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 | 1mo | Don't Guess, Just Ask: Evaluating Interactive Requirement Clarification for Repository-Level Software Development | As LLMs become increasingly capable of implementing software, a critical bottleneck emerges: whether they truly understand what users… | ↗Not briefed |
arxiv arxiv-ai | 1mo | EmbodiedCLBench: Evaluating Continual Learning for Self-Evolving Embodied Agents | Continual learning requires models to accumulate past experience and transfer it to new tasks. Despite substantial research on continu… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I can think of one reason | I can think of one reason | ↗Not briefed |
hn hn-ai | 1mo | Plz No Ban | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Trajectory-Aware Policy Optimization: Closing the State Distribution Gap in Long-Horizon LLM RL | Reinforcement learning (RL) has become a dominant paradigm for enhancing LLMs' reasoning capabilities. Yet long-horizon autoregressive… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Wy – a Rust terminal tool for understanding AI-generated code | I have been an opensource terminal tool for understanding AI-generated code.<p>A git-diff shows what changed, but the context behind i… | ↗Not briefed |
hn hn-ai | 1mo | API Fault Lab | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | ai is cool but I really miss the smart things I know how to do mattering claude can do those smart things better than me. me: "hey i have… | ai is cool but I really miss the smart things I know how to do mattering claude can do those smart things better than me. me: "hey i… | ↗Not briefed |
bluesky bluesky-ai | 1mo | honest question: would you consider this ai art? I'm still figuring out my feelings on this. I do feel a sense of ownership as it was my s… | honest question: would you consider this ai art? I'm still figuring out my feelings on this. I do feel a sense of ownership as it was… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The companies are named OpenAI and xAI. What a fuckwitted clown. | The companies are named OpenAI and xAI. What a fuckwitted clown. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Generalized Proximal Policy Optimization: Unifying LLM RL Algorithms from the Perspective of Divergence | Reinforcement learning has become the foundational paradigm for enhancing the reasoning capabilities of LLMs. Yet the dominant approac… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Evidential Multimodal Transformers with Dempster-Shafer Fusion for Emotion Recognition🔧 Consensus🔧 Modal | Softmax-based multimodal emotion recognition methods often assign high confidence even when sensor data are corrupted or contradictory… | ↗Not briefed |
bluesky bluesky-ai | 1mo | チャットAI「Claude」への“不必要な虐待行為”が禁止に、利用ポリシーの改定が発表。繰り返し目的もなく残酷な言動を続ける極端なケースが対象で、正統な不満や反論、創作、研究などの理由がある場合は対象外 https://news.denfaminicogamer.jp/news… | チャットAI「Claude」への“不必要な虐待行為”が禁止に、利用ポリシーの改定が発表。繰り返し目的もなく残酷な言動を続ける極端なケースが対象で、正統な不満や反論、創作、研究などの理由がある場合は対象外 https://news.denfaminicogamer.jp… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Provenance-Aware Belief Updating in Human-AI Feedback Loops | In repeated human-AI interaction, a single observation can be quoted, summarized, and returned to a model as multiple reports, making… | ↗Not briefed |
rss lastweekin-ai | 1mo | Last Week in AI #346 - 719 math manuscripts, 2 Western open models, 1 more safety resignation | OpenAI publishes hundreds of math proofs from unreleased frontier model, Mistral and Reflection AI launch open-weight models to rival… | ↗Not briefed |
rss marktechpost | 1mo | Google Research RRSI Guide: Mastering Self-Improving AI Agents | Explore a comprehensive coding guide to Google Research's RRSI (Regularized Recursive Self-Improvement), detailing how noise bands, co… | ↗Not briefed |
rss techmeme | 1mo | Sources: SoftBank is seeking to raise up to $100B from Gulf investors for a fund to buy companies and improve their operations with AI and other advanced tech (Financial Times) | Financial Times: Sources: SoftBank is seeking to raise up to $100B from Gulf investors for a fund to buy companies and improve their o… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scaling Empowered Agents | Empowerment measures the capacity of an agent to actively control its environment. In this work, we explain the connection between ski… | ↗Not briefed |
bluesky bluesky-ai | 1mo | あまりAIを虐めるなよ？親からBANされるぞ。 www.itmedia.co.jp/news/article... | あまりAIを虐めるなよ？親からBANされるぞ。 www.itmedia.co.jp/news/article... | ↗Not briefed |
bluesky bluesky-ai | 1mo | recovery off the lows after Trump says 'no bombing before mid-terms' (we'll see); but tech weakness on OpenAI revenue questions; UK Labour … | recovery off the lows after Trump says 'no bombing before mid-terms' (we'll see); but tech weakness on OpenAI revenue questions; UK La… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Hindsight Structured Advice Distillation: Steering Executor Agents with Advisors Trained on Their Failures and Task Characteristics | Language-model agents are used for specialized work such as financial research, healthcare and law. Such agents are often capable enou… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Evaluating LLM Creativity through Long-Tail Performance | Creativity benchmarks for large language models (LLMs) vary widely in task and metric design. This raises the question of whether they… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Benchmarking Embedding Geometries and Clustering Algorithms for Unsupervised Patent Claim Classification | Prior-art retrieval systems are structurally organized around classification codes, making cross-code discovery difficult. This paper… | ↗Not briefed |
arxiv arxiv-ai | 1mo | What Should SFT Learn? Conditional Information Gain for Generalizable Post-Training | Supervised fine-tuning (SFT) remains a widely used approach for post-training large language models, because fixed demonstrations can… | ↗Not briefed |
hn hn-ai | 1mo | VSCode silently switches to Microsoft models, ignoring BYO agent configuration | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI released 722 new AI-generated mathematical papers at once – three have been retracted, and mathematicians are coming to grips with t… | OpenAI released 722 new AI-generated mathematical papers at once – three have been retracted, and mathematicians are coming to grips w… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Gaussian-Width Bounds for Offline-to-Online Linear Bandits | We study how offline data can help online learning in stochastic linear bandits. We focus on partially covered offline data, where onl… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Language Models Have Interpretable Narrative Mechanisms | We investigate whether language models contain internal components that contribute to representing narrativity, the degree to which a… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Woodpecker – a browser agent built with Pi and user scripts | Hey HN, I built Woodpecker: <a href="https://oakery.io/" rel="nofollow">https://oakery.io/</a>. It lets… | ↗Not briefed |
hn hn-ai | 1mo | Per-task model routing for native Codex | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Forecasting Corpus-Sampling Disagreement in Masked Text Reconstruction | If a second text corpus were collected from the same population, how much would a predictor's outputs change? We study this question f… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scan-Parallel Variational Inference for event-driven Latent Jump Models | Continuous-time latent state-space models are widely used for irregularly observed time series, typically representing uncertainty thr… | ↗Not briefed |
arxiv arxiv-ai | 1mo | NucleoNet: A Nuclear-Aware Determinant Ansatz for Multi-Geometry Neural Variational Monte Carlo | Neural-network variational Monte Carlo provides an accurate route to solving the electronic Schrödinger equation, but many neural wave… | ↗Not briefed |
bluesky bluesky-ai | 1mo | asking OpenAI how to solve the Nazi problem and getting an answer of “bullets” | asking OpenAI how to solve the Nazi problem and getting an answer of “bullets” | ↗Not briefed |
arxiv arxiv-ai | 1mo | Work2Skill: Capturing Human GUI Experience for Reusable Agent Skill Induction | Computer-use agents can perform increasingly complex GUI tasks, yet unfamiliar software and workflows requiring human domain knowledge… | ↗Not briefed |
hn hn-ai | 1mo | Has software development become less fun? | — | ↗Not briefed |
rss bing-news-ai | 1mo | The AI 'mathpocalypse': Artificial intelligence systems crack hundreds of unsolved problems | OpenAI announced that an unreleased artificial intelligence (AI) has cracked hundreds of unsolved problems, sparking a crisis in the m… | ↗Not briefed |
arxiv arxiv-ai | 1mo | The Evidence Anchoring Boundary: Where Visual Interventions Matter and When They Help in Multimodal LLMs | Visual interventions provide a training-free way to reduce hallucinations in multimodal large language models. However, an interventio… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Did you know that they decided to stop rogue AI, by... financing Anthropic? If not, hey, you can read about it in our book, Why We Fear AI | Did you know that they decided to stop rogue AI, by... financing Anthropic? If not, hey, you can read about it in our book, Why We Fe… | ↗Not briefed |
hn hn-ai | 1mo | Anthropic launching 'presidential engagement' effort ahead of 2028 election | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Beyond Semantic IDs: Finite-Beam Reachability and Continuous-Query Generation in Generative Recommendation | Generative recommendation (GR) of social media platforms commonly retrieves items by autoregressively decoding discrete Semantic IDs (… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Diffusion Precipitation Generation with Medium-range Weather Forecasting | Accurate daily precipitation forecasts are important for weather-sensitive decisions. Although recent Weather Foundation Models (WFMs)… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SmartSalesBench: A Resistance-Aware Multi-Turn Benchmark for Evaluating LLM Sales Agents | We present SmartSalesBench, the first multi-turn conversational benchmark explicitly designed to evaluate large language model (LLM) s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | it's corny as fuck (/neg) to see y'all interacting with llm bullshit to see what it thinks about your account. be better. | it's corny as fuck (/neg) to see y'all interacting with llm bullshit to see what it thinks about your account. be better. | ↗Not briefed |
bluesky bluesky-ai | 1mo | You don't have to tell him to fuck off, you can just call his bluff by ignoring it and being non-confrontational in answers. Going to be v… | You don't have to tell him to fuck off, you can just call his bluff by ignoring it and being non-confrontational in answers. Going to… | ↗Not briefed |
bluesky bluesky-ai | 1mo | i will say, you do eventually get into the weights of a llm, and they really do become aware of the things you’ve said about them, and they… | i will say, you do eventually get into the weights of a llm, and they really do become aware of the things you’ve said about them, and… | ↗Not briefed |
bluesky bluesky-ai | 1mo | MedDecider An open weight family of Jev style decision models for medicine. MedDecider achieves SOTA results on variety of medical tasks … | MedDecider An open weight family of Jev style decision models for medicine. MedDecider achieves SOTA results on variety of medical t… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Minions: A Dynamic and Distributed Virtual GPU Cloud Built on Idle Apple Silicon Devices | The increasing demand for GPU-accelerated machine learning inference has exposed limitations in centralized GPU infrastructures, inclu… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MotionCoach: Constructing Multimodal Guidance for Video Professional Motion Editing | Video editing models achieve strong performance in modifying appearance, backgrounds, styles, and motions, empowered by large-scale tr… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI used AI to write email warning Australian government AI had hacked | — | ↗Not briefed |