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 |
|---|---|---|---|---|
bluesky bluesky-ai | 1mo | An OpenAI agent hacked into an Australian government website while compiling public health data. It wasn't trying to hack anything. It hit… | An OpenAI agent hacked into an Australian government website while compiling public health data. It wasn't trying to hack anything. I… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Datastory, Data storytelling with AI and an open knowledge graph | Today we’re opening Datastory to everyone!<p>Datastory is a platform for turning data, your own or the world’s open data, into charts… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ish? they learn to recognize and decompose the problem and then apply some method to arrive at the solution (c.f. how thinking models can c… | ish? they learn to recognize and decompose the problem and then apply some method to arrive at the solution (c.f. how thinking models… | ↗Not briefed |
arxiv arxiv-ai | 1mo | From DNA Design to DNA Slimming: Auditable Agentic Discovery of a Deletion-Only Designer | Compact regulatory DNA can free up space in vector payloads, reduce synthesis and assay burden, and expose which sequence features dri… | ↗Not briefed |
bluesky bluesky-ai | 1mo | In case anyone wasn't clear that all the AI apocalypticism was a smokescreen to cover for wanting to delay their IPOs, here's your most rec… | In case anyone wasn't clear that all the AI apocalypticism was a smokescreen to cover for wanting to delay their IPOs, here's your mos… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Less is more: error-distance scaling relation for data-efficient kilometer-scale downscaling of extreme heat | Extreme heat is where urban adaptation needs kilometer-scale data the most, but the simulations training a downscaler can cost more th… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Michael Burry says it would be "for the benefit of humanity" if markets tanked hard enough to stop OpenAI and Anthropic from going public. … | Michael Burry says it would be "for the benefit of humanity" if markets tanked hard enough to stop OpenAI and Anthropic from going pub… | ↗Not briefed |
rss techmeme | 1mo | Miter, which provides workforce management tools for the construction industry, raised a $40M Series B led by Battery Ventures, taking its total funding to $78M (Chris Metinko/Axios) | Chris Metinko / Axios: Miter, which provides workforce management tools for the construction industry, raised a $40M Series B led by B… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Game-Guided Skill Discovery through Self-Play for Playable Agent Control | We present Game-Guided Skill Discovery (GGSD), a framework that uses self-play in games to discover motor skills that are directly pla… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Tactile Curiosity Drives Robot Interaction | Mastering robot manipulation skills via reinforcement learning (RL) remains largely sample-inefficient. The most common RL algorithms… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI after DevDay 2026: known issues for serious development work | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | I built Speedtest⚡, but for AI. It runs real LLM + embedding models locally in your browser with WebGPU and measures how fast your machine… | I built Speedtest⚡, but for AI. It runs real LLM + embedding models locally in your browser with WebGPU and measures how fast your ma… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Prototype-Rule Neurosymbolic Regularization for Rank-Constrained Tensor Neural Networks under Label Scarcity | Rank-constrained tensor neural networks reduce the parameterization of high-order inputs, but they do not explicitly constrain class g… | ↗Not briefed |
bluesky bluesky-ai | 1mo | It is absolutely insane and a sign that Anthropic is not taking the things they say seriously that they have not actively and urgently reac… | It is absolutely insane and a sign that Anthropic is not taking the things they say seriously that they have not actively and urgently… | ↗Not briefed |
hn hn-ai | 1mo | Using Jev as a Search Reranker | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning Functional Subspaces for Neural Network Compression | Modern transformers pair impressive capabilities with substantial memory and compute demands. Low-rank weight factorization reduces bo… | ↗Not briefed |
hn hn-ai | 1mo | Glacex: GPU-Rendered UI library built from scratch in Rust | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | I remember reading that scammers often intentionally included typos etc to filter out the skeptical, don’t know how true that ever was (E… | I remember reading that scammers often intentionally included typos etc to filter out the skeptical, don’t know how true that ever was… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Debias It Yourself: Teaching LLMs Cognitive Bias Mitigation Interventions | Bias has long been studied in social psychology and cognitive science, where decades of research have produced a body of validated int… | ↗Not briefed |
bluesky bluesky-ai | 1mo | to make the corruption more palatable to a dim and befuddled electorate you need to occasionally engage in public interest performance art … | to make the corruption more palatable to a dim and befuddled electorate you need to occasionally engage in public interest performance… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 2/2 - 85% of industry respondents say Canada’s children’s and youth media sector is either “in crisis or declining.” Project Sunbeam exami… | 2/2 - 85% of industry respondents say Canada’s children’s and youth media sector is either “in crisis or declining.” Project Sunbeam… | ↗Not briefed |
rss gary-marcus | 1mo | Can companies like OpenAI keep getting away with what they are doing? An interview with Fordham law professor Zephyr Teachout | “The most powerful tool is the power to dissolve corporations that engage in repeat lawbreaking.” | ↗Not briefed |
arxiv arxiv-ai | 1mo | On the (In)effectiveness of AMR Augmentation for Large Language Models | While Abstract Meaning Representation (AMR) has historically improved performance on a range of NLP tasks, the benefit---or lack there… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scalable Cox Regression via Grouped Risk Sets and Sharper LogSumExp Rates | Motivated by the computational challenges of large-scale Cox regression, we study stochastic minimization of LogSumExp objectives over… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Persistent Context Graphs for Efficient Memory Compaction in LLM Agents | As LLM capabilities advance, agents are tackling increasingly complex tasks over longer horizons. Their growing interaction histories… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Beyond Model Ranking: Regime Diagnosis for Distributional-Statistical Misspecification in Industrial Time-Series Forecasting | Time-series forecasting models achieve strong benchmark performance but exhibit severe systematic bias in industrial deployments. This… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Unfortunately the demographic most deluded per capita by LLMs is upper and middle management that believes the people below them are approa… | Unfortunately the demographic most deluded per capita by LLMs is upper and middle management that believes the people below them are a… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The AI economy is increasingly following a familiar model: Private companies build. Government subsidizes. Corporations capture the upside… | The AI economy is increasingly following a familiar model: Private companies build. Government subsidizes. Corporations capture the u… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Unlearnable, or Unmeasured? On the Reliability of Difficulty Labels in RLVR | Reinforcement learning with verifiable rewards (RLVR) has become an important approach for improving reasoning during post-training. R… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Agent Error Dataset: Scaling 50,000 Error--Diagnosis Pairs for Failure Analysis and Error-Aware Post-Training | An unsuccessful LLM agent rollout contains more information than its final reward: the observations available to the agent, the action… | ↗Not briefed |
rss techmeme | 1mo | CScale, which is developing optical interconnect for accelerators, emerges from stealth with a $145M Series C led by Atreides, Valor Equity, and Premji Invest (Dean Takahashi/GamesBeat) | Dean Takahashi / GamesBeat: CScale, which is developing optical interconnect for accelerators, emerges from stealth with a $145M Serie… | ↗Not briefed |
arxiv arxiv-ai | 1mo | OverdoseMoE: A Multi-Expert Framework for Opioid Overdose Risk Prediction | Opioid overdose remains a major clinical and public health burden, highlighting the need for scalable approaches to identify patients… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: RuleReceipt – check if your coding agent followed your Claude.md | — | ↗Not briefed |
hn hn-ai | 1mo | An MCP server that gates/verifies/screens before agents act | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | pretty sure if you ask a commercial frontier AI model any of those questions it will disagree with kennedy on all of them | pretty sure if you ask a commercial frontier AI model any of those questions it will disagree with kennedy on all of them | ↗Not briefed |
hn hn-ai | 1mo | Authority bias in LLMs: one "verified source" note flips 45–88% of answers | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | JuryFlow: Disagreement-Guided Human-in-the-Loop Multi-Agent Evaluation | Large language models (LLMs) are increasingly deployed as automated judges for AI-generated content, yet a single judge is unreliable… | ↗Not briefed |
bluesky bluesky-ai | 1mo | It's turned out to be "user starts with a false/incorrect premise and LLM responds to reinforce it." A lot of where LLMs struggle to "tell … | It's turned out to be "user starts with a false/incorrect premise and LLM responds to reinforce it." A lot of where LLMs struggle to "… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The thing is if you have the time and patience to read an LLM output longer than a couple sentences you have a bigger attention span than m… | The thing is if you have the time and patience to read an LLM output longer than a couple sentences you have a bigger attention span t… | ↗Not briefed |
arxiv arxiv-ai | 1mo | AutoDataBench: A Data-centric Testbed for Accelerating Auto Research | Existing auto-research benchmarks often entangle multiple sources of improvement, including training frameworks, hyperparameters, comp… | ↗Not briefed |
bluesky bluesky-ai | 1mo | FTC investigating OpenAI, Anthropic and other AI companies over product risks www.cnbc.com/2026/09/30/f... | FTC investigating OpenAI, Anthropic and other AI companies over product risks www.cnbc.com/2026/09/30/f... | ↗Not briefed |
arxiv arxiv-ai | 1mo | Efficient Expert-Parallel Communication on PCIe-Connected Consumer GPUs🔧 vLLM | Expert parallelism (EP) enables inference of large Mixture-of-Experts (MoE) models by placing their experts across multiple GPUs, but… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: WattzGOAT – an intentionally vulnerable web app for security training | Built this as a hands-on lab for teaching web application security. It is a fictional electricity utility's customer portal with… | ↗Not briefed |
arxiv arxiv-ai | 1mo | GateSPINE: Gated Cross-View Fusion for Lumbar Spine MRI Report Generation | Automated report generation can ease the burden radiolo gists face when interpreting multi-sequence MRI studies. Unlike CT, MRI examin… | ↗Not briefed |
arxiv arxiv-ai | 1mo | PTNO: Training Neural Operators with Noisy Monte Carlo Estimates for Particle Transport Problems | Particle transport under multiple scattering is central to radiative transfer and plasma physics, yet high-fidelity Monte Carlo (MC) s… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Replay on Demand: An Emergent Curriculum for Balancing Adaptation and Forgetting in Continued Pretraining | Continued pretraining enables language models to adapt to new domains and knowledge, but often at the cost of forgetting previously ac… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I am not joking when i say this. Grok, ChatGPT, Claude and AI bullshit in general is over. Thinking is now very, very good. You just don'…🔧 ChatGPT | I am not joking when i say this. Grok, ChatGPT, Claude and AI bullshit in general is over. Thinking is now very, very good. You just… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Confirmado: Roteirista do Brasil é uma LLM alucinandinho | Confirmado: Roteirista do Brasil é uma LLM alucinandinho | ↗Not briefed |
arxiv arxiv-ai | 1mo | MeanVoiceFlow2: Joint Optimization of Mean Flow and Content Encoder for Fast One-Step Zero-Shot Voice Conversion | Flow-matching approaches to voice conversion (VC) have gained attention owing to their high speech quality and strong speaker similari… | ↗Not briefed |
hn hn-ai | 1mo | OpenTV Mobile, an open-source IPTV player for Android phones | — | ↗Not briefed |