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 | 1mo | Partial Evaluation in Truffle | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Un dernier point : un journaliste qui voudrait vraiment faire un papier anglé ainsi aurait certainement eu comme réflexe de solliciter un/d… | Un dernier point : un journaliste qui voudrait vraiment faire un papier anglé ainsi aurait certainement eu comme réflexe de solliciter… | ↗Not briefed |
rss the-decoder | 1mo | Three firings and a fourth departure shake up OpenAI's safety team | OpenAI has parted ways with three researchers who allegedly leaked confidential information to an outside AI safety organization, acco… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Collective Bias Mitigation via Model Routing and Collaboration | Large language models (LLMs) are increasingly deployed in public health, finance, and governance, requiring both accuracy and societal… | ↗Not briefed |
hn hn-ai | 1mo | Specialized Bibliographic Databases | — | ↗Not briefed |
hn hn-ai | 1mo | Ask HN: What to do with lots of spare compute in colocation? | I moved my homelab into a colocation rack (proper DC deployment with redundant power) a while ago. Things have been running smoothly,… | ↗Not briefed |
bluesky bluesky-ai | 1mo | They would have never passed muster had they been held to even the most basic standards. All of these STEM PhD candidates, on full-ride res… | They would have never passed muster had they been held to even the most basic standards. All of these STEM PhD candidates, on full-rid… | ↗Not briefed |
arxiv arxiv-ai | 1mo | WAMpy: Efficient Synthesis of Prolog Programs in Python | We present WAMpy, a Python framework optimized for synthesizing Prolog programs. Unlike general-purpose Prolog systems, WAMpy targets… | ↗Not briefed |
arxiv arxiv-ai | 1mo | D2K-Bench: Can LLM Agents Turn Expert Designs into Efficient GPU Kernels?🔧 GPT-5 | GPU kernels generated by large language model (LLM) agents can remain less efficient than expert implementations, but runtime alone do… | ↗Not briefed |
arxiv arxiv-ai | 1mo | The Neuro-Physical Inverter: A Modular Framework for Magnetotelluric Inversion Coupling Ensemble Conditioning with Residual Learning | We present the Neuro-Physical Inverter (NPI), a modular, uncertainty-aware framework for geophysical inversion that couples ensemble-b… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Based on this my suspicion is that they fed their conspiracy theory (about coordination with Fani Willis) to an LLM and asked it to look fo… | Based on this my suspicion is that they fed their conspiracy theory (about coordination with Fani Willis) to an LLM and asked it to lo… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Uncertainty as a Proxy for Semantic Correctness in Diffusion-Based Medical Image Synthesis | Diffusion models can synthesise contrast-enhanced CT (CECT) from non-contrast CT (NCCT), avoiding contrast administration and its envi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | It’s hard to get excited about AI model benchmarks anymore. Claude Opus 5 had high benchmark scores but was unusably bad. Muse is the top… | It’s hard to get excited about AI model benchmarks anymore. Claude Opus 5 had high benchmark scores but was unusably bad. Muse is th… | ↗Not briefed |
arxiv arxiv-ai | 1mo | AdaStep: Adaptive Step Credit Weighting for Agentic Reinforcement Learning | Long-horizon LLM agents are typically trained with sparse outcome rewards, making trajectory-level objectives too coarse to distinguis… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I get why people think so much of the grandiose statements from LLM world are a grift (because everything is these days). But I think it's… | I get why people think so much of the grandiose statements from LLM world are a grift (because everything is these days). But I think… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Near-Optimal Convex Optimization with Lazy Second-Order Oracles | This paper studies the complexity of convex optimization using lazy second-order oracles (Doikov, Chayti, and Jaggi, ICML 2023), where… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Built a little helper for answering questions while gaming | I've built wizbert because I was constantly switching screens from Valheim to a wiki page, and got annoyed with it. So I build a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Evolving Hybrid Quantum-Classical Architectures for Image Classification | Hybrid quantum classical neural networks integrate parameterized quantum circuits (PQCs) with established deep learning architectures,… | ↗Not briefed |
bluesky bluesky-ai | 1mo | “ai manages to solve millennium prize problem!” look, if you pour this amount of money onto a group of mathematicians to solve the problem,… | “ai manages to solve millennium prize problem!” look, if you pour this amount of money onto a group of mathematicians to solve the pro… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Ugh fucking AI is ruining my academic piracy alerts now | Ugh fucking AI is ruining my academic piracy alerts now | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic, qui s'apprête à devenir l'une des plus grosses capitalisations boursières de la planète, est une entreprise transhumaniste, qui … | Anthropic, qui s'apprête à devenir l'une des plus grosses capitalisations boursières de la planète, est une entreprise transhumaniste,… | ↗Not briefed |
hn hn-ai | 1mo | WSP WordPress MCP – Connect AI Agents to WordPress | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | futurism.com/artificial-i... AI companies have done next to nothing to earn their trust. Tech companies like OpenAI and Meta would love us… | futurism.com/artificial-i... AI companies have done next to nothing to earn their trust. Tech companies like OpenAI and Meta would lo… | ↗Not briefed |
bluesky bluesky-ai | 1mo | if I was an ai company, burning through unprecedented amounts of cash without a clear business model, I’d also hire scientists to do PR for… | if I was an ai company, burning through unprecedented amounts of cash without a clear business model, I’d also hire scientists to do P… | ↗Not briefed |
bluesky bluesky-ai | 1mo | donotsta.re/objects/3c86... | donotsta.re/objects/3c86... | ↗Not briefed |
arxiv arxiv-ai | 1mo | Kernel Singular Value Decomposition with Extension to Multiple Data Sources | Kernel Singular Value Decomposition (KSVD) learns a pair of singular vectors w.r.t. an asymmetric kernel matrix, which can be induced… | ↗Not briefed |
arxiv arxiv-ai | 1mo | StanceEval 2026: The Second Stance Detection Shared Task | StanceEval 2026 is the second edition of the StanceEval shared task series on stance detection in Arabic social media text. Stance det… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Toward SLM-based agentic task-tool intent matching | Tool-equipped AI agents use tool calls to access data and act on external systems. Horizontal growth of agentic systems increases the… | ↗Not briefed |
arxiv arxiv-ai | 1mo | HyperFuse: Fast Self-Supervised Node Embeddings for Attributed Hypergraphs | Self-supervised hypergraph representation learning can produce informative node embeddings, but existing methods often require deep en… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI publicly acknowledged the security flaw and fix in its system change log on September 25. “We continue to evolve our security practi… | OpenAI publicly acknowledged the security flaw and fix in its system change log on September 25. “We continue to evolve our security p… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Bare trælseren når sælgelsen ikke har virkeren på. ...kunne The Nordic Ticker dog ikke i det mindste bruge en LLM, der kan skrive bare li… | Bare trælseren når sælgelsen ikke har virkeren på. ...kunne The Nordic Ticker dog ikke i det mindste bruge en LLM, der kan skrive ba… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Hamiltonian locality testing and certification do not achieve the Heisenberg limit | We establish lower bounds for Hamiltonian property testing with access to the time-evolution operator but not its inverse. Each experi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | A Anthropic está planejando uma abertura de capital (IPO) e protocolou o rascunho do prospecto (S-1) em junho. A situação financeira da Ant… | A Anthropic está planejando uma abertura de capital (IPO) e protocolou o rascunho do prospecto (S-1) em junho. A situação financeira d… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Contextual Flow Matching: Adaptive Step Selection in Flow Models for Efficient Visual Generation | Flow Matching enables high-quality visual generation via continuous-time dynamics, but inference remains costly due to multiple sequen… | ↗Not briefed |
hn hn-ai | 1mo | Most recent AI-assisted coder makes first AI-assisted coder run again | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | The National AI Charter Act draws on financial regulation as a model. @grahamsteele.bsky.social recently wrote about the important lessons … | The National AI Charter Act draws on financial regulation as a model. @grahamsteele.bsky.social recently wrote about the important les… | ↗Not briefed |
bluesky bluesky-ai | 1mo | NEW: @jayapal.house.gov has released the National AI Charter Act, a framework for governing AI. It’s encouraging to see a framework that c… | NEW: @jayapal.house.gov has released the National AI Charter Act, a framework for governing AI. It’s encouraging to see a framework t… | ↗Not briefed |
bluesky bluesky-ai | 1mo | There's something so grim about having to work so hard to prove your art is real only for it to not really matter because it's going to be … | There's something so grim about having to work so hard to prove your art is real only for it to not really matter because it's going t… | ↗Not briefed |
bluesky bluesky-ai | 1mo | • « établi un lien entre Jordan Bardella et une officine d'extrême droite » (idem, avec introduction de flou, assez typique des reformulati… | • « établi un lien entre Jordan Bardella et une officine d'extrême droite » (idem, avec introduction de flou, assez typique des reform… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Predictively Oriented Gaussian Process Posteriors | Gaussian Processes (GPs) are a powerful tool for modelling and quantifying uncertainty in functional relationships. However, they requ… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: A Catalogue of Unknown Creatures | Hi there,<p>I've always loved creature movies since I was a kid and I had the idea a few weeks ago to see how models would do try… | ↗Not briefed |
bluesky bluesky-ai | 1mo | LLM agents are evolving their own skills and working together in swarms. But how well do they learn from each other? Our new paper on Recu… | LLM agents are evolving their own skills and working together in swarms. But how well do they learn from each other? Our new paper on… | ↗Not briefed |
hn hn-ai | 1mo | Harvard particle physicist Matthew Schwartz drops 36 papers authored with Claude | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Predicting and Repairing Merge Collapse in Large Language Models | Large language models fine-tuned from a shared base can be merged by averaging their task vectors, but some merges collapse far below… | ↗Not briefed |
arxiv arxiv-ai | 1mo | KV$^2$: A Self-Refining KV Cache | The memory footprint of the key-value (KV) cache constrains the practical use of long-context models, and it dominates cost when one p… | ↗Not briefed |
bluesky bluesky-ai | 1mo | “Officials from companies including OpenAI & Anthropic said they will testify—but every company excluding Meta agreed to do so only after t… | “Officials from companies including OpenAI & Anthropic said they will testify—but every company excluding Meta agreed to do so only af… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Source Preference in the Wild: How LLM Agents Favor Items by Source, and How to Reduce It | As LLM agents decide on users' behalf which product to buy, which hotel to book, or which paper to cite, a preference for items from c… | ↗Not briefed |
bluesky bluesky-ai | 1mo | pryvero.net/blog/anthrop... #Crypto #BTC #XMR #Monero #Bitcoin #Ethereum #ETH | pryvero.net/blog/anthrop... #Crypto #BTC #XMR #Monero #Bitcoin #Ethereum #ETH | ↗Not briefed |
arxiv arxiv-ai | 1mo | Not Until the Evidence Says So: Teaching LLM Investigators When to Close a Case | Accident, defect and outage investigations end with a decision that ordinary question answering never faces: whether the evidence gath… | ↗Not briefed |
bluesky bluesky-ai | 1mo | like if you think this was Aisi, repost if you think it was METR | like if you think this was Aisi, repost if you think it was METR | ↗Not briefed |