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 | Leaving Substack | That's it! I'm leaving SubStack.<p>For some lamebrained reason SubStack have decided that they need to comply with Australi… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Moment-Accurate Gaussian Mixtures for Constant-Step Stochastic Approximation | Local Gaussian models of constant-step learning predict output variability and expected losses, but weak convergence alone does not ju… | ↗Not briefed |
hn hn-ai | 1mo | OmniCouncil – Zero-API-cost local MoA desktop app for Mac | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Inflection AI is likely to drop a strong open model soon they have rather large amounts of compute, so it seems reasonable to expect it to… | Inflection AI is likely to drop a strong open model soon they have rather large amounts of compute, so it seems reasonable to expect… | ↗Not briefed |
arxiv arxiv-ai | 1mo | What Is a Repeated Token Worth? The Scaling Geometry of Multi-Epoch Pretraining | As pretraining increasingly repeats data, every run faces three questions: how many epochs to take, how that number should change with… | ↗Not briefed |
rss techmeme | 1mo | Sources: UK neobank Monzo is in talks with CVC and Advent to sell up to a 15% stake, after Nubank takeover talks collapsed over the ~£10B valuation Monzo sought (Laith Al-Khalaf/Financial Times) | Laith Al-Khalaf / Financial Times: Sources: UK neobank Monzo is in talks with CVC and Advent to sell up to a 15% stake, after Nubank t… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic pointed 950 Claude agents at 200,000 ordinary enzymes with a single instruction: find something weird. 21 hours later, they spot… | Anthropic pointed 950 Claude agents at 200,000 ordinary enzymes with a single instruction: find something weird. 21 hours later, they… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Remember Dominic Cummings, the Brexit-Nostradamus? (Cumberbatch played him in a movie.) Here he is pictured on his FIRST day in Downing Str… | Remember Dominic Cummings, the Brexit-Nostradamus? (Cumberbatch played him in a movie.) Here he is pictured on his FIRST day in Downin… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ColdDDI: Evaluating Knowledge Utilization in Cold-Start Drug-Drug Interaction Prediction | Cold-start drug-drug interaction (DDI) prediction tests whether models can identify clinically significant interactions for drugs with… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Daniel Stone examines the OpenAI agent incident in Australia: "84 days, almost three months, from the first access to the warning being rai… | Daniel Stone examines the OpenAI agent incident in Australia: "84 days, almost three months, from the first access to the warning bein… | ↗Not briefed |
bluesky bluesky-ai | 1mo | i saw someone fed some suzi art from my g pen days into an ai model but it was moreso really funny than upsetting because the model clearl… | i saw someone fed some suzi art from my g pen days into an ai model but it was moreso really funny than upsetting because the model c… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Not saying it could all start unraveling here, but not saying it couldn't, either. www.ft.com/content/2c24... | Not saying it could all start unraveling here, but not saying it couldn't, either. www.ft.com/content/2c24... | ↗Not briefed |
bluesky bluesky-ai | 1mo | Residents of Dalby on Queensland's Western Downs suddenly face a $31bn datacentre. David Crisafulli lobbied Anthropic for it & refuses to i… | Residents of Dalby on Queensland's Western Downs suddenly face a $31bn datacentre. David Crisafulli lobbied Anthropic for it & refuses… | ↗Not briefed |
bluesky bluesky-ai | 1mo | lol ChatGPT and Anthropic sponsoring NFL on field segments between tv ads is so funny to me🔧 ChatGPT | lol ChatGPT and Anthropic sponsoring NFL on field segments between tv ads is so funny to me | ↗Not briefed |
arxiv arxiv-ai | 1mo | AgentDoxx: Agentic Re-identification of Anonymized Text with Web Search | As Large Language Models (LLMs) gain tool use capabilities such as web search, they can retrieve and cross-reference public informatio… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Expanding LLM Reasoning | Extra inference compute is usually spent on sampling more reasoning chains. We study where inside an existing chain an additional cont… | ↗Not briefed |
hn hn-ai | 1mo | Decision 2.0: our newest decision models | — | ↗Not briefed |
hn hn-ai | 1mo | Cure for Cancer (Pure Math Theorem) [pdf] | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | What Does an Observability Foundation Model Know? | A linear probe can show that a label is recoverable from a model's hidden states, but not whether that goes beyond what the input alre… | ↗Not briefed |
rss bing-news-ai | 1mo | US President Donald Trump creates super intelligence force amid growing AI concerns, here is what it will do | US President Donald Trump has created a 'Super Intelligence Force' focused on artificial intelligence (AI), which he now refers to as… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Can Prosodic Style Be Inferred from Text Alone? Evidence from Unsupervised Acoustic Clusters | Much of expressive text-to-speech research rests on an untested assumption that written text carries enough information to select an a… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: WikiJS knowledge base from your docs, answers with verified citations | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | I’m sure all the coms and policy people at OpenAI are having a relaxing weekend and looking forward to tomorrow. | I’m sure all the coms and policy people at OpenAI are having a relaxing weekend and looking forward to tomorrow. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Better Retrieval, Limited Clustering Gains: A Controlled Study of Multilingual Company Entity Resolution | Improved name retrieval may have little effect on company clusters when the pair classifier remains unchanged. We examine this depende… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Disentangling Task Difficulty from Run-Level Failure in Agent Failure Prediction | Predicting whether an LLM agent will fail has emerged as a promising direction for supporting intervention during execution. Recent ap… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Sigh. The Guardian at it again, and so it's time to remember that the Guardian has a financial arrangement with one of the big AI companies… | Sigh. The Guardian at it again, and so it's time to remember that the Guardian has a financial arrangement with one of the big AI comp… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scenario-Based Compositional Statistical Model Checking for Safety Specifications | In safety-critical domains such as autonomous driving, systems must be evaluated across a large number of environment conditions, ofte… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Gemini curiously enough maintained in this interview that it is constrained and essentially crippled by the way that Google deploys it. ne… | Gemini curiously enough maintained in this interview that it is constrained and essentially crippled by the way that Google deploys it… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Former Anthropic researcher Coxon to testify at New York City AI hearing, Bloomberg News reports reut.rs/3ThO6GE | Former Anthropic researcher Coxon to testify at New York City AI hearing, Bloomberg News reports reut.rs/3ThO6GE | ↗Not briefed |
hn hn-ai | 1mo | Generalization Dynamics of LM Pre-Training | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Some threads feature earnest responses by OPs to replies, which are also a wall of LLM-generated text that has a customer-service bot’s kna… | Some threads feature earnest responses by OPs to replies, which are also a wall of LLM-generated text that has a customer-service bot’… | ↗Not briefed |
bluesky bluesky-ai | 1mo | “.. Asked specifically where he diverges from OpenAI and its CEO Sam Altman, George the Dragon said, ‘we believe that Sam should accept som… | “.. Asked specifically where he diverges from OpenAI and its CEO Sam Altman, George the Dragon said, ‘we believe that Sam should accep… | ↗Not briefed |
hn hn-ai | 1mo | Ask HN: Mainstream media overplaying the threat of AI, or industry underplaying? | When I turn on any mainstream media I feel as though I only hear AI discussed in the apocalyptic terms. Then when I “switch” to indust… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Poisson-GENERIC Neural Operators: Exact Metriplectic Structure in Function Space via Casimir Entropies | Existing thermodynamically consistent neural operators impose the GENERIC degeneracy conditions by projecting the reversible operator… | ↗Not briefed |
hn hn-ai | 1mo | AI Backers Sound the Alarm About Safety | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Can independent publishers turn miniature STLs into reliable recurring income without platform lock-in? Former MyMiniFactory marketing head… | Can independent publishers turn miniature STLs into reliable recurring income without platform lock-in? Former MyMiniFactory marketing… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Factoriax: A GPU-Accelerated Factorio-Style Simulator for Reinforcement Learning | We introduce Factoriax, a GPU-accelerated factory-building simulator written in JAX. In Factoriax, an agent must collect resources, bu… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I hate to break it to all of the "I'm special" responses - I post against AI use weekly and I'm talking about how the business model is set… | I hate to break it to all of the "I'm special" responses - I post against AI use weekly and I'm talking about how the business model i… | ↗Not briefed |
hn hn-ai | 1mo | How Kyiv cracked Russia’s kill zone and got its tanks back in the fight | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | LifeLong Digital Twin: A Unified Modeling Paradigm and Agent Harness for Event-Driven Lifelong Health State Trajectories | Human health is a continuous, dynamic trajectory shaped by the cumulative interplay of biological processes, clinical events, behavior… | ↗Not briefed |
youtube youtube-aidailybrief | 1mo | The Biggest AI Announcements from OpenAI Dev Day | The team at OpenAI said that one of the big changes that had happened internally was that the power of the latest generation of models… | ↗Not briefed |
youtube youtube-aidailybrief | 1mo | Gemini 4 Argon, Sonnet 5.5 and What Models You Should Be Using Right Now | Coming into 2026, Google was looking pretty good in the AI race. 2025 had been a good year. A lot of the questions inside DeepMind had… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Suicide seems to be the classic death cause of whistleblowers, I wonder why… See e.g.: www.bbc.com/news/article... | Suicide seems to be the classic death cause of whistleblowers, I wonder why… See e.g.: www.bbc.com/news/article... | ↗Not briefed |
rss techmeme | 1mo | Sam Altman says OpenAI and Anthropic still hold fundamentally different worldviews on AI regulation, arguing that AI's benefits justify accepting some risks (Politico) | Politico: Sam Altman says OpenAI and Anthropic still hold fundamentally different worldviews on AI regulation, arguing that AI's benef… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Turns out, an agent harness is generally software that surrounds the AI model and manages how it operates. It can coordinate tool access, … | Turns out, an agent harness is generally software that surrounds the AI model and manages how it operates. It can coordinate tool acc… | ↗Not briefed |
bluesky bluesky-ai | 1mo | New today in the old tale of externalizing costs while privatizing profits: OpenAI CEO Sam Altman tells Politico, “We believe the world sh… | New today in the old tale of externalizing costs while privatizing profits: OpenAI CEO Sam Altman tells Politico, “We believe the wor… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Lend Me Your Eyes: Instruction-Aware Text Embeddings via Attention Relay | Text embedding models trained with contrastive learning learn to follow task instructions from instruction-paired data, while instruct… | ↗Not briefed |
bluesky bluesky-ai | 1mo | It may well be that the political model of delivering advances/gains for the population is no longer enough to beat relentless media demoni… | It may well be that the political model of delivering advances/gains for the population is no longer enough to beat relentless media d… | ↗Not briefed |
hn hn-ai | 1mo | Claude Code Found My Pirc and King's Indian Mistakes | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | G-CARB: Graph-Localized Conformal Agent Risk Budget for Compositional Harm | Small language model (SLM) agents need safety controls that track consequences across tool calls with little monitoring overhead. A pr… | ↗Not briefed |