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 |
|---|---|---|---|---|
hn hn-ai | 12d | Show HN: 3D to 2D, point clouds flattening (geometry images) | — | ↗Not briefed |
rss the-decoder | 12d | Tencent's AI chief says AI bioweapons are coming, but closing off models isn't the answer | Tencent chief AI scientist Vinces Yao Shunyu warns that AI will soon be able to develop bioweapons and argues for a middle ground betw… | ↗Not briefed |
bluesky bluesky-ai | 12d | OpenAI reports 948 working Linux root exploits from a prototype pipeline run costing $1,213,430 in API credits. WordPress and SonicWall ent… | OpenAI reports 948 working Linux root exploits from a prototype pipeline run costing $1,213,430 in API credits. WordPress and SonicWal… | ↗Not briefed |
bluesky bluesky-ai | 12d | @ramirez.house.gov Kathryn is excellent and would be happy to support with sample language for LLM logs as part of subpoenas :) | @ramirez.house.gov Kathryn is excellent and would be happy to support with sample language for LLM logs as part of subpoenas :) | ↗Not briefed |
arxiv arxiv-ai | 12d | Fairness in Explainable AI Narratives: Decomposing Social Disparities in LLM-Generated Explanations | Large language models (LLMs) are increasingly used to translate technical explainable AI (XAI) artefacts such as SHAP into explanatory… | ↗Not briefed |
hn hn-ai | 12d | Show HN: Lampjack - live 3D visuals that follow your music | — | ↗Not briefed |
arxiv arxiv-ai | 12d | TRACE: Adaptive Trust-Region Bayesian Optimization for Open-Ended Molecular Design | Optimizing molecules under small evaluation budgets requires deciding both where in chemical space to search and which candidate withi… | ↗Not briefed |
bluesky bluesky-ai | 12d | @jayapal.house.gov you’re one of my Reps, ask these dorks for their LLM logs when you’re subpoenaing them please! I can share exemplar lang… | @jayapal.house.gov you’re one of my Reps, ask these dorks for their LLM logs when you’re subpoenaing them please! I can share exemplar… | ↗Not briefed |
arxiv arxiv-ai | 12d | AgentOPSD: Turn-Level Credit Assignment via Recursive Belief Revision | Reinforcement learning with verifiable rewards ordinarily spreads a single advantage over a whole trajectory, a granularity too coarse… | ↗Not briefed |
bluesky bluesky-ai | 12d | Tthis is just one example among SO many examples: Sam Altman and OpenAI are currently bribing Chuck Schumer to get their way on the "LET US… | Tthis is just one example among SO many examples: Sam Altman and OpenAI are currently bribing Chuck Schumer to get their way on the "L… | ↗Not briefed |
rss techmeme | 12d | Draft law: the EU's Digital Fairness Act would require companies to let adults disable addictive features and will target AI bots that simulate relationships (Bloomberg) | Bloomberg: Draft law: the EU's Digital Fairness Act would require companies to let adults disable addictive features and will target A… | ↗Not briefed |
hn hn-ai | 12d | Rust Harness for Abliterated Models | — | ↗Not briefed |
bluesky bluesky-ai | 12d | Anthropic hat in der ersten Hälfte des Jahres zwar einen Umsatz von rund 16 Milliarden USD gemacht, gleichzeitig aber auch Verluste von cir… | Anthropic hat in der ersten Hälfte des Jahres zwar einen Umsatz von rund 16 Milliarden USD gemacht, gleichzeitig aber auch Verluste vo… | ↗Not briefed |
bluesky bluesky-ai | 12d | Die großen Tech-Konzerne wetten Billionen US-Dollar darauf, dass KI die Zukunft ist. Diese Wette wird aber zunehmend durch Schulden und Kre… | Die großen Tech-Konzerne wetten Billionen US-Dollar darauf, dass KI die Zukunft ist. Diese Wette wird aber zunehmend durch Schulden un… | ↗Not briefed |
hn hn-ai | 12d | The same H100 costs $2.69 or $12.29 an hour depending where you rent it | — | ↗Not briefed |
rss theverge-ai | 12d | Disney and Adobe are building generative AI tools together🔧 Adobe Firefly | Who wants a hand-designed and creatively curated monorail anyway. | Image: Disney / Adobe The Walt Disney Company has teamed up with… | ↗Not briefed |
bluesky bluesky-ai | 12d | Yeah. Like, some of this LLM stuff is technologically interesting, and part of me would love to work on it, but then these dudes open their… | Yeah. Like, some of this LLM stuff is technologically interesting, and part of me would love to work on it, but then these dudes open… | ↗Not briefed |
bluesky bluesky-ai | 12d | To bardzo dobry pomysł, żeby taki mural powstał. Obrazek z FB. I to jest dobre wykorzystanie LLM. | To bardzo dobry pomysł, żeby taki mural powstał. Obrazek z FB. I to jest dobre wykorzystanie LLM. | ↗Not briefed |
hn hn-ai | 12d | Chrome 153 has a native MP4 tab recorder | — | ↗Not briefed |
arxiv arxiv-ai | 12d | From Likelihood Convergence to Parameter Convergence in POMDPs | Likelihood-based methods are widely used to learn the parameters of partially observable Markov decision processes (POMDPs). While fin… | ↗Not briefed |
arxiv arxiv-ai | 12d | Evaluating Cross-Signal Coherence in Mammography Vision-Language Models | As vision-language models (VLMs) move toward interpreting medical images directly, their outputs increasingly include not just a label… | ↗Not briefed |
bluesky bluesky-ai | 12d | Token usage at OpenAI and Anthropic slipping in the last couple weeks. | Token usage at OpenAI and Anthropic slipping in the last couple weeks. | ↗Not briefed |
bluesky bluesky-ai | 12d | ‘In any mania there are always the small handful of weathervanes that point to the shifting winds as bubble goes to peak and then bust. In … | ‘In any mania there are always the small handful of weathervanes that point to the shifting winds as bubble goes to peak and then bust… | ↗Not briefed |
hn hn-ai | 12d | Show HN: One TUI to Manage MCP Servers for Claude Code and Codex | — | ↗Not briefed |
bluesky bluesky-ai | 12d | Mini-rant: I am so tired of getting slop-papers to review. I have no problem with people using AI to help with coding, writing, etc. I DO h… | Mini-rant: I am so tired of getting slop-papers to review. I have no problem with people using AI to help with coding, writing, etc. I… | ↗Not briefed |
rss theverge-ai | 12d | Hold up — I had a great time talking to the RIAA guy🔧 Udio🔧 FLUX🔧 Suno | Today, I’m talking with Mitch Glazier, who is CEO and chairman of the RIAA. This conversation was a real full-circle moment for me: I… | ↗Not briefed |
arxiv arxiv-ai | 12d | LawKAN: Group-Law Layers for Composition -- When Imposing a Known Law Pays Off | Many tasks compose transformations by a known rule: chained rotations in robotics and navigation. Learned models relearn it, compoundi… | ↗Not briefed |
bluesky bluesky-ai | 12d | And for bonus points, they especially like it if you start accusing him of pre-emptive corruption with his daughter now being a lobbyist fo… | And for bonus points, they especially like it if you start accusing him of pre-emptive corruption with his daughter now being a lobbyi… | ↗Not briefed |
bluesky bluesky-ai | 12d | for how much money & hype has gone into LLM chatbots the UI/UX situation is fucking criminal imho | for how much money & hype has gone into LLM chatbots the UI/UX situation is fucking criminal imho | ↗Not briefed |
bluesky bluesky-ai | 12d | here are the 45 findings Anthropic reported to #curl last week: https://github.com/curl/curl/issues?q=is%3Aissue+ANT-2026 | here are the 45 findings Anthropic reported to #curl last week: https://github.com/curl/curl/issues?q=is%3Aissue+ANT-2026 | ↗Not briefed |
arxiv arxiv-ai | 12d | WanderBench: Can World Models Explore Dynamic Worlds from Video? | Video world models conditioned on an input video make it possible to explore real, recorded dynamic scenes, enabling re-shooting of fo… | ↗Not briefed |
bluesky bluesky-ai | 12d | “Language is Now an Attack Surface.” My take on OpenAI and Hugging Face, in which I argue we don’t really know what text is anymore and i…🔧 Hugging Face | “Language is Now an Attack Surface.” My take on OpenAI and Hugging Face, in which I argue we don’t really know what text is anymore… | ↗Not briefed |
hn hn-ai | 12d | FairPlay Has Been Cracked | — | ↗Not briefed |
bluesky bluesky-ai | 12d | Supondo que a tecnologia de LLM amadureça, chegue num estágio de só ser usada quando necessário e tenha custos e degradação ambiental r… | Supondo que a tecnologia de LLM amadureça, chegue num estágio de só ser usada quando necessário e tenha custos e degradação ambien… | ↗Not briefed |
arxiv arxiv-ai | 12d | SciMARS: Multi-Agent Recursive Self-Improvement for Closed-Loop Scientific Discovery | We propose SciMARS, a framework for recursive task-local adaptation of multi-agent harnesses in closed-loop scientific discovery. Mult… | ↗Not briefed |
bluesky bluesky-ai | 12d | Synaptics threatens developer who used Claude AI to create open-source DisplayLink driver | Synaptics threatens developer who used Claude AI to create open-source DisplayLink driver | ↗Not briefed |
bluesky bluesky-ai | 12d | I’ve reviewed adversaries’ LLM logs in discovery and it is fucking mind blowing what people will tell their robot DoctorLawyerPsychicPriest… | I’ve reviewed adversaries’ LLM logs in discovery and it is fucking mind blowing what people will tell their robot DoctorLawyerPsychicP… | ↗Not briefed |
arxiv arxiv-ai | 12d | A Strict Upper Bound for Three-Point Weighted Selection in Planar Vector-Valued Regression | We prove $F_{\mathrm{weighted}}(2,2,3)<11/6$ for minimum-Frobenius-norm least-squares regression with two input and two output coordin… | ↗Not briefed |
bluesky bluesky-ai | 12d | somebody typed "in a hand painted font" once & got a good result, since LLMs are feedback loops now it's what everybody gets for that promp… | somebody typed "in a hand painted font" once & got a good result, since LLMs are feedback loops now it's what everybody gets for that… | ↗Not briefed |
hn hn-ai | 12d | The 79% Problem: Why the People Who Need Data Still Cannot Get It | — | ↗Not briefed |
hn hn-ai | 12d | Viper4Linux-Re – PipeWire Audio Enhancer | — | ↗Not briefed |
bluesky bluesky-ai | 12d | Llama 4 Maverick (17B-128E MoE) vs Llama 4 Scout (10M Context MoE): Which LLM API actually protects your SaaS gross margin at 10,000 DAU? …🔧 Llama 4 | Llama 4 Maverick (17B-128E MoE) vs Llama 4 Scout (10M Context MoE): Which LLM API actually protects your SaaS gross margin at 10,000 D… | ↗Not briefed |
rss techmeme | 12d | Medium launches TK, an AI-powered, minimalist note-taking, journaling, and private social networking app with daily prompts, its first major launch since 2017 (David Pierce/The Verge) | David Pierce / The Verge: Medium launches TK, an AI-powered, minimalist note-taking, journaling, and private social networking app wit… | ↗Not briefed |
bluesky bluesky-ai | 12d | you were like "she can't be a spy she loves LLM and nano robotics" | you were like "she can't be a spy she loves LLM and nano robotics" | ↗Not briefed |
arxiv arxiv-ai | 12d | Learning inexact alternating minimization for convex problems with quadratic coupling | Alternating minimization is a standard tool for optimization problems with two coupled variable blocks. Its cost is dominated by the i… | ↗Not briefed |
hn hn-ai | 12d | An LLM revived a 2016 game with one byte | — | ↗Not briefed |
bluesky bluesky-ai | 12d | Do I know anyone who has connections to the folks in Congress who are going to be writing subpoenas? If I do, can y’all tell those folks t… | Do I know anyone who has connections to the folks in Congress who are going to be writing subpoenas? If I do, can y’all tell those fo… | ↗Not briefed |
bluesky bluesky-ai | 12d | Lol so this is actually a very funny test and Claude is the ONLY ai that does not launch | Lol so this is actually a very funny test and Claude is the ONLY ai that does not launch | ↗Not briefed |
hn hn-ai | 12d | Perceptive Humanoid Motion from Simulation | — | ↗Not briefed |
rss huggingface-blog | 12d | Nemotron-Personas-Finance-Japan: A Synthetic Dataset Capturing Japan's Demographics and Financial Context | — | ↗Not briefed |