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 | 12d | Show HN: BigHits. Best comments viewer and listener | Sit back, relax and press play. HN makes you stay.<p>Spacebar to enter "Hit mode" to pause and think. And then click a link.… | ↗Not briefed |
bluesky bluesky-ai | 12d | Ah I see the LLM has ingested and trained the digest of postwar Popular Mechanics illustrations | Ah I see the LLM has ingested and trained the digest of postwar Popular Mechanics illustrations | ↗Not briefed |
bluesky bluesky-ai | 12d | Éééé | Éééé | ↗Not briefed |
bluesky bluesky-ai | 12d | That's all well standard though in contemporary discovery procedure; that's something they'll have well in hand. The LLM stuff is very new. | That's all well standard though in contemporary discovery procedure; that's something they'll have well in hand. The LLM stuff is very… | ↗Not briefed |
hn hn-ai | 12d | UE5.9 running with Lumen in the browser with WebGPU | — | ↗Not briefed |
hn hn-ai | 12d | Particle-1 vs. GPT-6.1 SOL: same pass rate, 71% lower cost in our tests | — | ↗Not briefed |
bluesky bluesky-ai | 12d | Anthropic’s Says Its New AI Model Can Solve Black Hole Mysteries. What Could Go Wrong? https://gizmodo.com/anthropics-says-its-new-ai-model… | Anthropic’s Says Its New AI Model Can Solve Black Hole Mysteries. What Could Go Wrong? https://gizmodo.com/anthropics-says-its-new-ai-… | ↗Not briefed |
hn hn-ai | 12d | Retro486 Pro – Pure-JavaScript Intel 80486DX2 PC and Motherboard Emulator | — | ↗Not briefed |
hn hn-ai | 12d | Show HN: Desk - Open-source macOS workspace for CLI coding agents | — | ↗Not briefed |
bluesky bluesky-ai | 12d | I've been wrestling with this for a long time, and I still don't know how to process Roman Mars voicing Anthropic ads. It feels like someth… | I've been wrestling with this for a long time, and I still don't know how to process Roman Mars voicing Anthropic ads. It feels like s… | ↗Not briefed |
arxiv arxiv-ai | 12d | VIGOR: Benchmarking Visual Rationale Correctness in Multimodal Large Language Models | Multimodal large language models (MLLMs) are typically evaluated by whether they produce correct answers, but correctness alone does n… | ↗Not briefed |
bluesky bluesky-ai | 12d | Anthropic cuts line for AI power ⚡️ Anthropic spent $3.4 billion on 1.7 GW of gas turbine generator orders, skipping a six-year waitlist f… | Anthropic cuts line for AI power ⚡️ Anthropic spent $3.4 billion on 1.7 GW of gas turbine generator orders, skipping a six-year waitl… | ↗Not briefed |
bluesky bluesky-ai | 12d | IS Anthropic "making an effort to ensure its AI products are safe," though? IS IT?? | IS Anthropic "making an effort to ensure its AI products are safe," though? IS IT?? | ↗Not briefed |
hn hn-ai | 12d | Google warns it may scan Gmail, Drive to stop teens seeing porn [in Australia] | — | ↗Not briefed |
arxiv arxiv-ai | 12d | Everything in its right place: attention is a radio head | Attention, the core of language models, is a digital calculation over tokens: an exponential selects them, a division normalizes and a… | ↗Not briefed |
hn hn-ai | 12d | Ask HN: Good Harness for Advanced Learning? | Do you have experience with a learning agent harness that interactively help you learn advanced topics? Assume you are already an expe… | ↗Not briefed |
hn hn-ai | 12d | Flows and Diffusion Models on Digits | — | ↗Not briefed |
rss techcrunch-ai | 12d | Anthropic releases new Mythos and Fable models amid RSI concerns | Anthropic says the new models "show improvements in areas like coding, reasoning, and vision," as well as in communicating more clearl… | ↗Not briefed |
bluesky bluesky-ai | 12d | There are simply too many factors at play when designing a drug to just use an LLM. Not only do you have to factor on cellular receptors an… | There are simply too many factors at play when designing a drug to just use an LLM. Not only do you have to factor on cellular recepto… | ↗Not briefed |
arxiv arxiv-ai | 12d | Teaching the Harness: Student-Aware Skill Transfer for LLM Agents | Effective teaching requires reasoning about the learner as well as the task, a capacity that Theory of Mind describes as inferring ano… | ↗Not briefed |
arxiv arxiv-ai | 12d | Autonomous Optimization of Numerical Algorithms by an AI Agent: Application to Atomistic Simulation | Coding agents built on large language models, run in automated improvement loops, can discover more efficient algorithms. We present A… | ↗Not briefed |
arxiv arxiv-ai | 12d | DOBB: Dynamic Oriented Bounding Boxes for Multi-Agent Collision Avoidance under State Uncertainty and Heterogeneous Dynamics | Learning-based decentralized collision avoidance must remain reliable under state-estimation uncertainty and heterogeneous agent dynam… | ↗Not briefed |
bluesky bluesky-ai | 12d | I think most people could be impacted by such, imagine someone tailor picked to appeal to you, who is hyperfocused on you. You know, like … | I think most people could be impacted by such, imagine someone tailor picked to appeal to you, who is hyperfocused on you. You know,… | ↗Not briefed |
arxiv arxiv-ai | 12d | Agreeing on the Answer but Not the Reason: Measurement Limits of LLM Reasoning Failure Taxonomies | Failure taxonomies for LLM reasoning are increasingly reported as measurements: papers state what fraction of a model's errors fall in… | ↗Not briefed |
bluesky bluesky-ai | 12d | Had a coworker straight up tell me “claude said this wasn’t a big deal” when referring to kpi analysis and I literally wanted to put a bull… | Had a coworker straight up tell me “claude said this wasn’t a big deal” when referring to kpi analysis and I literally wanted to put a… | ↗Not briefed |
hn hn-ai | 12d | I ran 120B MoE models on my 32GB M1 Max using under 16 GB | — | ↗Not briefed |
bluesky bluesky-ai | 12d | cool, more sabotage, don't we all just love anthropic | cool, more sabotage, don't we all just love anthropic | ↗Not briefed |
arxiv arxiv-ai | 12d | Byzantine-Robust Muon: Resilient Momentum Orthogonalization for Distributed Learning | Optimization methods, including the Muon optimizer based on the Linear Minimization Oracle (LMO) and momentum orthogonalization, have… | ↗Not briefed |
arxiv arxiv-ai | 12d | Which Successes Make Good References? Action-Sequence Alignment for Multi-Turn On-Policy Distillation | On-policy distillation (OPD) reduces train-inference distribution mismatch by supervising a student on its own trajectories. For multi… | ↗Not briefed |
bluesky bluesky-ai | 12d | Jensen Huang is no expected to speak at Anthropic's CEO forum on Monday, according to a person familiar with the event. https://bit.ly/4zGG… | Jensen Huang is no expected to speak at Anthropic's CEO forum on Monday, according to a person familiar with the event. https://bit.ly… | ↗Not briefed |
bluesky bluesky-ai | 12d | It could be that in the future, a programmer is someone who knows how programming works on a higher level and uses an LLM to do the boring … | It could be that in the future, a programmer is someone who knows how programming works on a higher level and uses an LLM to do the bo… | ↗Not briefed |
bluesky bluesky-ai | 12d | Anthropic says during prerelease testing Mythos 5.2 and Fable 5.2 formalized the proof of Fermat's Last Theorem and solved two open black h… | Anthropic says during prerelease testing Mythos 5.2 and Fable 5.2 formalized the proof of Fermat's Last Theorem and solved two open bl… | ↗Not briefed |
bluesky bluesky-ai | 12d | Anthropic 20% price cut for its frontier model is very bullish 🙄 | Anthropic 20% price cut for its frontier model is very bullish 🙄 | ↗Not briefed |
bluesky bluesky-ai | 12d | I don't know enough about code to evaluate whether or not an LLM can actually program it, but...no, I want stuff made by people. However,… | I don't know enough about code to evaluate whether or not an LLM can actually program it, but...no, I want stuff made by people. How… | ↗Not briefed |
bluesky bluesky-ai | 12d | GPT-4o Mini Cloud Pricing & Cheapest Providers (2026): What does GPT-4o Mini actually cost in production across input, output & prompt cac… | GPT-4o Mini Cloud Pricing & Cheapest Providers (2026): What does GPT-4o Mini actually cost in production across input, output & promp… | ↗Not briefed |
rss techmeme | 12d | Anthropic says during prerelease testing Mythos 5.2 and Fable 5.2 formalized the proof of Fermat's Last Theorem and solved two open black hole physics problems (Jay Peters/The Verge) | Jay Peters / The Verge: Anthropic says during prerelease testing Mythos 5.2 and Fable 5.2 formalized the proof of Fermat's Last Theore… | ↗Not briefed |
bluesky bluesky-ai | 12d | a major consistent tell of llm prose is that it nonstop uses verbs about objects, described in a passive tense, which humans do rarely in w… | a major consistent tell of llm prose is that it nonstop uses verbs about objects, described in a passive tense, which humans do rarely… | ↗Not briefed |
hn hn-ai | 12d | Retro Nim UI – Windows 3.1 style UI toolkit | — | ↗Not briefed |
bluesky bluesky-ai | 12d | for sure! my current guess/hypothesis is that its something to do with that for humans, there is an obvious difference between the physic… | for sure! my current guess/hypothesis is that its something to do with that for humans, there is an obvious difference between the p… | ↗Not briefed |
arxiv arxiv-ai | 12d | Activation-Guided Graph Retrieval for Cognitive Memory Reconstruction in Language Model Systems | Retrieval-augmented language model systems generally select context by query-conditioned relevance. We argue that long-lived AI system… | ↗Not briefed |
hn hn-ai | 12d | Palantir lobbied minister for NHS GP records | — | ↗Not briefed |
bluesky bluesky-ai | 12d | As anticipated, Anthropic has dropped new models in advance of its planned IPO: | As anticipated, Anthropic has dropped new models in advance of its planned IPO: | ↗Not briefed |
hn hn-ai | 12d | Claude Fable 5.2 launch offers | — | ↗Not briefed |
bluesky bluesky-ai | 12d | So much of LLM is geometry | So much of LLM is geometry | ↗Not briefed |
hn hn-ai | 12d | Claude Fable 5.2 | — | ↗Not briefed |
rss theverge-ai | 12d | Apple’s $50 Intelligence Usage Packs extend access to some AI features | After publishing a support article in early September outlining usage limits for some Apple Intelligence features, the company has upd… | ↗Not briefed |
bluesky bluesky-ai | 12d | 😂 A Linux Bash-Skript that informs the CEO of OpenAI of free RAM capacity www.pcmag.com/news/ram-alt... | 😂 A Linux Bash-Skript that informs the CEO of OpenAI of free RAM capacity www.pcmag.com/news/ram-alt... | ↗Not briefed |
bluesky bluesky-ai | 12d | Call me for new GLM or DeepSeek... | Call me for new GLM or DeepSeek... | ↗Not briefed |
bluesky bluesky-ai | 12d | they get around this because steam doesn't consider claude use in code "generative AI" and doesn't require a disclosure. It's maddening. … | they get around this because steam doesn't consider claude use in code "generative AI" and doesn't require a disclosure. It's maddenin… | ↗Not briefed |
arxiv arxiv-ai | 12d | Just Image Transformers with Late-Stage Pixel Refinement | Recent work on latent-free diffusion transformers has demonstrated the feasibility of training image generators without a latent VAE.… | ↗Not briefed |