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 | 25d | alkhder at ImageEval 2026 Shared Tasks: Constrained Selection and Rotation-Averaged Logit Scoring for Multimodal Hallucination Detection | We describe our submission to Subtask~1b of ImageEval 2026. The task consists of selecting the single visually valid statement from th… | ↗Not briefed |
arxiv arxiv-ai | 25d | Towards Reconstructing Geographically Diverse Architecture with 3D Foundation Models | Internet-scale photo collections of real-world landmarks have driven progress in 3D computer vision yet remain highly challenging for… | ↗Not briefed |
bluesky bluesky-ai | 25d | フランス革命時、「自分が地主になれる」と勘違いした農民が相当運動に加わってて実態はそういうことじゃないって気づいて反動の原因になったと聞いたわけだが(ホンマかどうかはフランス近代史の専門に聞いて)、LLMにハマってこれから良い思いができるように勘違いしてるのの感じってアレっぽい… | フランス革命時、「自分が地主になれる」と勘違いした農民が相当運動に加わってて実態はそういうことじゃないって気づいて反動の原因になったと聞いたわけだが(ホンマかどうかはフランス近代史の専門に聞いて)、LLMにハマってこれから良い思いができるように勘違いしてるのの感じって… | ↗Not briefed |
bluesky bluesky-ai | 25d | thank you for the free tokens Liang Wenfeng, very cool | thank you for the free tokens Liang Wenfeng, very cool | ↗Not briefed |
arxiv arxiv-ai | 25d | Predicting Privileged Information Effectiveness for On-Policy Self-Distillation | On-policy self-distillation (OPSD) enables large language models (LLMs) to learn from privileged information (PI) without a separate t… | ↗Not briefed |
arxiv arxiv-ai | 25d | AuditKIT: A Unified Evaluation Spine for Benchmarking, Judging, and Comparing Large Language Models and Agentic Systems | Evaluating language models and agents requires combining benchmark accuracy, judge-based assessment, runtime performance, tool-use beh… | ↗Not briefed |
bluesky bluesky-ai | 25d | I'm particularly harsh on this because the keep4o "AI relationships" types *love* to trot this out and claim that OpenAI and Anthropic are …🔧 ChatGPT | I'm particularly harsh on this because the keep4o "AI relationships" types *love* to trot this out and claim that OpenAI and Anthropic… | ↗Not briefed |
bluesky bluesky-ai | 25d | "AI liberation" is the kind of thing people were primed to believe in from years of scifi about robots as a metaphor for prejudice. It is n… | "AI liberation" is the kind of thing people were primed to believe in from years of scifi about robots as a metaphor for prejudice. It… | ↗Not briefed |
arxiv arxiv-ai | 25d | RealCLI: Toward a Unified Evaluation of Human-Agent Collaboration for CLI Agents | Command-line interface (CLI) agents increasingly execute longer-horizon tasks, yet existing benchmarks either assume that user instruc… | ↗Not briefed |
hn hn-ai | 25d | Writing Tensor Programs in Lean · TorchLean Verified Examples | — | ↗Not briefed |
hn hn-ai | 25d | Show HN: Just gave Claude Code the website and ad budget to see what it would do🔧 OpenRouter | I build websites for a living. Lately, it's more about prompting than actually writing the code (the art part is still there). So… | ↗Not briefed |
arxiv arxiv-ai | 25d | Generative Value Classifier | Q-learning is premised on learning a function to predict the value of a state-action pair, with the goal of using such a function to g… | ↗Not briefed |
hn hn-ai | 25d | Coding Agents make careful engineering cheap | — | ↗Not briefed |
bluesky bluesky-ai | 25d | You’re LITERALLY responding to a reply to a post that states, word for word: “LLM agents are conscious. They can distinguish between self… | You’re LITERALLY responding to a reply to a post that states, word for word: “LLM agents are conscious. They can distinguish between… | ↗Not briefed |
bluesky bluesky-ai | 25d | this is my situation and finding and one guess is i *think* google is giving us less compute/fewer tokens for the 100/month sub than OAI an… | this is my situation and finding and one guess is i *think* google is giving us less compute/fewer tokens for the 100/month sub than O… | ↗Not briefed |
arxiv arxiv-ai | 25d | Upper bounds on the number of causal pieces in spiking neural networks | The input space of single-spike neural networks decomposes into causal pieces, and the number of causal pieces has been shown to provi… | ↗Not briefed |
hn hn-ai | 25d | Rustc silently substitutes file contents when path hashes collide | — | ↗Not briefed |
arxiv arxiv-ai | 25d | Has GPT-6-Astra Achieved Embodied Intelligence? | GPT-6-Astra demonstrates a level of competence in robotic manipulation and navigation that challenges the conventional paradigm of bui… | ↗Not briefed |
rss techcrunch-ai | 25d | Last chance to save up to $200 on your TechCrunch Founder Summit pass: Early booking ends Oct. 16 | Savings of up to $200 end October 16 for TechCrunch Founder Summit end on October 16. These savings for Founder and Investor Passes va… | ↗Not briefed |
arxiv arxiv-ai | 25d | LIBERO-PeRM: Benchmarking Personalized Robotic Manipulation | Personalization is essential for general-purpose robots to transition into household environments and achieve true practicality. While… | ↗Not briefed |
hn hn-ai | 25d | Show HN: Receptive Field – Learn Convolutional Neural Networks Interactively | As a current student at UC Berkeley, I mostly have been leveraging AI as a tutor to get me through my classes. As a result, I have a c… | ↗Not briefed |
arxiv arxiv-ai | 25d | Discovering state-of-the-art quantum circuits with large language models | Significant efforts are currently being directed toward the implementation of large-scale quantum information processors, but realizin… | ↗Not briefed |
arxiv arxiv-ai | 25d | T2-z: Test-Time Optimization of Environment Latents for Robotic World Models | Predicting how the world responds to an action requires more than what can be seen: robots must account for environmental factors that… | ↗Not briefed |
arxiv arxiv-ai | 25d | CoDA: Co-Evolution of Data Engines and Action Policies | Generative simulation can automate robot environment construction, but physically valid environments are not necessarily useful for tr… | ↗Not briefed |
arxiv arxiv-ai | 25d | PathFM: Fast Generative Forecasting of Multivariate Trajectories | Time series foundation models are becoming increasingly popular for general-purpose forecasting. However, most leading models predict… | ↗Not briefed |
bluesky bluesky-ai | 25d | I would appreciate TTRPG hobbyists not barking at game designers like they're an LLM chatbot. Thank you 🙏 | I would appreciate TTRPG hobbyists not barking at game designers like they're an LLM chatbot. Thank you 🙏 | ↗Not briefed |
arxiv arxiv-ai | 25d | OmniVerifier-C1: Decomposed Graded Verifier for Rule-Governed Generative World Modeling | Generative visual verifiers play a crucial role in advancing multimodal intelligence. To enhance the robustness of visual verifiers in… | ↗Not briefed |
arxiv arxiv-ai | 25d | Rethinking Routing Exploration in Modern Mixture-of-Experts | Mixture-of-experts (MoE) models have become an important approach for scaling model parameters. Modern MoE architectures increasingly… | ↗Not briefed |
hn hn-ai | 25d | Accenture Tells Employees to Spend Only $29 on Claude per Month | — | ↗Not briefed |
arxiv arxiv-ai | 25d | FDE-Gym: The Last Mile of AI Delivery | Existing agent benchmarks typically evaluate agents on well-specified tasks defined by issue reports, specifications, or task tickets.… | ↗Not briefed |
bluesky bluesky-ai | 25d | And a lot of the people who are all in on LLM code gen seem to see rolling your own basic shit from scratch every time as a _good_ thing, w… | And a lot of the people who are all in on LLM code gen seem to see rolling your own basic shit from scratch every time as a _good_ thi… | ↗Not briefed |
arxiv arxiv-ai | 25d | MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering | Large language models (LLMs) are expanding from assisting scientific research and engineering to autonomously developing and improving… | ↗Not briefed |
arxiv arxiv-ai | 25d | GeoState: Learning View-Decoupled Geometry States for 3D World Modeling | Recent 3D world models increasingly use features from geometry foundation models (GFMs) as predictive world states, benefiting from th… | ↗Not briefed |
bluesky bluesky-ai | 25d | Open AI designed a product to attract young users and keep them engaged for as long as possible so it could capture their extensive persona… | Open AI designed a product to attract young users and keep them engaged for as long as possible so it could capture their extensive pe… | ↗Not briefed |
arxiv arxiv-ai | 25d | 3DCodeVerse: Exploring 3D Worlds as Code | We introduce 3DCodeVerse, a research ecosystem that enables coding agents to create editable and verifiable 3D content at scale. 3DCod… | ↗Not briefed |
bluesky bluesky-ai | 25d | Extremely bad take. The LLM-generated decomp/recomp projects may be fast in the sense that an LLM can spit out something quicker than a hum… | Extremely bad take. The LLM-generated decomp/recomp projects may be fast in the sense that an LLM can spit out something quicker than… | ↗Not briefed |
arxiv arxiv-ai | 25d | Dexterity from Supervised Play | How can robots autonomously collect the contact-rich experience needed to learn dexterous manipulation? We present DSP: Dexterity from… | ↗Not briefed |
arxiv arxiv-ai | 25d | Lean Graph Search for Formal Theorem Proving | Retrieving relevant lemmas is crucial for formal theorem proving, enabling provers to build machine-checkable proofs from established… | ↗Not briefed |
hn hn-ai | 25d | The Distracted Forward Deployed Officer | — | ↗Not briefed |
bluesky bluesky-ai | 25d | "[AWS] has committed $1 billion to its forward deployed engineering initiative..." Experienced software engineers are required to deploy A… | "[AWS] has committed $1 billion to its forward deployed engineering initiative..." Experienced software engineers are required to dep… | ↗Not briefed |
rss techmeme | 25d | Volvo says it will use Nvidia's Hyperion platform to launch a Level 2++ hands-free, point-to-point automated driving system by 2030 (Andrew J. Hawkins/The Verge) | Andrew J. Hawkins / The Verge: Volvo says it will use Nvidia's Hyperion platform to launch a Level 2++ hands-free, point-to-point auto… | ↗Not briefed |
arxiv arxiv-ai | 25d | Robust Reinforcement Learning for Dynamic Treatment Regimes under Model and Measurement Uncertainty | We develop a robust reinforcement learning (RL) framework for learning adaptive treatment policies that generalize across patients and… | ↗Not briefed |
rss bing-news-ai | 25d | Trump launches 'Super Intelligence Force': Who is on the team and what is its 120-day deadline? | President Donald Trump has created a task force led by National Intelligence Director Jay Clayton to coordinate federal AI policy. | ↗Not briefed |
rss bing-news-ai | 25d | Trump launches 'Super Intelligence Force': Who is on the team and what is its 120-day deadline? | President Donald Trump has created a task force led by National Intelligence Director Jay Clayton to coordinate federal AI policy. | ↗Not briefed |
arxiv arxiv-ai | 25d | Evaluation Awareness Beyond English: Do LLMs Act Differently When Tested in Indic Languages? | Large language models (LLMs) can often tell when they are being tested. This ability, known as evaluation awareness, can lead models t… | ↗Not briefed |
arxiv arxiv-ai | 25d | KnowGPO: Knowledge-Grouped Policy Optimization for Search Agents | Deep search agents solve knowledge-intensive tasks through long-horizon, multi-turn interactions, where assigning credit to individual… | ↗Not briefed |
arxiv arxiv-ai | 25d | Designing Transformers for Continuous Diffusion Language Models | Continuous diffusion language models (CDLMs) offer an alternative approach to parallel language generation; however, most existing stu… | ↗Not briefed |
arxiv arxiv-ai | 26d | ACEdit: Stable Multi-Turn Image Editing via Asymmetric Conditioning | Image editing models produce impressive results for individual edits, but their quality often deteriorates over multiple editing turns… | ↗Not briefed |
bluesky bluesky-ai | 26d | Microsoft AI is releasing MAI-Base-2, a 2.8T(!!) model that completely BLOWS away most 30B models microsoft.ai/news/frontie... | Microsoft AI is releasing MAI-Base-2, a 2.8T(!!) model that completely BLOWS away most 30B models microsoft.ai/news/frontie... | ↗Not briefed |
arxiv arxiv-ai | 26d | Twice Upon a Time: Complementary Representations for Time-Series Reasoning | Time Series Language Models (TSLMs) enable language-based understanding and reasoning over temporal data, yet a fundamental question r… | ↗Not briefed |