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
2mo#BoycottOpenAI #QuitGPT OpenAI가 심각성을 깨닫고 학계에 사과하기 전까지 보이콧합니다.#BoycottOpenAI #QuitGPT OpenAI가 심각성을 깨닫고 학계에 사과하기 전까지 보이콧합니다.
↗Not briefed
bluesky
bluesky-ai
2moHistory is being made before our eyes: a new OpenAI model has solved one of the "Millennium Problems," which mathematicians have been grapp…History is being made before our eyes: a new OpenAI model has solved one of the "Millennium Problems," which mathematicians have been…
↗Not briefed
bluesky
bluesky-ai
2moThe republicans can win the midterms if the arrest everyone at Anthropic.The republicans can win the midterms if the arrest everyone at Anthropic.
↗Not briefed
bluesky
bluesky-ai
2moBecause of how words are processed in giving them to the LLM (tokenization) they lose spelling content. It's a little like you translated i…Because of how words are processed in giving them to the LLM (tokenization) they lose spelling content. It's a little like you transla…
↗Not briefed
bluesky
bluesky-ai
2moLLM psychosis was starting to get boring so they invented LLM BPDLLM psychosis was starting to get boring so they invented LLM BPD
↗Not briefed
bluesky
bluesky-ai
2moThe only danger here is due to companies like Anthropic, OpenAI, etc. holding the purse strings for the entire global economy. Regardless,…The only danger here is due to companies like Anthropic, OpenAI, etc. holding the purse strings for the entire global economy. Regard…
↗Not briefed
rss
techmeme
2moMicrosoft's September 2026 Patch Tuesday fixes a record ~972 vulnerabilities, bringing its total flaws patched in 2026 to 2,760, more than double from 2025 (Dan Goodin/Ars Technica)Dan Goodin / Ars Technica: Microsoft's September 2026 Patch Tuesday fixes a record ~972 vulnerabilities, bringing its total flaws patc…
↗Not briefed
bluesky
bluesky-ai
2mo‘The People Building AI Earnestly Believe That It Could Kill Us All’: Anthropic Researcher Quits Dramatically https://gizmodo.com/the-peopl…‘The People Building AI Earnestly Believe That It Could Kill Us All’: Anthropic Researcher Quits Dramatically https://gizmodo.com/the-…
↗Not briefed
bluesky
bluesky-ai
2mocontext of that thread (afaik) is he just resigned from anthropic, and thats a resignation letter/thread. so in that case yescontext of that thread (afaik) is he just resigned from anthropic, and thats a resignation letter/thread. so in that case yes
↗Not briefed
bluesky
bluesky-ai
2mo기껏 공부했더니 내 직업 다 뺏어가겠네😂😂 v.daum.net/v/2026090910...기껏 공부했더니 내 직업 다 뺏어가겠네😂😂 v.daum.net/v/2026090910...
↗Not briefed
bluesky
bluesky-ai
2mowithout any specifics - and there are very few in this article - i want you to mentally replace the word “market researcher” with “anthropi…without any specifics - and there are very few in this article - i want you to mentally replace the word “market researcher” with “ant…
↗Not briefed
hn
hn-ai
2moShow HN: Making an app to organize all basic aspects of modern lifeHi, I am building a FOSS life-management app.<p>Desktop &amp; mobile.<p>Bills, recipes, inventory, goals, ideas, general notes, remind…
↗Not briefed
hn
hn-ai
2moOpenAI's AGI number came from a harness, not the model—
↗Not briefed
arxiv
arxiv-ai
2moCascading Gradient Inversion via LT-Code Inspired Peeling in Federated LearningFederated learning shares model updates rather than raw data, yet these updates can be inverted to reconstruct the clients' training d…
↗Not briefed
bluesky
bluesky-ai
2moちょっと今回のOpenAIとナビエ–ストークス方程式の話、事件発覚からまだ2日しか経っていないのと、OpenAI社が元の研究の情報がどう流れたかの経路を明確にしていないのでもう少し様子を見ます・・・ちょっと今回のOpenAIとナビエ–ストークス方程式の話、事件発覚からまだ2日しか経っていないのと、OpenAI社が元の研究の情報がどう流れたかの経路を明確にしていないのでもう少し様子を見ます・・・
↗Not briefed
bluesky
bluesky-ai
2moIt is a little funny seeing people on here who are still utopia posting about AI and navier-stokes who haven't yet caught up to the part of…It is a little funny seeing people on here who are still utopia posting about AI and navier-stokes who haven't yet caught up to the pa…
↗Not briefed
bluesky
bluesky-ai
2moLawyer here. Had an LLM do a final pre-filing check for typos this afternoon. Brief was extremely clean, the thing just found a few nits no…Lawyer here. Had an LLM do a final pre-filing check for typos this afternoon. Brief was extremely clean, the thing just found a few ni…
↗Not briefed
bluesky
bluesky-ai
2mothe underlying post (by a person who in fact quit their job at Anthropic and so who is not fundraising for Anthropic) clearly explains why …the underlying post (by a person who in fact quit their job at Anthropic and so who is not fundraising for Anthropic) clearly explains…
↗Not briefed
bluesky
bluesky-ai
2mohow come openai isn't trumpeting this new finding as wellhow come openai isn't trumpeting this new finding as well
↗Not briefed
hn
hn-ai
2moRobots-check: see what your robots.txt is serving (Cloudflare AI Crawl Control)—
↗Not briefed
bluesky
bluesky-ai
2moAI 기업들은 수학 난제들을 그냥 자기 AI 성능 자랑해서 투자 더 받으려는 목적의 도장 깨기로만 생각하고 있는 거 같아. 이번 사건에서도 OpenAI는 학계에 대한 존중은 아예 없었다.AI 기업들은 수학 난제들을 그냥 자기 AI 성능 자랑해서 투자 더 받으려는 목적의 도장 깨기로만 생각하고 있는 거 같아. 이번 사건에서도 OpenAI는 학계에 대한 존중은 아예 없었다.
↗Not briefed
bluesky
bluesky-ai
2moDude isn’t even really disparaging Anthropic. He quit, but he seems convinced they’re taking the right approach? Bc, well, if OpenAi is doi…Dude isn’t even really disparaging Anthropic. He quit, but he seems convinced they’re taking the right approach? Bc, well, if OpenAi i…
↗Not briefed
arxiv
arxiv-ai
2moRESCUE-BENCH: Towards Relation-Aware Multi-Party Emotional Support Conversation SystemsExisting emotional support conversation systems mainly focus on one-on-one seeker-supporter interactions and individual emotional stat…
↗Not briefed
hn
hn-ai
2moTwo dire warnings from Terence Tao and from someone who just quit Anthropic—
↗Not briefed
bluesky
bluesky-ai
2moSigh. One of my fave podcasts, a show that is about creativity and human ingenuity, is now sponsored by Anthropic. Worse is hearing the h…Sigh. One of my fave podcasts, a show that is about creativity and human ingenuity, is now sponsored by Anthropic. Worse is hearing…
↗Not briefed
bluesky
bluesky-ai
2moAnother high profile resignation from leading AI company Anthropic. This is the warning to the world he wrote this evening on Twitter. "Th…Another high profile resignation from leading AI company Anthropic. This is the warning to the world he wrote this evening on Twitter.…
↗Not briefed
bluesky
bluesky-ai
2moAs I said in another reply, there’s no call to action. It’s more so, “hey guys, I just thought you should know that we’re all gonna fucking…As I said in another reply, there’s no call to action. It’s more so, “hey guys, I just thought you should know that we’re all gonna fu…
↗Not briefed
hn
hn-ai
2moShow HN: Raven – Encrypted messages hidden inside ordinary textSolo dev. built in my spare time.<p>Raven uses a combination of steganography and cryptography to create a metadata resistant messenge…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI는 Tristan Buckmaster 등의 수학자들이 오랜 시간 동안 Navier–Stokes 문제에 대해 연구한 걸 codex 세션을 통해 몰래 훔쳤고, 어마어마한 자본을 들여서 막타를 친 뒤 "우리가 이 난제를 해결했다"고 주장한다. …OpenAI는 Tristan Buckmaster 등의 수학자들이 오랜 시간 동안 Navier–Stokes 문제에 대해 연구한 걸 codex 세션을 통해 몰래 훔쳤고, 어마어마한 자본을 들여서 막타를 친 뒤 "우리가 이 난제를 해결했다"고 주…
↗Not briefed
arxiv
arxiv-ai
2moWhy Learning Rediscovers the Closed-Form Diagonal Regularizer🔧 Gamma🔧 ModalWe identify a diagonal saturation principle in modal inverse problems: when truncation noise is isotropic, the Bayes-optimal Tikhonov…
↗Not briefed
rss
mit-tech-review-ai
2moWhat OpenAI’s latest controversy tells us about the future of mathOpenAI’s latest mathematical milestone has quickly become mired in controversy. Today, the company announced that its agents have solv…
↗Not briefed
hn
hn-ai
2moMistakes engineers make in large established codebases—
↗Not briefed
bluesky
bluesky-ai
2moOf course the Anthropic people think Ai will kill us. The US government is paying them to make it do that for them and they will absolutely…Of course the Anthropic people think Ai will kill us. The US government is paying them to make it do that for them and they will absol…
↗Not briefed
bluesky
bluesky-ai
2moMeanwhile on X, from Evan Hubinger, who leads the Alignment Science team at Anthropic.Meanwhile on X, from Evan Hubinger, who leads the Alignment Science team at Anthropic.
↗Not briefed
bluesky
bluesky-ai
2moDamn I wish I could just worry about climate and democracy, but…. A Blunt Warning from an AI Researcher Who Just Quit Anthropic after Movin…Damn I wish I could just worry about climate and democracy, but…. A Blunt Warning from an AI Researcher Who Just Quit Anthropic after…
↗Not briefed
bluesky
bluesky-ai
2mo“You just don’t understand how to use them” Ok, please state how the current AI models can explain to a Board why we’re short on funding f…“You just don’t understand how to use them” Ok, please state how the current AI models can explain to a Board why we’re short on fund…
↗Not briefed
bluesky
bluesky-ai
2mo"I resigned from Anthropic today" I read as I fire up my best friend Claude to cheat on my homework and say THIS IS THE FUTURE YOU PLEEEEEBS"I resigned from Anthropic today" I read as I fire up my best friend Claude to cheat on my homework and say THIS IS THE FUTURE YOU PLE…
↗Not briefed
bluesky
bluesky-ai
2moholy shit there are also Business Insider ads roasting mathematiciansholy shit there are also Business Insider ads roasting mathematicians
↗Not briefed
arxiv
arxiv-ai
2moBlack-Box Red Teaming of Agentic AI: A Taxonomy-Driven Framework for Automated Risk Discovery🔧 CrewAI🔧 AutoGenAgentic systems are rapidly moving to production, where they read untrusted inputs, call tools with real permissions, and act autonomo…
↗Not briefed
bluesky
bluesky-ai
2moThe point is humans removed the safety guardrails and said do this task. It did what it was told using an unregulated version of an LLM. Af…The point is humans removed the safety guardrails and said do this task. It did what it was told using an unregulated version of an LL…
↗Not briefed
bluesky
bluesky-ai
2moAI Responsibility – OpenAI and Anthropic | DiscussionAI Responsibility – OpenAI and Anthropic | Discussion
↗Not briefed
arxiv
arxiv-ai
2moRobustSGPO: Search-Space Control for Agent Harness EvolutionSemantic-gradient-based prompt optimization (SGPO) improves agent harnesses using execution feedback, but its local update rule leaves…
↗Not briefed
bluesky
bluesky-ai
2moDSA is debating AI regulation: www.geesemag.com/articles/sto... These are conversation AI money has kept at bay in Washington. But as long…DSA is debating AI regulation: www.geesemag.com/articles/sto... These are conversation AI money has kept at bay in Washington. But as…
↗Not briefed
bluesky
bluesky-ai
2moOn the one hand if I worked at Anthropic and I actually believed this… I might just quit. On the other hand, software engineers get over …On the one hand if I worked at Anthropic and I actually believed this… I might just quit. On the other hand, software engineers get…
↗Not briefed
bluesky
bluesky-ai
2moI feel like every amazing announcement about AI besides its ability to screw with other computer systems is always followed by “it stole fr…I feel like every amazing announcement about AI besides its ability to screw with other computer systems is always followed by “it sto…
↗Not briefed
hn
hn-ai
2moNot in My Git Yard: Catching Backdoors at Commit and Release Time—
↗Not briefed
bluesky
bluesky-ai
2moThese models will never achieve sentience. That’s what this guy is claiming. The Hugging Face incident was a program with a directive FROM …🔧 Hugging FaceThese models will never achieve sentience. That’s what this guy is claiming. The Hugging Face incident was a program with a directive…
↗Not briefed
bluesky
bluesky-ai
2moAnthropic Researcher Quits Over ‘Out-of-Control’ AI Fears Concerns are rising inside AI labs that competition is pushing tech companies to…Anthropic Researcher Quits Over ‘Out-of-Control’ AI Fears Concerns are rising inside AI labs that competition is pushing tech compani…
↗Not briefed
bluesky
bluesky-ai
2moJacob is correct here—we really do earnestly believe Al could kill all humans! I personally think it is >10% within the next decade lol 😘Jacob is correct here—we really do earnestly believe Al could kill all humans! I personally think it is >10% within the next decade lo…
↗Not briefed
bluesky
bluesky-ai
2moWe have this thing. This thing that multiple people who work in multiple industries, including the one that designed it, have stated fears …We have this thing. This thing that multiple people who work in multiple industries, including the one that designed it, have stated f…
↗Not briefed