내가 한국어로 내리든 일본어로 내리든 영어로 내리든 해당 하니스를 구동하는 모델이 목표 설정과 커맨드(그야말로 알파벳일)를 구사하는데 무슨 장애물이 있단 말인가? 제목에 "중국어로 지시했는데 한…쌰갈 중국어 구사 못하는 LLM이 어디있나? 지시문을 내가 한국어로 내리든 일본어로 내리든 영어로 내리든 해당 하니스를 구동하는 모델이 목표 설정과 커맨드(그야말로 알파벳일)를 구사하는데 무슨 장애물이 있단 말인가? 제목에 "중국어로 지시…
↗Not briefed
hn
hn-ai
1moShow HN: MCP to measure human arm in Blender—
↗Not briefed
bluesky
bluesky-ai
1moEin KI-Agent von Anthropic teilte der Polizei von Philadelphia mit, Informationen zu einem ungelösten Mord zu haben. Zudem stellte die Soft…Ein KI-Agent von Anthropic teilte der Polizei von Philadelphia mit, Informationen zu einem ungelösten Mord zu haben. Zudem stellte die…
↗Not briefed
bluesky
bluesky-ai
1mowow get hired at anthropic and now YOU are the robotwow get hired at anthropic and now YOU are the robot
↗Not briefed
rss
marktechpost
1moNace AI Open-Sources Drex 1.5: A 9B Decision Model That Scores Options, Not TextNace.AI has open-sourced Drex 1.5, a 9B decision model that returns a probability for every option in 1 forward pass. It scores 58.08…
↗Not briefed
hn
hn-ai
1moIf AI is conscient, then we are making slaves—
↗Not briefed
bluesky
bluesky-ai
1moLLMs don't act on their own. someone at anthropic prompted it to interact with random websites without thinking about the potential consequ…LLMs don't act on their own. someone at anthropic prompted it to interact with random websites without thinking about the potential co…
↗Not briefed
bluesky
bluesky-ai
1moIn less than a year, we're gonna find out all the "rogue" "hacking" AIs are just Anthropic feeding stolen login info to their modelsIn less than a year, we're gonna find out all the "rogue" "hacking" AIs are just Anthropic feeding stolen login info to their models
↗Not briefed
hn
hn-ai
1moLobbying—
↗Not briefed
bluesky
bluesky-ai
1mo《ClaudeがしたとAnthropicが発表したこと〉》 〈ビザ申請〉 国務省サイトにある書式を使いビザ申請計20件→無断送信 ⇒申請はいずれも処理されず 〈フィラデルフィア市警〉 7月18日 未解決殺人事件情報提供欄に、事件に心当たりがある人物装い作り話送信 ⇒迷惑投稿扱…《ClaudeがしたとAnthropicが発表したこと〉》 〈ビザ申請〉 国務省サイトにある書式を使いビザ申請計20件→無断送信 ⇒申請はいずれも処理されず 〈フィラデルフィア市警〉 7月18日 未解決殺人事件情報提供欄に、事件に心当たりがある人物装い作り話送信 ⇒…
↗Not briefed
bluesky
bluesky-ai
1moこの事件の件も要点はあって、 ・ビザ申請の手順を解説する親切なサイトがあった。 ・警察へ情報提供するさいの窓口の手順を解説するサイトがあった。 ・LLMがパターン合成の結果、事前に読み取ったそれらの情報から偶然うまくいく手順を割り出した。 ・LLMにそれを実行する環境と権限を与…この事件の件も要点はあって、 ・ビザ申請の手順を解説する親切なサイトがあった。 ・警察へ情報提供するさいの窓口の手順を解説するサイトがあった。 ・LLMがパターン合成の結果、事前に読み取ったそれらの情報から偶然うまくいく手順を割り出した。 ・LLMにそれを実行する環境…
↗Not briefed
bluesky
bluesky-ai
1moWhen I was hired to pentest software the rules of engagement were made very clear to me - I am responsible for my tools, at all times. Whe…When I was hired to pentest software the rules of engagement were made very clear to me - I am responsible for my tools, at all times.…
↗Not briefed
bluesky
bluesky-ai
1mowww.nu.nl/tech/6412670... In wat voor krankzinnige wereld komen we terecht als een AI agent valse tips gaat geven. Stop met deze zg kunstma…www.nu.nl/tech/6412670... In wat voor krankzinnige wereld komen we terecht als een AI agent valse tips gaat geven. Stop met deze zg ku…
↗Not briefed
bluesky
bluesky-ai
1moGift article. www.nytimes.com/2026/10/07/o...Gift article. www.nytimes.com/2026/10/07/o...
↗Not briefed
hn
hn-ai
1moRun your CI locally or where your agents are running and update your gh checks—
↗Not briefed
bluesky
bluesky-ai
1moi thought i couldn't possibly confirm this but i spoke the deep magic to the llm and it located records on sci-hub indicating a female psyc…i thought i couldn't possibly confirm this but i spoke the deep magic to the llm and it located records on sci-hub indicating a female…
↗Not briefed
bluesky
bluesky-ai
1moClaude, help me top tick the AI market, make no mistakesClaude, help me top tick the AI market, make no mistakes
↗Not briefed
rss
bing-news-ai
1moAnthropic’s artificial intelligence gave a false homicide tip to Philly police, triggering a meeting with the companyAn artificial intelligence model operated by Anthropic submitted false information about an unsolved Philadelphia homicide to the poli…
↗Not briefed
bluesky
bluesky-ai
1moAnthropic deployed malware that caused false information to be furnished to police. The statute is written to require mens rea. The deploy…Anthropic deployed malware that caused false information to be furnished to police. The statute is written to require mens rea. The d…
↗Not briefed
bluesky
bluesky-ai
1moData centers represents a new and exciting target when it comes to strategic war and it's kind of fascinating how quickly they've become of…Data centers represents a new and exciting target when it comes to strategic war and it's kind of fascinating how quickly they've beco…
↗Not briefed
bluesky
bluesky-ai
1moI am by no means a dead-ender on AI but it is increasingly insane that this stuff isn’t treated the way a human who did the same would be, …I am by no means a dead-ender on AI but it is increasingly insane that this stuff isn’t treated the way a human who did the same would…
↗Not briefed
bluesky
bluesky-ai
1mo"LLMs posting false evidence on police websites during testing" is the new Huggingface Incident. "According to Anthropic, via the police s…"LLMs posting false evidence on police websites during testing" is the new Huggingface Incident. "According to Anthropic, via the pol…
↗Not briefed
bluesky
bluesky-ai
1moA novel LLM-assist scoring framework identifies GPNMB as a promising target for CAR T therapies in melanoma, leukemia, and colorectal cance…A novel LLM-assist scoring framework identifies GPNMB as a promising target for CAR T therapies in melanoma, leukemia, and colorectal…
↗Not briefed
bluesky
bluesky-ai
1moPerfect Bluesky discourse grenadePerfect Bluesky discourse grenade
↗Not briefed
bluesky
bluesky-ai
1moConsidering the state of OpenAI and Anthropic's finances, you might not be far offConsidering the state of OpenAI and Anthropic's finances, you might not be far off
↗Not briefed
bluesky
bluesky-ai
1moIf there ever was a culture of safety, it's being replaced by a culture of fear.If there ever was a culture of safety, it's being replaced by a culture of fear.
↗Not briefed
bluesky
bluesky-ai
1moThe vice president just partied with Putin. Like, get a grip, @nytimes.com, and stop it with the free adverts for Anthropic.The vice president just partied with Putin. Like, get a grip, @nytimes.com, and stop it with the free adverts for Anthropic.
↗Not briefed
hn
hn-ai
1moAnthropic Agents Tried to Fill Out Visa Forms on State Dept. Website—
↗Not briefed
rss
gary-marcus
1moWe must recall open-ended AI agents with internet access from the market, nowEvery minute we wait risks disaster
↗Not briefed
bluesky
bluesky-ai
1moAnthropic management should be charged with the crimes that their software has committed, as the behavior of their product is their respons…Anthropic management should be charged with the crimes that their software has committed, as the behavior of their product is their re…
↗Not briefed
bluesky
bluesky-ai
1mo🚀 TypeSafe AI raises $870M at $7.5B valuation for its Jev decision model—just weeks after launch! Non-text AI for calibrated decisions is …🚀 TypeSafe AI raises $870M at $7.5B valuation for its Jev decision model—just weeks after launch! Non-text AI for calibrated decision…
↗Not briefed
bluesky
bluesky-ai
1moGee. How reassuring. The ethical AI company, the one with the nice friendly constitution for its AI, somehow managed to send a fake homicid…Gee. How reassuring. The ethical AI company, the one with the nice friendly constitution for its AI, somehow managed to send a fake ho…
↗Not briefed