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 | 12d | Distilling Instructions into Weights: Closed-Form Edits Fix a Robot Policy's Misreading | Generalist robot policies often fail at one step of a task they can otherwise perform. We ask whether such a failure can be repaired b… | ↗Not briefed |
bluesky bluesky-ai | 12d | OpenAI Batch API vs Anthropic Message Batches: 50% Discount FinOps Calculator How to split asynchronous workloads (evals, document extract… | OpenAI Batch API vs Anthropic Message Batches: 50% Discount FinOps Calculator How to split asynchronous workloads (evals, document ex… | ↗Not briefed |
rss techmeme | 12d | When talking about AI safety, President Trump says "I don't mind guardrails. But I don't wanna put us in a deep second position", and "China has no guardrails" (CBS News) | CBS News: When talking about AI safety, President Trump says “I don't mind guardrails. But I don't wanna put us in a deep second posit… | ↗Not briefed |
arxiv arxiv-ai | 12d | When Answer Decodability Is Not a Decision: Separating Final-Answer and Imminent-Emission Signals in Chain-of-Thought | Answer probes applied to chain-of-thought hidden states are often read as evidence that a language model has already decided its final… | ↗Not briefed |
arxiv arxiv-ai | 12d | A Large-Scale Dataset for Cross-Lingual Multi-Jurisdiction Patent Drafting from Patent Cooperation Treaty Applications | Securing patent protection across countries requires drafting claims separately for each national patent office, in the relevant langu… | ↗Not briefed |
bluesky bluesky-ai | 12d | i for one think it's good that someone outside anthropic thinks it's important to keep ai world powers at pairity, | i for one think it's good that someone outside anthropic thinks it's important to keep ai world powers at pairity, | ↗Not briefed |
bluesky bluesky-ai | 12d | Anthropicの非公開AIモデルを使って18年間未解決だった量子コンピューターの難問が証明される https://gigazine.net/news/20261109-anthropic-psi-aaronson-ambainis/ | Anthropicの非公開AIモデルを使って18年間未解決だった量子コンピューターの難問が証明される https://gigazine.net/news/20261109-anthropic-psi-aaronson-ambainis/ | ↗Not briefed |
rss techmeme | 12d | Sources and filings detail how Jensen Huang has brought his children and other family members into Nvidia, making it unusually family-run for Silicon Valley (Wall Street Journal) | Wall Street Journal: Sources and filings detail how Jensen Huang has brought his children and other family members into Nvidia, making… | ↗Not briefed |
arxiv arxiv-ai | 12d | How Many Candidates Does a One-Step World Model Need? A Random-Codebook Account of Nearest-of-m Training on Frozen DINO Tokens | One-step stochastic world models avoid the mode averaging of deterministic feature predictors without the tens of network evaluations… | ↗Not briefed |
bluesky bluesky-ai | 12d | Listen, I know FFXI has a barrier to entry, but genuinely, if you need a FFXI-LLM assistant to give you information which would be easier t… | Listen, I know FFXI has a barrier to entry, but genuinely, if you need a FFXI-LLM assistant to give you information which would be eas… | ↗Not briefed |
bluesky bluesky-ai | 13d | "this product is unreliable an unknown but non-negligable percentage of the time" is functionally equivalent to "this product is worthless … | "this product is unreliable an unknown but non-negligable percentage of the time" is functionally equivalent to "this product is worth… | ↗Not briefed |
rss bing-news-ai | 13d | Not AMD. Not Intel. This Is the Artificial Intelligence (AI) Chip Stock Nvidia Should Fear Most. | AMD and Intel both offer alternative chip architectures to Nvidia's GPUs, but there's a bigger threat in the background. | ↗Not briefed |
rss bing-news-ai | 13d | Not AMD. Not Intel. This Is the Artificial Intelligence (AI) Chip Stock Nvidia Should Fear Most. | AMD and Intel both offer alternative chip architectures to Nvidia's GPUs, but there's a bigger threat in the background. | ↗Not briefed |
rss bing-news-ai | 13d | Not AMD. Not Intel. This Is the Artificial Intelligence (AI) Chip Stock Nvidia Should Fear Most. | AMD and Intel both offer alternative chip architectures to Nvidia's GPUs, but there's a bigger threat in the background. | ↗Not briefed |
arxiv arxiv-ai | 13d | When Does Physics Beat Data? Data-Efficient Traffic State Estimation with Amortized PINNs | Traffic state estimation (TSE) reconstructs traffic density at unmonitored locations from a sparse set of sensors. Data-driven models… | ↗Not briefed |
arxiv arxiv-ai | 13d | Tropical Gaussian Anticoncentration: Optimal Instance-Dependent Bounds for Online Learning in Extensive-Form Games | We study the minimax regret of full-information online decision making in extensive-form games. For prediction with expert advice, the… | ↗Not briefed |
bluesky bluesky-ai | 13d | futurism.com/artificial-i... In all, she doesn’t believe that the election drastically changes anything for the company. To her, it seems,… | futurism.com/artificial-i... In all, she doesn’t believe that the election drastically changes anything for the company. To her, it s… | ↗Not briefed |
rss bing-news-ai | 13d | Artificial Intelligence in IoT Market to Reach US$ 58.02 Billion by 2034 at 19.56% CAGR, Driven by Edge Computing and Real-Time Data Intelligence| Exclusive Report by The ... | The global Artificial Intelligence in IoT Market is projected to grow from US$ 11.62 billion in 2025 to US$ 58.02 billion by 2034, reg… | ↗Not briefed |
rss bing-news-ai | 13d | Artificial Intelligence in IoT Market to Reach US$ 58.02 Billion by 2034 at 19.56% CAGR, Driven by Edge Computing and Real-Time Data Intelligence| Exclusive Report by The ... | The global Artificial Intelligence in IoT Market is projected to grow from US$ 11.62 billion in 2025 to US$ 58.02 billion by 2034, reg… | ↗Not briefed |
hn hn-ai | 13d | Show HN: clap.cpp – Audio embedding inference in C++ (GGML) | — | ↗Not briefed |
bluesky bluesky-ai | 13d | 腕のある有志の技術者達が何年もかけて古いゲームの新しいエンジンを開発しました → かわいい、すごい。 LLMパワーで最新ゲームを一週間でリバースエンジニアリングしました!あらゆるハードでプレイできるよ!! → 可愛くない。じゃあく😡 | 腕のある有志の技術者達が何年もかけて古いゲームの新しいエンジンを開発しました → かわいい、すごい。 LLMパワーで最新ゲームを一週間でリバースエンジニアリングしました!あらゆるハードでプレイできるよ!! → 可愛くない。じゃあく😡 | ↗Not briefed |
arxiv arxiv-ai | 13d | Fade to... Red? Scheduling Coloured Noise Exploration Supports Efficient Robot Learning with Reduced Movement Jerk | Exploration in deep reinforcement learning (RL) for continuous control is commonly driven by temporally uncorrelated Gaussian (white)… | ↗Not briefed |
bluesky bluesky-ai | 13d | The finest minds have not yet found a “cure” for cancer; some LLM isn’t going to succeed where they failed. If it does succeed, it will be … | The finest minds have not yet found a “cure” for cancer; some LLM isn’t going to succeed where they failed. If it does succeed, it wil… | ↗Not briefed |
arxiv arxiv-ai | 13d | Competing Influence in LLM Populations: Budget Timing and Baseline Drift | When competing campaigns target a population of large language model (LLM) agents, observed opinion changes may reflect both campaign… | ↗Not briefed |
arxiv arxiv-ai | 13d | DeFSM: Robust and Efficient Decentralization of Federated Submodular Maximization🔧 Consensus | Submodular maximization seeks a small subset best representing a large dataset. In its federated setting, the server is a single point… | ↗Not briefed |
hn hn-ai | 13d | Show HN: Flide: Make Jev-like decision models up to 9.5x faster🔧 vLLM | Decision models need their own inference engine.<p>So we developed Flide, a high-performance and exact inference engine for Jev-like d… | ↗Not briefed |
arxiv arxiv-ai | 13d | Can Hindsight Critics Guide Delayed Actors? | Sensory delays create temporal misalignments between the state of the environment and the observations available to an agent, turning… | ↗Not briefed |
bluesky bluesky-ai | 13d | I actually think he’s referring to Dario Amodei though I don’t it happened in the way Trump claims. Remember when Anthropic objected to DoD… | I actually think he’s referring to Dario Amodei though I don’t it happened in the way Trump claims. Remember when Anthropic objected t… | ↗Not briefed |
bluesky bluesky-ai | 13d | I've recently started hiding the bulk of my AI-generated GitHub PR descriptions behind a collapsed <details><summary> HTML tag (thankfully … | I've recently started hiding the bulk of my AI-generated GitHub PR descriptions behind a collapsed <details><summary> HTML tag (thankf… | ↗Not briefed |
arxiv arxiv-ai | 13d | Are UFOs Real? Introducing Unified Factored Organizers | Implicit in much research in deep learning is the assumption that if a network achieves low loss, then it must have learned a good rep… | ↗Not briefed |
arxiv arxiv-ai | 13d | Scaling Adaptive Prompt Injection Attacks with Agent Swarms | Prompt injections have emerged as one of the top security risks faced by AI agents today. Early work in prompt injection attacks looke… | ↗Not briefed |
rss simon-willison | 13d | The Web Is More Than Enough, by Paul Ford | — | ↗Not briefed |
rss simon-willison | 13d | Hide your slop PR descriptions behind <details><summary> | — | ↗Not briefed |
rss techmeme | 13d | Jensen Huang's scandal poses a rare crisis in US corporate history, given the national security stakes, his centrality to Nvidia, and its role in the AI race (Andrew Edgecliffe-Johnson/Semafor) | Andrew Edgecliffe-Johnson / Semafor: Jensen Huang's scandal poses a rare crisis in US corporate history, given the national security s… | ↗Not briefed |
arxiv arxiv-ai | 13d | DeltaForge: A Post-hoc Mixture-of-Experts VLA for Multi-Domain Autonomous Driving | Fleets of driving vision-language-action (VLA) models are developed one domain at a time: each operator finetunes a shared base model… | ↗Not briefed |
arxiv arxiv-ai | 13d | Decoupled Asynchronous System for Humanoids (DASH): Latency-Agnostic Scaling of Vision-Language-Action Models for Humanoid Manipulation | Vision-language-action (VLA) models pair a vision-language backbone with an action head, but in closed-loop control the backbone sits… | ↗Not briefed |
bluesky bluesky-ai | 13d | Claude is the only LLM verified to have actually been use in war crimes tbf | Claude is the only LLM verified to have actually been use in war crimes tbf | ↗Not briefed |
bluesky bluesky-ai | 13d | Just about reached the point where it's not worth even opening the Steam Store. So many titles w. no AI disclosure with obviously vibe-code… | Just about reached the point where it's not worth even opening the Steam Store. So many titles w. no AI disclosure with obviously vibe… | ↗Not briefed |
arxiv arxiv-ai | 13d | Identifiable but Hard to Estimate: Rotation in Equally Spaced Overcomplete ICA | How many observations are needed to estimate a common rotation in two-dimensional overcomplete independent component analysis? We stud… | ↗Not briefed |
hn hn-ai | 13d | Show HN: A diffusion model that runs in pure CSS, no JavaScript | — | ↗Not briefed |
bluesky bluesky-ai | 13d | DeepSeek R1 (671B Reasoning) Cloud Pricing & Cheapest Providers (2026): What does DeepSeek R1 (671B Reasoning) actually cost in production… | DeepSeek R1 (671B Reasoning) Cloud Pricing & Cheapest Providers (2026): What does DeepSeek R1 (671B Reasoning) actually cost in produ… | ↗Not briefed |
bluesky bluesky-ai | 13d | im so fucking stoked for the big LLM chatbots to go down cuz the companies go under and the complete and total collapse of slop being uploa… | im so fucking stoked for the big LLM chatbots to go down cuz the companies go under and the complete and total collapse of slop being… | ↗Not briefed |
hn hn-ai | 13d | Best-Effort Extent-Same, a btrfs dedupe agent | — | ↗Not briefed |
arxiv arxiv-ai | 13d | Causal Signal Credit Assignment for Verilog Code Generation via Counterfactual Hardware Verification🔧 GPT-5 | Large language models have improved Verilog generation, yet functional correctness remains difficult to optimize because simulation co… | ↗Not briefed |
arxiv arxiv-ai | 13d | Severity-Controlled Prediction Sets for Medication Recommendation | Medication recommendation is a multi-label prediction task. However, existing prediction-set evaluation and calibration methods rely m… | ↗Not briefed |
arxiv arxiv-ai | 13d | MOTIF: LLM-Defined Material Taxonomies for OOD Property Prediction via Invariant Feature Selection | Out-of-distribution (OOD) generalization remains a central challenge in materials property prediction, where models must extrapolate t… | ↗Not briefed |
rss techmeme | 13d | An interview with Cognition CISO Alex Stamos on how AI is upending cybersecurity, AI-powered red teaming, vulnerabilities in legacy codebases, and more (Nilay Patel/The Verge) | Nilay Patel / The Verge: An interview with Cognition CISO Alex Stamos on how AI is upending cybersecurity, AI-powered red teaming, vul… | ↗Not briefed |
bluesky bluesky-ai | 13d | Look at these coward's using "superintelligence" to mean "LLM". What a bunch of spinelessness. | Look at these coward's using "superintelligence" to mean "LLM". What a bunch of spinelessness. | ↗Not briefed |
bluesky bluesky-ai | 13d | Hear me out: a remake of WarGames where a kid’s LLM test suite accidentally breaks into NORAD and starts triggering a missile launch agains… | Hear me out: a remake of WarGames where a kid’s LLM test suite accidentally breaks into NORAD and starts triggering a missile launch a… | ↗Not briefed |
rss bing-news-ai | 13d | Not AMD. Not Intel. This Is the Artificial Intelligence (AI) Chip Stock Nvidia Should Fear Most. | AMD and Intel both offer alternative chip architectures to Nvidia's GPUs, but there's a bigger threat in the background. | ↗Not briefed |