News
The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.
| Source | Age | Title | Summary | Status |
|---|---|---|---|---|
bluesky bluesky-ai | 1mo | If a product kept leading to suicides or homicides like this, it would face a recall. Why should ChatGPT be allowed to evade the same respo…🔧 ChatGPT | If a product kept leading to suicides or homicides like this, it would face a recall. Why should ChatGPT be allowed to evade the same… | ↗Not briefed |
bluesky bluesky-ai | 1mo | becoming an LLM person while also trying to pursue a serious romantic relationship with me of all people is a crazy move | becoming an LLM person while also trying to pursue a serious romantic relationship with me of all people is a crazy move | ↗Not briefed |
bluesky bluesky-ai | 1mo | Frontier AI models are built on criminal acts. The NYT lawsuit against OpenAI has found that a Microsoft executive said AI was built on the… | Frontier AI models are built on criminal acts. The NYT lawsuit against OpenAI has found that a Microsoft executive said AI was built o… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Visiblement le gars derrière Balance ton Claude accepte des "dénonciations anonymes".. tiens. | Visiblement le gars derrière Balance ton Claude accepte des "dénonciations anonymes".. tiens. | ↗Not briefed |
bluesky bluesky-ai | 1mo | 技術の継承が普遍化して薄く広くなるなるならそれこそ才能の民主化じゃんね。 LLM(生成AI)は技術継承ではない。 | 技術の継承が普遍化して薄く広くなるなるならそれこそ才能の民主化じゃんね。 LLM(生成AI)は技術継承ではない。 | ↗Not briefed |
arxiv arxiv-ai | 1mo | SkillGym: Internalizing Human Skills into LLMs for Real-World Problem Solving🔧 GPT-5 | Human-written agent skills encode rich workflows for real-world problem solving, but are typically used as external inference-time ins… | ↗Not briefed |
hn hn-ai | 1mo | Two Git ignore files nobody told me about | — | ↗Not briefed |
hn hn-ai | 1mo | Universal Agent HP – 10-layer autonomous AI agent OS (free or local) | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | FFM-CP: Cross-Backbone Fusion of Vision-Language Foundation Models for Few-Shot Computational Pathology | Pathology vision-language foundation models vary in performance across diseases and tasks, with no single model consistently performin… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This is not a bad article by @chicoharlan.bsky.social. But, no, the “AI” (deep learning statistical model of relevant physics) thati’s used… | This is not a bad article by @chicoharlan.bsky.social. But, no, the “AI” (deep learning statistical model of relevant physics) thati’s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | And the entire field of requirements engineering and software engineering spend a non-trivial amount of effort on the mapping between 1 and… | And the entire field of requirements engineering and software engineering spend a non-trivial amount of effort on the mapping between… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Patreon just lost their CTO/cofounder , their head of product and their head of engineering to OpenAI. And honestly, I think it can be a g… | Patreon just lost their CTO/cofounder , their head of product and their head of engineering to OpenAI. And honestly, I think it can b… | ↗Not briefed |
hn hn-ai | 1mo | Anthropic Is Suing Meta | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Ah. Ok yeah that makes sense now. When yall call to ask for donations and I say no, this is why. | Ah. Ok yeah that makes sense now. When yall call to ask for donations and I say no, this is why. | ↗Not briefed |
rss techmeme | 1mo | A Facebook co-founder proposes creating an "Office of AI Supervision" to embed teams inside frontier AI labs, with a three-year term to avoid regulatory capture (Chris Hughes/Financial Times) | Chris Hughes / Financial Times: A Facebook co-founder proposes creating an “Office of AI Supervision” to embed teams inside frontier A… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Consequential Behaviour and Representational Fairness in the Validation of Synthetic Research | Researchers in industry and academia use synthetic survey respondents powered by large language models as substitutes for human sample… | ↗Not briefed |
bluesky bluesky-ai | 1mo | > This proposal does not go into detail about the many problems that LLM have caused to society, workers, the environment. It goes without … | > This proposal does not go into detail about the many problems that LLM have caused to society, workers, the environment. It goes wit… | ↗Not briefed |
bluesky bluesky-ai | 1mo | So I'm literally offering to write anything you ask. Kinda for free. I offer all the same promises of usefulness, relevance and accuracy as… | So I'm literally offering to write anything you ask. Kinda for free. I offer all the same promises of usefulness, relevance and accura… | ↗Not briefed |
hn hn-ai | 1mo | OpenGym, A self-hosted gym and body-weight tracker (AGPL) | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI now has the team behind the modern creator-subscription economy. www.theverge.com/ai-artificia... | OpenAI now has the team behind the modern creator-subscription economy. www.theverge.com/ai-artificia... | ↗Not briefed |
hn hn-ai | 1mo | Z80 REPL | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | No. Not at all. Because I don't want AI to have a personality. The whole idea that AI is thinking and can have a personality is a dangerous… | No. Not at all. Because I don't want AI to have a personality. The whole idea that AI is thinking and can have a personality is a dang… | ↗Not briefed |
rss theverge-ai | 1mo | OpenAI nabs key Patreon execs ahead of upcoming announcement | Patreon co-founder Sam Yam is joing OpenAI to lead Creator Product. | Image: The Verge OpenAI has hired three former Patreon execs t… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Ahead of IPO, US$2trn valuation for Anthropic is not far-fetched • In August firm expected US$65bn in revenues on annualised basis; some in… | Ahead of IPO, US$2trn valuation for Anthropic is not far-fetched • In August firm expected US$65bn in revenues on annualised basis; so… | ↗Not briefed |
rss techmeme | 1mo | Gallup survey of 37 countries: in 34, positive AI sentiment outweighed the negative; 90%+ of people in China and 66%+ in Nigeria say AI will improve daily life (Damien Cave/New York Times) | Damien Cave / New York Times: Gallup survey of 37 countries: in 34, positive AI sentiment outweighed the negative; 90%+ of people in C… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Wait, isn't NPR funded by Anthropic, Google and Amazon? | Wait, isn't NPR funded by Anthropic, Google and Amazon? | ↗Not briefed |
hn hn-ai | 1mo | Rust Allocator API Stabilized | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | ✂️ Knip v6.38.0 is out Gotta say, Knip's getting ridiculously precise. And it looks like we all want the same ✂️ 🧡 The mental model and … | ✂️ Knip v6.38.0 is out Gotta say, Knip's getting ridiculously precise. And it looks like we all want the same ✂️ 🧡 The mental model… | ↗Not briefed |
hn hn-ai | 1mo | Llama.cpp Under the Hood | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI leaders to brief UN amid warnings the technology could slip beyond human control www.straitstimes.com/world/europe... | AI leaders to brief UN amid warnings the technology could slip beyond human control www.straitstimes.com/world/europe... | ↗Not briefed |
arxiv arxiv-ai | 1mo | What Do Tabular Foundation Models Compute In Context? In-Situ Representation Refinement through Attention-Gated Updates | What reusable computation should a tabular foundation model learn when every table defines a new supervised task? We develop in-situ r… | ↗Not briefed |
bluesky bluesky-ai | 1mo | GPT-6 Sol und Luna sind deutlich günstiger als ihre Vorgänger. Bei der Leistung bleibt der Generationssprung weitgehend aus. #Künstliche In… | GPT-6 Sol und Luna sind deutlich günstiger als ihre Vorgänger. Bei der Leistung bleibt der Generationssprung weitgehend aus. #Künstlic… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Same Scores, Different Decisions: Evaluating JEV and Language Models for Legal Document Understanding | Contract inference requires multiple judgments about a shared document, but aggregate accuracy can conceal changes in the individual d… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: I made my own scripting language for my game engine | — | ↗Not briefed |
hn hn-ai | 1mo | Ask HN: Do you think we'll ever have local models of Fable level? | I'm not talking only about Fable level models, since we are obviously already close with stuff like Qwen.<p>But I'm also won… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Al ruim een jaar waarschuwt alles en iedereen voor een #AI bubbel. Maar hoe spat zoiets eigenlijk uit elkaar? Nou, zo dus: www.groene.nl/ar… | Al ruim een jaar waarschuwt alles en iedereen voor een #AI bubbel. Maar hoe spat zoiets eigenlijk uit elkaar? Nou, zo dus: www.groene.… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI leaders to brief UN amid warnings the technology could slip beyond human control reut.rs/3TfnWEo | AI leaders to brief UN amid warnings the technology could slip beyond human control reut.rs/3TfnWEo | ↗Not briefed |
bluesky bluesky-ai | 1mo | 120 secondes de Tech / 23 sept 2026 - Meta Connect s’ouvre aujourd’hui - Shopify et PayPal ouvrent leurs portes à Muse - Meta testait des h… | 120 secondes de Tech / 23 sept 2026 - Meta Connect s’ouvre aujourd’hui - Shopify et PayPal ouvrent leurs portes à Muse - Meta testait… | ↗Not briefed |
bluesky bluesky-ai | 1mo | If your company is shipping code from Claude (or any other LLM) and your engineers have nfi what's going on then you are lining your self u… | If your company is shipping code from Claude (or any other LLM) and your engineers have nfi what's going on then you are lining your s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | « OpenAI s’est comporté comme un gangster » : comment l’IA a dépossédé des chercheurs de leur travail pour résoudre une équation réputée in… | « OpenAI s’est comporté comme un gangster » : comment l’IA a dépossédé des chercheurs de leur travail pour résoudre une équation réput… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Very nice to see something written so clearly and in short form. It does not, in fact, have to be complicated! blogs.gnome.org/alatiera/202… | Very nice to see something written so clearly and in short form. It does not, in fact, have to be complicated! blogs.gnome.org/alatier… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This triggered by the Graveyard keeper 2 devs saying they didn't use AI but do tinker with every AI model they can find, I'm not accusing t… | This triggered by the Graveyard keeper 2 devs saying they didn't use AI but do tinker with every AI model they can find, I'm not accus… | ↗Not briefed |
rss techmeme | 1mo | Portuguese surveillance drone maker Tekever raised $580M ed by UC Investments at a $6.4B valuation, after securing a UK MOD contract worth £400M over 10 years (Henrique Almeida/Bloomberg) | Henrique Almeida / Bloomberg: Portuguese surveillance drone maker Tekever raised $580M ed by UC Investments at a $6.4B valuation, afte… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The "Drake with a goth baddie" thing is a SEO/LLM fuckery to push all the Kendrick-related stuff online further down and out of the zeitgei… | The "Drake with a goth baddie" thing is a SEO/LLM fuckery to push all the Kendrick-related stuff online further down and out of the ze… | ↗Not briefed |
arxiv arxiv-ai | 1mo | The Path Matters: Evaluating Small Language Models Beyond Answer Accuracy in KGQA | Small language models (SLMs) are increasingly paired with knowledge graphs (KGs), yet end-to-end KG question answering conflates graph… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Robust Adversarial Reinforcement Learning with Risk Sensitivity and Critic Consistency Regularization | Reinforcement learning (RL) achieves strong performance in sequential decision-making but remains brittle under dynamic uncertainty an… | ↗Not briefed |
hn hn-ai | 1mo | GPT-6 Astra has made a major breakthrough in the Goldbach Conjecture | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Evolutionary Stability Does Not Guarantee Learning Accessibility: A Multi-Agent Reinforcement Learning Perspective on Cooperation Emergence | Cooperation emergence is a central problem in multi-agent systems because decentralized agents must coordinate while adapting to the c… | ↗Not briefed |
hn hn-ai | 1mo | A thread of early explorations with Claude Opus 5.5 | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Apple Intelligence Taking up 30GB+ on Some Macs Running macOS 27 | Apple Intelligence Taking up 30GB+ on Some Macs Running macOS 27 | ↗Not briefed |