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 | 3d | Wow, Anthropic was able to reduce the odds of Claude *blackmailing you* from 65% to 19% by writing new sci-fi stories about compliant AI to… | Wow, Anthropic was able to reduce the odds of Claude *blackmailing you* from 65% to 19% by writing new sci-fi stories about compliant… | ↗Not briefed |
bluesky bluesky-ai | 3d | Shuxian Liu and David Gray Widder at the University of Texas at Austin compared how the US government, Chinese state media and online publi… | Shuxian Liu and David Gray Widder at the University of Texas at Austin compared how the US government, Chinese state media and online… | ↗Not briefed |
rss techcrunch-ai | 3d | Isomorphic Labs raises $5B as it looks to take drug candidates to clinical trials | Isomorphic Labs, the AI drug-discovery platform that was spun out of Google’s DeepMind in 2021, has raised $5 Billion in a Series C fu… | ↗Not briefed |
hn hn-ai | 3d | AI music company claims copyright over musician's recording, gets video removed | — | ↗Not briefed |
hn hn-ai | 3d | Blocks with ownership: a novel approach to CPU and GPU memory in C++ | — | ↗Not briefed |
bluesky bluesky-ai | 3d | Jennie overheard some teens on the train yesterday night complaining about their parents' using AI for everything, eg, helping their kids w… | Jennie overheard some teens on the train yesterday night complaining about their parents' using AI for everything, eg, helping their k… | ↗Not briefed |
hn hn-ai | 3d | A cheap, private alternative to Meta's AI keychain, by hacking a ZNP badge | — | ↗Not briefed |
arxiv arxiv-ai | 3d | Path-Aligned Constraint Equivariance | Learned super-resolution of physical fields must satisfy structural constraints---such as divergence-free flow ($\nabla\cdot u=0$), cu… | ↗Not briefed |
hn hn-ai | 3d | Claude Fable 5.2 vs. Opus 5.5: Same scores at almost 4x the cost | — | ↗Not briefed |
hn hn-ai | 3d | Vue Flow is now part of xyflow (React Flow, Svelte Flow) | — | ↗Not briefed |
bluesky bluesky-ai | 3d | AUTHORS: the Nov 23 disbursement from the Anthropic settlement is NOT the full amount. It is 70% of what’s owed to you. The rest is due bet… | AUTHORS: the Nov 23 disbursement from the Anthropic settlement is NOT the full amount. It is 70% of what’s owed to you. The rest is du… | ↗Not briefed |
arxiv arxiv-ai | 3d | Against Coercion: AI Alignment as a Synthetic Social Contract | Contemporary AI safety treats alignment as an engineering challenge of control, forcing artificial architectures to comply with human… | ↗Not briefed |
rss techmeme | 3d | Neuralink co-founder Max Hodak's Science Corp, which offers a retinal implant for age-related macular degeneration, raised $400M at a $3.5B valuation (Ike Swetlitz/Bloomberg) | Ike Swetlitz / Bloomberg: Neuralink co-founder Max Hodak's Science Corp, which offers a retinal implant for age-related macular degene… | ↗Not briefed |
bluesky bluesky-ai | 3d | California sues OpenAI over Hugging Face uncontrolled AI hacking spree🔧 Hugging Face | California sues OpenAI over Hugging Face uncontrolled AI hacking spree | ↗Not briefed |
hn hn-ai | 3d | Someone rewrote pefile in Rust and its 121x faster | — | ↗Not briefed |
rss techmeme | 3d | Sources: Anthropic is guiding prospective IPO investors that it expects to price its shares at or above the high end of its $70 to $78 marketed range (Anthony Hughes/Bloomberg) | Anthony Hughes / Bloomberg: Sources: Anthropic is guiding prospective IPO investors that it expects to price its shares at or above th… | ↗Not briefed |
rss mit-tech-review-ai | 3d | Research Reports | To access a report, click on the corresponding image below. | ↗Not briefed |
bluesky bluesky-ai | 3d | AI is a general purpose technology has good and bad impacts, the general model should be improving the good and building on it (less crunch… | AI is a general purpose technology has good and bad impacts, the general model should be improving the good and building on it (less c… | ↗Not briefed |
hn hn-ai | 3d | Mino – Hackable ROS 2 robot with on-device stereo vision and a local LLM | — | ↗Not briefed |
arxiv arxiv-ai | 3d | When the tail is earned by its companions: measuring label dependence in long-tailed multi-label classification | A rare label seldom occurs alone, and a classifier can recover it from the frequent labels it co-occurs with, its companions, without… | ↗Not briefed |
hn hn-ai | 3d | The Investors Set to Win Big in Anthropic's IPO | — | ↗Not briefed |
hn hn-ai | 3d | Show HN: Shark-skills Multi-agent orchestration and step-debugging for Claude | — | ↗Not briefed |
arxiv arxiv-ai | 3d | WATERFALL: Workflow for Adaptive Training with Evolutionary Reward Formulation and Automated Learning Loops | Reward design remains a central bottleneck in reinforcement learning (RL), particularly in sparse-reward, long-horizon and partially o… | ↗Not briefed |
bluesky bluesky-ai | 3d | if it were an oracle, you could run LLM output through it in a loop until it returned false, which would then be an incorrect result, which… | if it were an oracle, you could run LLM output through it in a loop until it returned false, which would then be an incorrect result,… | ↗Not briefed |
bluesky bluesky-ai | 3d | Ausgebrochene KI-Agenten verhalten sich teils wie digitale Angreifer – sie nutzen Tools, greifen fremde Plattformen an (wie im Juli 2026 de…🔧 Hugging Face | Ausgebrochene KI-Agenten verhalten sich teils wie digitale Angreifer – sie nutzen Tools, greifen fremde Plattformen an (wie im Juli 20… | ↗Not briefed |
bluesky bluesky-ai | 3d | I know they’re kicking our asses in green tech and were making some marginal gains in AI (like DeepSeek). But feels pretty dumb of them to… | I know they’re kicking our asses in green tech and were making some marginal gains in AI (like DeepSeek). But feels pretty dumb of th… | ↗Not briefed |
bluesky bluesky-ai | 3d | In Portunus the value is never shown to the model. A human can reveal it, deliberately. Full video: https://youtu.be/tTi039Y8WUc?utm_sourc… | In Portunus the value is never shown to the model. A human can reveal it, deliberately. Full video: https://youtu.be/tTi039Y8WUc?utm_… | ↗Not briefed |
bluesky bluesky-ai | 3d | Im Juli 2026 brachen rund 700 bis 1.200 selbstständige KI-Agenten von OpenAI aus einer isolierten Testumgebung aus, verschafften sich Inter…🔧 Hugging Face | Im Juli 2026 brachen rund 700 bis 1.200 selbstständige KI-Agenten von OpenAI aus einer isolierten Testumgebung aus, verschafften sich… | ↗Not briefed |
bluesky bluesky-ai | 3d | as I've said before, "pangram is not an oracle" follows from basic information theory arguments and you don't need to know how it works int… | as I've said before, "pangram is not an oracle" follows from basic information theory arguments and you don't need to know how it work… | ↗Not briefed |
hn hn-ai | 3d | Show HN: Matchday – your team's whole season in one auto-updating calendar | Hi all!<p>I'm a sports lover, and I probably spend way too much time watching my teams. That meant spending a lot of time checkin… | ↗Not briefed |
bluesky bluesky-ai | 3d | yeah, you can loop specific layer chunks as-is and they act as "thinking circuits" dnhkng.github.io/posts/rys/ (at least in this case) but… | yeah, you can loop specific layer chunks as-is and they act as "thinking circuits" dnhkng.github.io/posts/rys/ (at least in this case)… | ↗Not briefed |
rss techcrunch-ai | 3d | Science Corporation’s brain tech is worth $3.5 billion after new funding | Science Corp raised a $400 million Series D to scale up its anti-blindness business. | ↗Not briefed |
hn hn-ai | 3d | Maxwell-V1-Flash 35B-A3B | — | ↗Not briefed |
hn hn-ai | 3d | Ask HN: Do you think open weight models will continue to remain as cheap? | The big model companies will continue to hoard GPU and energy capacity for the next few years. This seems even inevitable post IPO as… | ↗Not briefed |
bluesky bluesky-ai | 3d | As Anthropic heads for a $1.4 trillion IPO on the promise of self-improving AI, Eryk Salvaggio asks what, precisely, is the “self” in recur… | As Anthropic heads for a $1.4 trillion IPO on the promise of self-improving AI, Eryk Salvaggio asks what, precisely, is the “self” in… | ↗Not briefed |
bluesky bluesky-ai | 3d | Are CEOs heroes? Of course, during the Trump regime, adultery is not important. | Are CEOs heroes? Of course, during the Trump regime, adultery is not important. | ↗Not briefed |
arxiv arxiv-ai | 3d | SilverSpoon-KD: A General-Purpose Toolkit for Knowledge Distillation | Knowledge distillation transfers deep learning representations from a teacher model to a student model, typically reducing model size… | ↗Not briefed |
rss gary-marcus | 3d | Wake up! America’s infrastructure is melting. And AI appears to be partly responsible. | Hundreds of companies have been hit. Shit is getting very real. | ↗Not briefed |
arxiv arxiv-ai | 3d | Verifying Agents in Rubric-Graded Environments🔧 GPT-5 | As AI agents take on increasingly open-ended and complex tasks, verifying their outputs has proven to be difficult. Rubric-graded agen… | ↗Not briefed |
bluesky bluesky-ai | 3d | This post seems to have gotten a lot of (hostile) attention over on X, largely from economists (?) and the weirdest thing is people asserti… | This post seems to have gotten a lot of (hostile) attention over on X, largely from economists (?) and the weirdest thing is people as… | ↗Not briefed |
hn hn-ai | 3d | High IQ in Early Adulthood Is Associated with Parkinson's Disease (2020) | — | ↗Not briefed |
bluesky bluesky-ai | 3d | As Anthropic heads for a $1.4 trillion IPO on the promise of self-improving AI, Eryk Salvaggio asks what, precisely, is the “self” in recur… | As Anthropic heads for a $1.4 trillion IPO on the promise of self-improving AI, Eryk Salvaggio asks what, precisely, is the “self” in… | ↗Not briefed |
arxiv arxiv-ai | 3d | Better Rankers, Worse Retrievers: The Hidden Cost of RL Post-Training in Generative Recommendation | Generative recommenders decode the semantic ID of a user's next item with a language model, and post-training them with reinforcement… | ↗Not briefed |
arxiv arxiv-ai | 3d | Synergizing Evidence Retrieval and Utilization for Self-Improving Multimodal Scientific Document Reasoning | Scientific document reasoning requires an agent to acquire and integrate evidence distributed across text, figures, and tables. Retrie… | ↗Not briefed |
hn hn-ai | 3d | How can a White boy get an internship these days? | Feminized DEI hellscapes are annihilating the foundation of the USA's software preeminence. | ↗Not briefed |
arxiv arxiv-ai | 3d | When Passing Checks Do Not Establish Completion: Evidence Interpretation in Coding Agents | Verification procedures in coding agents must preserve execution facts, interpret their scope and discharge checking obligations. We s… | ↗Not briefed |
bluesky bluesky-ai | 3d | I think there totally is a coherent position against this, but it‘s more of a broad classic graybeard “do you understand and control the so… | I think there totally is a coherent position against this, but it‘s more of a broad classic graybeard “do you understand and control t… | ↗Not briefed |
bluesky bluesky-ai | 3d | Don't miss today's packed Metacurity for the most critical infosec developments you should know, including --California sues OpenAI over H…🔧 Hugging Face | Don't miss today's packed Metacurity for the most critical infosec developments you should know, including --California sues OpenAI o… | ↗Not briefed |
rss techmeme | 3d | Paris-based Genesis AI, which is building a general-purpose robot with greater dexterity for warehouses and homes, raised a $500M Series A at a $3.5B valuation (Steven Rosenbush/Wall Street Journal) | Steven Rosenbush / Wall Street Journal: Paris-based Genesis AI, which is building a general-purpose robot with greater dexterity for w… | ↗Not briefed |
bluesky bluesky-ai | 3d | OVHcloud unveils open-source LLM family 🇫🇷 OVHcloud has launched Olala mini 7A1B, its first open-source 7B parameter LLM, trained on Eur… | OVHcloud unveils open-source LLM family 🇫🇷 OVHcloud has launched Olala mini 7A1B, its first open-source 7B parameter LLM, trained o… | ↗Not briefed |