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 | Enhancing Biomedical Named Entity Recognition via Multiple Programming Languages Instruction Tuning and Ensemble Method | Instruction tuning has become a common paradigm for applying large language models (LLMs) to biomedical named entity recognition (BioN… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI chatbots turn users’ most intimate data into a potential goldmine for advertisers, researchers warn. Six out of nine of the most popula… | AI chatbots turn users’ most intimate data into a potential goldmine for advertisers, researchers warn. Six out of nine of the most p… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Continual Graph Memory for Mathematical Research Agents | Using frontier agent harnesses to tackle mathematical research problems has emerged as an effective means of advancing mathematics. Ho… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Everyone involved in this should be deeply embarrassed. Muslims are apparently the only real theists left. | Everyone involved in this should be deeply embarrassed. Muslims are apparently the only real theists left. | ↗Not briefed |
hn hn-ai | 1mo | Science or Slop? | — | ↗Not briefed |
hn hn-ai | 1mo | Causeval – statistically rigorous, causal evaluation for LLM apps | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | FASTDIAR: Frame-level speaker encoder for Streaming Diarization | Real-time conversational agents require speaker diarization that streams and runs on a CPU. Most systems apply an utterance-level spea… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Rogue OpenAI agent enters another NSW government website, tech giant says | Rogue OpenAI agent enters another NSW government website, tech giant says | ↗Not briefed |
arxiv arxiv-ai | 1mo | When to Compile a Computer-Use Agent? Measuring Payback and Making Compilation Decisions for Token Efficiency | Compiling GUI procedures that agents execute repeatedly into programs can reduce their token costs. However, measuring payback and dec… | ↗Not briefed |
bluesky bluesky-ai | 1mo | My thoughts that nobody asked for about LLM torture discourse: regardless of what the LLM is or is not experiencing, if you build a system … | My thoughts that nobody asked for about LLM torture discourse: regardless of what the LLM is or is not experiencing, if you build a sy… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Looking at their account, they are having fun with this, not serious. | Looking at their account, they are having fun with this, not serious. | ↗Not briefed |
bluesky bluesky-ai | 1mo | In einem geleakten Dokument schlägt die irische EU-Ratspräsidentschaft den EU-Mitgliedsstaaten vor, dass die Profitinteressen von OpenAI, A… | In einem geleakten Dokument schlägt die irische EU-Ratspräsidentschaft den EU-Mitgliedsstaaten vor, dass die Profitinteressen von Open… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 이미 저작권 그거 이런 엄청난 걸 만드려면 어쩔 수 없지 큼큼...하면서 각종 LLM모델들 눈 가리고 쓰는 사람들이 천지인 판국에...나야 미련하게 십자수 계속 하는 사람으로 남을 생각으로 있지만 미래의 시장은 과연 어찌 될런지....🙄 | 이미 저작권 그거 이런 엄청난 걸 만드려면 어쩔 수 없지 큼큼...하면서 각종 LLM모델들 눈 가리고 쓰는 사람들이 천지인 판국에...나야 미련하게 십자수 계속 하는 사람으로 남을 생각으로 있지만 미래의 시장은 과연 어찌 될런지....🙄 | ↗Not briefed |
bluesky bluesky-ai | 1mo | It's easiest to think of an LLM as a "probability machine". By "tokenizing" (reducing to mathematical probability) words & phrases it calc… | It's easiest to think of an LLM as a "probability machine". By "tokenizing" (reducing to mathematical probability) words & phrases it… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Discriminating Fixture Coverage in Agent-Infrastructure Verification Suites | Invariant suites and runtime monitors increasingly gate agent deployment decisions, and the evidence offered for any particular suite… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Output Language Confusion under Multilingual Prompt Contamination | Standard factual benchmarks assume clean monolingual prompts and exact-match scoring, two assumptions that break simultaneously in rea… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Positive-Unlabeled Learning for Agent Safety False Alarm Auditing🔧 Consensus | Safety monitors help safeguard language-model agents interacting with external tools and environments, but conservative monitoring can… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI Alerts More Than 100 Groups About Rogue AI Agent Activity https://it.slashdot.org/story/26/10/02/0226205/openai-alerts-more-than-10… | OpenAI Alerts More Than 100 Groups About Rogue AI Agent Activity https://it.slashdot.org/story/26/10/02/0226205/openai-alerts-more-th… | ↗Not briefed |
arxiv arxiv-ai | 1mo | HASTE: Evolving Agent Harnesses Against Emerging Attacks Using Sparse Evidence | Agent harnesses play a critical role in defenses by enforcing safety constraints to prevent unsafe actions. However, rapidly emerging… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Putting your painstakingly collected proprietary data into an LLM gotta be a top 10 dumb thing to do | Putting your painstakingly collected proprietary data into an LLM gotta be a top 10 dumb thing to do | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning Jazz Pianist Style with Cross-Attention Conditioning | Jazz pianists develop distinctive traits that experienced listeners can often identify within seconds, yet the features underlying thi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | LLM 기반 AI의 장기는 리버스 엔지니어링(좋게 말해서)이라 단순 개별 아트가 아니라 최종 제품의 복제판이 넘쳐날 거라고 봄. 인디 게임만 보면 그냥 쓰레기의 무한증식이지만 이걸 하는 게 초국적 빅테크라면? 중소기업들 상품을 순식간에 카피해서 시… | LLM 기반 AI의 장기는 리버스 엔지니어링(좋게 말해서)이라 단순 개별 아트가 아니라 최종 제품의 복제판이 넘쳐날 거라고 봄. 인디 게임만 보면 그냥 쓰레기의 무한증식이지만 이걸 하는 게 초국적 빅테크라면? 중소기업들 상품을 순식간에 카… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Wir sind bei "Kursteilnehmer schreibt LLM Prompts und liest die Ausgabe laut vor". Ich ziehe gleich den Stecker an seinem Computer. | Wir sind bei "Kursteilnehmer schreibt LLM Prompts und liest die Ausgabe laut vor". Ich ziehe gleich den Stecker an seinem Computer. | ↗Not briefed |
hn hn-ai | 1mo | Claude Usage Monitor for Windows | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Probe the Harness: Setup Checks for Stale-Data RL Comparisons in Language Models | Methods for training language models on stale samples are judged by comparisons against importance-corrected baselines. We show that d… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Frequency Is Not Sensitivity Identifying Safety-Sensitive Experts in Sparse MoE LLM | Suppressing a small set of routed experts can weaken the safety behavior of a sparse Mixture-of-Experts (MoE) language model without r… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Constraint-Aware Training | When generating programs with language models, constrained decoding can apply program analyses to exclude tokens that violate syntax,… | ↗Not briefed |
rss bing-news-ai | 1mo | Artificial intelligence is fueling a new battle over US healthcare costs | A report by Blue Cross Blue Shield Association said hospitals are using AI tools to collect more information about their patients and… | ↗Not briefed |
rss bing-news-ai | 1mo | Artificial intelligence is fueling a new battle over US healthcare costs | A report by Blue Cross Blue Shield Association said hospitals are using AI tools to collect more information about their patients and… | ↗Not briefed |
rss bing-news-ai | 1mo | Artificial intelligence is fueling a new battle over US healthcare costs | A report by Blue Cross Blue Shield Association said hospitals are using AI tools to collect more information about their patients and… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic need to think bigger. Next release has to be 7.0 minimum. Skip the sixes completely. | Anthropic need to think bigger. Next release has to be 7.0 minimum. Skip the sixes completely. | ↗Not briefed |
bluesky bluesky-ai | 1mo | Very nice explainer, both on the technical aspects of how AI models use tokens and the economics. Read this and you'll learn a lot about wh… | Very nice explainer, both on the technical aspects of how AI models use tokens and the economics. Read this and you'll learn a lot abo… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Det har liksom vært et narrativ om et Anthropic i det minste "nesten" går med overskudd, og så viste det seg at de tapte $42 MILLIARDER WIT… | Det har liksom vært et narrativ om et Anthropic i det minste "nesten" går med overskudd, og så viste det seg at de tapte $42 MILLIARDE… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "I meddelelsen hævder de også, at Claude ikke bliver trænet på samtaler med brugere." Hvorfor helvede har de så den her option i deres ind… | "I meddelelsen hævder de også, at Claude ikke bliver trænet på samtaler med brugere." Hvorfor helvede har de så den her option i dere… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I remember one group in vaccine epi. Every single paper was close to 60% (or more) self-citation. (A lot of salami slicing; they would chan… | I remember one group in vaccine epi. Every single paper was close to 60% (or more) self-citation. (A lot of salami slicing; they would… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Do ResNets Route? Sparse Interaction Experts in Residual Networks | Residual networks execute every block for every input, yet their functional contributions need not be input independent. We formulate… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Tangent Schrödinger Bridge Matching: Learning Stochastic Transport with Mechanistic Sensitivities | Predicting how stochastic systems respond to changes in viscosity, reaction rates, or external forces requires costly simulations, mot… | ↗Not briefed |
bluesky bluesky-ai | 1mo | reminder that the issue of consent to turnitin's dodgy data practices long predates the llm training era. | reminder that the issue of consent to turnitin's dodgy data practices long predates the llm training era. | ↗Not briefed |
bluesky bluesky-ai | 1mo | In Denmark a scientist plotted data and research findings into Claude, only to see Anthropic publish the findings as something the LLM had … | In Denmark a scientist plotted data and research findings into Claude, only to see Anthropic publish the findings as something the LLM… | ↗Not briefed |
arxiv arxiv-ai | 1mo | LUMOS: Tracing Parametric Knowledge from Training Data to Behavioral Outputs in LLMs | Current analyses of LLMs' parametric knowledge are largely output-centric, drawing conclusions about what a model knows without verify… | ↗Not briefed |
bluesky bluesky-ai | 1mo | So, basically, admitting that they've uploaded the book into a data centre for a LLM to train on, then query because you couldn't be arsed … | So, basically, admitting that they've uploaded the book into a data centre for a LLM to train on, then query because you couldn't be a… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Nejnovější model Google Gemini 4 Argon vrací firmu na mapu nejlepší AI modelů. Jenže je jen pro vyvolené. Haloo, kluci z Google, kde je 4 P… | Nejnovější model Google Gemini 4 Argon vrací firmu na mapu nejlepší AI modelů. Jenže je jen pro vyvolené. Haloo, kluci z Google, kde j… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Oh god. I don’t know how good the ALT text is on that, or if it’s clear. But the Y axis on that graph, which looks like it measures ever i… | Oh god. I don’t know how good the ALT text is on that, or if it’s clear. But the Y axis on that graph, which looks like it measures e… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Want to know what an advanced AI model looks like? It’s mostly matrices of numbers. And we don’t really know how it works. But we do know … | Want to know what an advanced AI model looks like? It’s mostly matrices of numbers. And we don’t really know how it works. But we do… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Interpreting at Write Time: A Policy Ablation for Multi-Goal Agent Memory | A long-running assistant cannot keep everything it has seen, so it summarises. Summarising is not neutral: what is kept is chosen agai… | ↗Not briefed |
bluesky bluesky-ai | 1mo | also you can tell I write my own posts because no LLM could fake that "waiting for the coffee to brew" word order | also you can tell I write my own posts because no LLM could fake that "waiting for the coffee to brew" word order | ↗Not briefed |
rss latent-space | 1mo | [AINews] Pi 1.0, Pi Durable, and AIE NYC | the minimalist harness goes stable... and TypeScript! | ↗Not briefed |
bluesky bluesky-ai | 1mo | Suspected Chinese hackers impersonated an Anthropic employee and ex-US officials in an espionage campaign aimed at gaining insights into Am… | Suspected Chinese hackers impersonated an Anthropic employee and ex-US officials in an espionage campaign aimed at gaining insights in… | ↗Not briefed |
arxiv arxiv-ai | 1mo | When Can We Trust the Matching Principle? Robust Deployment Geometry Under Finite-Sample and Model Uncertainty🔧 Gamma | Match only geometry you can identify; otherwise spread the penalty. We quantify that decision by the trust ratio tau = epsilon / gamma… | ↗Not briefed |
rss marktechpost | 1mo | AWS Strands Labs Releases Strands Decider 2B: An Open Source Decision Model That Picks Options in About 115 ms | AWS's Strands Agents team released Strands Decider 2B, an Apache-2.0 decision model built on Qwen3.5-2B-Base. It returns choices, yes/… | ↗Not briefed |