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 |
|---|---|---|---|---|
arxiv arxiv-ai | 1mo | Surviving the Router: Optimizing Skill Injections for Retrieval and Execution | AI agents increasingly rely on modular third-party "skills" that are dynamically selected by skill routers to execute complex tasks. W… | ↗Not briefed |
arxiv arxiv-ai | 1mo | When Tools Lie: Reliability of Mathematical Agents Under Corrupted Tool Feedback | Mathematical problem solving often requires deterministic computational steps that agents delegate to tools and implicitly trust. Yet… | ↗Not briefed |
rss techmeme | 1mo | AMD CEO Lisa Su says the company plans to "substantially increase" supply in 2027 and needs "more advanced wafer capacity", as she meets Foxconn and TSMC (Wen-Yee Lee/Reuters) | Wen-Yee Lee / Reuters: AMD CEO Lisa Su says the company plans to “substantially increase” supply in 2027 and needs “more advanced wafe… | ↗Not briefed |
hn hn-ai | 1mo | Is pnp-baremetal/pnp-esp32 1.5.0 production ready? | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Natural Language Questions as an Interface for Knowledge Graphs: QRAKEN Graph Distillation and Semantic Self-Healing🔧 GPT-5 | Natural-language access to RDF knowledge graphs is a core Semantic Web ambition. Large language models (LLMs) have advanced Text-to-SP… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SAGE: Semantic Anchor-Guided Evolution for Grounded Medical QA Data Synthesis | Developing reliable models for clinical tasks, such as Medical Question Answering (QA), is severely constrained by the limited availab… | ↗Not briefed |
hn hn-ai | 1mo | I built a GUI for managing systemd-nspawn containers | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | C'est l'heure du RDV Tech ! Avec @cedric2luca.bsky.social et @korben.info on couvre: - La conf Meta (Muse & Meta VR) - La conf OpenAI (Dots… | C'est l'heure du RDV Tech ! Avec @cedric2luca.bsky.social et @korben.info on couvre: - La conf Meta (Muse & Meta VR) - La conf OpenAI… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Explainable Rule Mining of IPv6 Extension-Header Presence Patterns from Paired-Vantage Captures | IPv6 extension headers (EHs), such as fragmentation, segment routing, and in-situ telemetry, are operationally important yetwidely dro… | ↗Not briefed |
arxiv arxiv-ai | 1mo | When Plans Change Answers: Formalizing Cost-Accuracy Optimization for Semantic Queries | In semantic query engines, predicates are evaluated by machine-learned models, and the choice of a query plan affects not only the cos… | ↗Not briefed |
arxiv arxiv-ai | 1mo | DirectSpeech2LLM: A Simple End-to-End Framework to Mitigate Prompt Overfitting in Speech-LLMs | Speech-LLMs often exhibit prompt overfitting, where models solely trained on automatic speech recognition (ASR) instruction fail to ge… | ↗Not briefed |
bluesky bluesky-ai | 1mo | How AI model collapse works in real time. Technically, I describe the problem here www.theregister.com/software/202... Or you could jus… | How AI model collapse works in real time. Technically, I describe the problem here www.theregister.com/software/202... Or you coul… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: A complete technical SEO audit in an MCP wrapper | Install, and sync with your google search console - analytses from a local SQlite database dependency. Would love your comments. | ↗Not briefed |
arxiv arxiv-ai | 1mo | POLAR: Ontology-Guided Risk Prevention for Tool-Calling LLM Agents | LLM tool-use agents operate in dynamic environments where many actions carry operational risk. However, most safety mechanisms react o… | ↗Not briefed |
hn hn-ai | 1mo | Tools Are Getting Better. Are We? | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | At an Australian hearing, OpenAI Chief Strategy Officer Jason Kwon pledged faster safety incident disclosures; Anthropic representatives ma… | At an Australian hearing, OpenAI Chief Strategy Officer Jason Kwon pledged faster safety incident disclosures; Anthropic representativ… | ↗Not briefed |
rss techmeme | 1mo | At an Australian hearing, OpenAI Chief Strategy Officer Jason Kwon pledged faster safety incident disclosures; Anthropic representatives made similar pledges (ABC) | ABC: At an Australian hearing, OpenAI Chief Strategy Officer Jason Kwon pledged faster safety incident disclosures; Anthropic represen… | ↗Not briefed |
hn hn-ai | 1mo | Predicting Peace | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | For example; Anthropic looks like it’s going to get a $2tn valuation at IPO, when from all the external evidence it’s an industry that is s… | For example; Anthropic looks like it’s going to get a $2tn valuation at IPO, when from all the external evidence it’s an industry that… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Kunde: Aus Datenschutzgründen setzen wir auf ein lokales LLM, das bei einem europäischen Provider gehosted wird. Das lokale LLM: OpenAI GP…🔧 GPT-5 | Kunde: Aus Datenschutzgründen setzen wir auf ein lokales LLM, das bei einem europäischen Provider gehosted wird. Das lokale LLM: Open… | ↗Not briefed |
rss stratechery | 1mo | Game Decompilation, Is This Legal?, A Well-Trodden Path | Games are being decompiled, but the real risk to gaming is new games and increased personalization. | ↗Not briefed |
arxiv arxiv-ai | 1mo | ProximalFM: Amortized Proximal Causal Inference under Hidden Confounding | Standard causal identification methods often assume no unmeasured confounding and can fail when relevant confounders are unobserved. P… | ↗Not briefed |
hn hn-ai | 1mo | Auto-review of agent actions without synchronous human oversight | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Self-Retrospection Distillation: Turning Post-hoc Experiences into Prior Foresight | Reinforcement learning with verifiable rewards (RLVR) turns agent experience into learning signals primarily through scalar outcome re… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SpeedrunBench: Challenging LLM Agents with Video Game Speedrunning | Frontier LLM agents have been shown to be capable of solving increasingly complex tasks for which humans have measurable solutions. Th… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Optimization Encoders: Rethinking Second-Order Meta-Learning for Neural Fields | Conditional neural fields represent signals continuously, but their effectiveness depends on how the conditional latent representation… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 'Australia's adoption rate of OpenAI's ChatGPT, Mr Kwon did not have an answer for why some polls show that a majority of the nation does n…🔧 ChatGPT | 'Australia's adoption rate of OpenAI's ChatGPT, Mr Kwon did not have an answer for why some polls show that a majority of the nation d… | ↗Not briefed |
hn hn-ai | 1mo | Payload CMS | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Acceptodds a prediction market on which ICLR 2027 papers get accepted | Hi, I built acceptodds, a prediction market on conference peer review. Every ICLR 2027 submission (42k papers) has a market: Accept vs… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Detecting a Shift Is Not Enough: Exact Minimax Limits of Linear Representation Repair | A mean shift between two data sources can be easy to detect but hard to remove without substantially changing their representations. W… | ↗Not briefed |
hn hn-ai | 1mo | Language Models That Play Chess and Explain Their Moves | — | ↗Not briefed |
hn hn-ai | 1mo | KI-LLM Has Reached the Pareto Frontier of Recklessness | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Parity,check if a smaller model matches your fine-tuned one right? | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | South Korea plans to launch a ~$3.5B program to develop a frontier AI model starting as early as March 2027 and will open a competition to … | South Korea plans to launch a ~$3.5B program to develop a frontier AI model starting as early as March 2027 and will open a competitio… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Days before the Vatican launched Magnifica Humanitas, Anthropic’s Chris Olah saw an advance copy and proposed withdrawing over its stance o… | Days before the Vatican launched Magnifica Humanitas, Anthropic’s Chris Olah saw an advance copy and proposed withdrawing over its sta… | ↗Not briefed |
rss techmeme | 1mo | South Korea plans to launch a ~$3.5B program to develop a frontier AI model starting as early as March 2027 and will open a competition to select participants (Jie Ye-eun/The Korea Herald) | Jie Ye-eun / The Korea Herald: South Korea plans to launch a ~$3.5B program to develop a frontier AI model starting as early as March… | ↗Not briefed |
arxiv arxiv-ai | 1mo | HINTT Submission to the 2nd MLC-SLM Challenge: Comparing Cascaded and Unified Approaches to Diarization and ASR | This paper presents the HINTT system submitted to the 2nd Challenge and Workshop on Multilingual Conversational Speech Language Model… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Parece ser que Artificial es cojonuda y que va a hacer por Altman lo que «La red social» hizo por Zuckerberg: llevar a las masas la certeza… | Parece ser que Artificial es cojonuda y que va a hacer por Altman lo que «La red social» hizo por Zuckerberg: llevar a las masas la ce… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Language Carries the Expert's Impression: Instrument-Anchored LLM Judges Transfer Counseling-Quality Assessment and Beat In-Domain Training | Automatic assessment of communication quality in dyadic counseling conversations is bottlenecked by data: expert-rated corpora are sma… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Google did not share its latest Gemini 4 Argon AI model with the U.K. government’s AI Security Institute for safety testing ahead of its in… | Google did not share its latest Gemini 4 Argon AI model with the U.K. government’s AI Security Institute for safety testing ahead of i… | ↗Not briefed |
rss techmeme | 1mo | The UK's Ofcom opens an OSA probe into Meta over Instagram Instants, launched in May; Meta says it "briefed Ofcom about this feature on a number of occasions" (Reuters) | Reuters: The UK's Ofcom opens an OSA probe into Meta over Instagram Instants, launched in May; Meta says it “briefed Ofcom about this… | ↗Not briefed |
arxiv arxiv-ai | 1mo | A Riemannian Geometry for Low-rank Adaptation | Low-rank adaptation (LoRA) is widely used as a parameter-efficient fine-tuning technique for pre-trained deep neural networks, which a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | DAEDALUS: Bootstrapping Agent Memory from Self-Generated Tasks | LLM agents often lack the operational knowledge to act reliably in new environments, as they must discover specific tool behaviors or… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SepsisLens: Structure-Preserving Sequence Modelling for Decomposable Early Sepsis Warning | Early sepsis warning from ICU records can be cast as a structure-preserving prediction problem. A model needs to detect deterioration… | ↗Not briefed |
hn hn-ai | 1mo | Peter Norvig on AI's Future and the Limits of Trust | — | ↗Not briefed |
hn hn-ai | 1mo | LittleJS Example: Make your own Minecraft | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | So my understanding from the "leak" of the Anthropic Financials is they are saying they will never recover the investment at the levels nee… | So my understanding from the "leak" of the Anthropic Financials is they are saying they will never recover the investment at the level… | ↗Not briefed |
bluesky bluesky-ai | 1mo | As critical as I am of *gestures wildly at everything*, I don't actually believe that LLM assistance in projects is an automatic disqualifi… | As critical as I am of *gestures wildly at everything*, I don't actually believe that LLM assistance in projects is an automatic disqu… | ↗Not briefed |
bluesky bluesky-ai | 1mo | A reminder that by far the largest use case for LLM-based AI products is "therapy and companionship". The Epstein class's hostile takeover… | A reminder that by far the largest use case for LLM-based AI products is "therapy and companionship". The Epstein class's hostile tak… | ↗Not briefed |
bluesky bluesky-ai | 1mo | oh we should do this thing and this other thing and enact extremely harmful and costly processes to endow presumptive innocence and benefit… | oh we should do this thing and this other thing and enact extremely harmful and costly processes to endow presumptive innocence and be… | ↗Not briefed |