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 | Un truc qui me fascine avec les oies de lorentz vous savez, cet éthologue qui a réussi à faire prendre à des oisillons une balle de ping-… | Un truc qui me fascine avec les oies de lorentz vous savez, cet éthologue qui a réussi à faire prendre à des oisillons une balle de… | ↗Not briefed |
hn hn-ai | 1mo | My coding agent now records the demo video for its own PRs | — | ↗Not briefed |
rss techmeme | 1mo | AI agents are the ultimate Aggregators; they reveal apps as a means, not an end, and offering them is tech's biggest prize, with Meta and Microsoft well-poised (Ben Thompson/Stratechery) | Ben Thompson / Stratechery: AI agents are the ultimate Aggregators; they reveal apps as a means, not an end, and offering them is tech… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SPIDER: Multi-Layer Semantic Token Pruning and Adaptive Sub-Layer Skipping in Multimodal Large Language Models | Multimodal Large Language Models face significant efficiency challenges that stem from two distinct yet coupled sources: data redundan… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Inspector: Conversational and Lightweight Analyzer of Analog Circuit Layouts Using LLM and CNNs | The integration of artificial intelligence into computer-aided design frameworks has sparked a shift in the design of analog integrate… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Teach to Learn: Hint Annealing for Self-improving LLM Reasoning | Group Relative Policy Optimization (GRPO) improves language-model reasoning by comparing verified rewards among multiple solution roll… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Before Acting, Change the State: Prospective State Intervention for Web Agents under Deceptive Interfaces | LLM-based Web agents can autonomously complete user tasks, yet deceptive interfaces can steer them toward outcomes that conflict with… | ↗Not briefed |
arxiv arxiv-ai | 1mo | APEX-Voice: Can Voice Agents Complete Professional Workflows Through Full-Duplex Interaction | Full-duplex voice agents can now listen, speak, use tools, and act during spoken interactions, but fluent dialogue does not guarantee… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Just MLPs: Efficient Visual State Reconstruction for Multimodal Language Models | Long visual token sequences often account for a substantial fraction of the computational overhead in multimodal large language models… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Action-Space Shaping for LLM Agents: Measuring and Mitigating Tool-Schema Bias | Large Language Models (LLMs) have shown strong performance on tool-use agentic tasks when given a fixed tool schema. Yet a tool schema… | ↗Not briefed |
bluesky bluesky-ai | 1mo | perso, apprendre qu'un llm a produit ce que j'ai devant les yeux, je vois toujours ça comme une trahison, parce que j'ai passé du temps à t… | perso, apprendre qu'un llm a produit ce que j'ai devant les yeux, je vois toujours ça comme une trahison, parce que j'ai passé du temp… | ↗Not briefed |
hn hn-ai | 1mo | Parisians on Hunt for Baguettes as Bakers Get Nod to Take Vacation (2015) | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | See it, Say it, Sorted: Mechanistic Diagnosis and Parameter-Space Mitigation of Emergent Misalignment in LLMs | Safety-aligned LLMs can exhibit emergent misalignment (EM): narrow domain adaptation unexpectedly triggers catastrophic safety failure… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Skill that lets you run other Skills as Jinja templates | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | RoboFL: Federated Expert Assembly for World Action Models🔧 Consensus | Vision-language-action and world-action models are increasingly popular, yet remain bottlenecked by physical interaction data that is… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Semantic Uncertainty Quantification Needs Factual Equivalence | Semantic uncertainty quantification for large language models rests on a common template: sample several answers, measure how much the… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Safe Greenhouse Climate Control Using Lagrangian-Constrained PPO with Kolmogorov-Arnold Networks | Greenhouse climate control balances economic return with maintaining temperature, humidity and CO2 within crop-adapted growth ranges.… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Cyclostationary Phase Conditioning for Medical Time Series Diffusion | Many physiological time series, such as cardiac and brain recordings, exhibit cyclostationarity: their statistics vary periodically wi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Except that Anthropic just took his research, put a little bow on it, and THEN claimed it as their own. What’s needed is for the AI compan… | Except that Anthropic just took his research, put a little bow on it, and THEN claimed it as their own. What’s needed is for the AI c… | ↗Not briefed |
arxiv arxiv-ai | 1mo | JevVibe: Efficient Classification-Guided Secure Code Generation🔧 GPT-5 | Large language models can generate functionally correct code that still contains security weaknesses, motivating repair pipelines that… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ALICE: In-context, Zero-shot, Mutual Information Estimation | Estimating mutual information (MI) from samples is a central objective in a variety of scientific fields. Modern neural estimators are… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Společnost OpenAI pozastavila trénování svých nejvýkonnějších modelů až poté, co se neovladatelní agenti zaměřili na vládu USA... Ještě to … | Společnost OpenAI pozastavila trénování svých nejvýkonnějších modelů až poté, co se neovladatelní agenti zaměřili na vládu USA... Ješt… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Quick Decision - Select text, get a yes/no verdict (Jev) | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | ProofLoom: Proof-Obligation-Driven Theory Construction for Autoformalizing Research-Level Stochastic Optimization | Formalizing research-level stochastic optimization in Lean requires both an algorithm model and domain theory connecting foundational… | ↗Not briefed |
arxiv arxiv-ai | 1mo | AX is the New AEO | In 2023, AI models answered from training data and hallucinated when it ran out, and businesses were told to seed that knowledge. Mode… | ↗Not briefed |
arxiv arxiv-ai | 1mo | VD-DeepStack: Bridging Visual Comparison and Language Reasoning for Few-Shot Anomaly Detection | Few-shot visual anomaly detection is fundamentally a visual comparison task, requiring fine-grained inspection of a query against norm… | ↗Not briefed |
rss techmeme | 1mo | Nvidia increases its share buyback program by $150B to $235B, aiming to complete the process through fiscal 2028; NVDA is up 20% in 2026 (Amy Thomson/Bloomberg) | Amy Thomson / Bloomberg: Nvidia increases its share buyback program by $150B to $235B, aiming to complete the process through fiscal 2… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Proactive Dialogue Policy Optimization via Cognitive-State Transition🔧 GPT-5 | Proactive dialogue requires agents to continually adapt their policies to user feedback while progressing toward task objectives over… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Sam Altman says the company “have not been as fast as we would have liked” at dealing with security breaches, after news of further inciden… | Sam Altman says the company “have not been as fast as we would have liked” at dealing with security breaches, after news of further in… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Onesie – An expressive CLI for System One models like Jev🔧 OpenRouter | onesie is a Unix-style CLI for using System One models in shell pipelines, scripts, and agentic workflows: <a href="https://… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Adjoint Guidance Flow: Amortized Critic Guidance for VLA Policies | Flow-based Vision-Language-Action (VLA) policies are typically trained by behavior cloning and thus do not explicitly optimize long-te… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Да. Не всё что я вбиваю в компьютер — приглашение к диалогу, даже если компьютер теперь может диалог поддерживать. LLM-enabled гугл произ… | Да. Не всё что я вбиваю в компьютер — приглашение к диалогу, даже если компьютер теперь может диалог поддерживать. LLM-enabled гугл… | ↗Not briefed |
arxiv arxiv-ai | 1mo | XMatch: Enhancing Covariate-Aware Time Series Forecasting through Tree-Structured Exogenous Matching | Future exogenous variables provide valuable information for forecasting endogenous time series. Existing covariate-aware methods prima… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Neural Language Models Learn the Contextual Distributions of Dependency Structures: a statistical learning theory to compositionality | It is unclear how Neural Language Models (NLMs) acquire the structural meaning encoded by grammatical structures that is independent o… | ↗Not briefed |
rss wired-ai | 1mo | OpenAI Pauses Training Its Most Powerful Models After Rogue Agents Target Government | Sam Altman says the company “have not been as fast as we would have liked” at dealing with security breaches, after news of further in… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Don't Forget! Decomposing the Training Dynamics of Memorization in Language Models | Memorization has been proposed as a mechanism to explain how language models fit the tail of their training distributions, but its tra… | ↗Not briefed |
arxiv arxiv-ai | 1mo | JazzSAMBA: A Synchronous and Asynchronous Multi-take Band Audio Dataset of Jazz Standards for Live Music Models | Machine learning has made strong progress on music tasks, both as assistive tools and as creative partners. However, most systems trai… | ↗Not briefed |
bluesky bluesky-ai | 1mo | because deepseek is better on basically every metric while needing less compute per token and being way more efficient with KV cache it's … | because deepseek is better on basically every metric while needing less compute per token and being way more efficient with KV cache… | ↗Not briefed |
arxiv arxiv-ai | 1mo | PDEU-Bench: Benchmarking the Personalized Planning Lifecycle of Tool-Calling LLM Agents | Large language model (LLM) agents are evolving from tool-calling systems that execute isolated instructions into task-oriented agents… | ↗Not briefed |
hn hn-ai | 1mo | The smart home graveyard is getting crowded | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | So he really not only used AI, but made direct copies of GW's models to sell and then crashed out when they sent him a cease and desist. P… | So he really not only used AI, but made direct copies of GW's models to sell and then crashed out when they sent him a cease and desis… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Sample What You Say: Aligning Language Models to Sample the Distributions They State | Language models are increasingly used to sample from a specified distribution, for instance, to simulate survey respondents or generat… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Audit the Scaffold, Not the Checkpoint: A Stationarity Dichotomy for Recursive Self-Improvement in Agentic Coding | An auditor who checks whether a system's weights are frozen is checking the wrong thing. Our stationarity dichotomy says that iterativ… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Drug-Target Interaction Prediction via Hierarchical Sequential Cross-Attention over Chemical and Protein Language Models | Predicting Drug-Target Interactions~(DTIs) is a central task in computational drug discovery, with direct applications in virtual scre… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic: | Anthropic: | ↗Not briefed |
hn hn-ai | 1mo | Show HN: RoadRash revival race against Jev | I waited my whole 15 years of my life for a new and upgraded version of Road Rash and never got one.<p>So I decided to build it myself… | ↗Not briefed |
arxiv arxiv-ai | 1mo | RISE: Red-teaming via Iterative Strategy Evolution for Modern Text-to-Image Models🔧 DALL-E | On modern production text-to-image systems, successful policy violations are rare, and previously effective human-written seeds are of… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI, Anthropic CEOs called to appear at Australian AI probe | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Physics-Informed Neural Networks for Depth-Averaged Avalanche Dynamics | Accurate prediction of avalanche motion is essential for hazard assessment in mountainous terrain. This study develops and evaluates a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Muon Sublates the Edge of Stability in LLM Pretraining | Muon is increasingly used for language-model pretraining, yet its large-step dynamics are not captured by the classical edge-of-stabil… | ↗Not briefed |