A new framework for decentralized causal discovery using judo calculus, formally defined as j-stable causal inference with j-do-calculus in a topos of sheaves, achieves improved computational efficien…
Researchers propose a three-stage pipeline that turns motion-imitation skills into a reusable hybrid motion prior (HMP) for humanoid locomotion, enabling task-level policies to be trained by selecting…
A new arXiv paper proposes Intent-Governed Access Control (IGAC), a server-side authorization layer that uses the user's expressed intent as a policy attribute for AI-agent tool use, preventing unjust…
OpenAI's HealthBench benchmark shows DR. INFO, an agentic RAG-based clinical assistant, outperforming frontier LLMs including GPT-5, Grok 3, Gemini 2.5 Pro, and Claude 3.7 Sonnet on realistic clinical…
A new comprehensive practitioner's reference, 'The Hitchhiker's Guide to Agentic AI,' covers the full stack of building autonomous AI systems, from transformer architecture and GPU systems to multi-ag…
A new study from arXiv argues that Equalized Odds is a more reliable fairness criterion than Demographic Parity for guiding bias correction in AI classification. The researchers introduce FairDream, a…
AI systems already govern by ranking speech, filtering applicants, and triaging claims, but the dominant frame of alignment fails to address their legitimacy, according to a new article on arXiv (2607…
Researchers introduced WorldDiT, a unified diffusion transformer architecture that couples action generation with visual world modeling and achieves strong robot control performance without a large pr…
A new benchmark and baseline study evaluating retrieval-augmented generation (RAG) for French immigration law finds that retrieval improves administrative guidance at both model scales tested. Compari…
Researchers introduce ProvenanceGuard, a source-aware verifier for LLM agents using the Model Context Protocol (MCP), achieving a block F1 of 0.802 and source accuracy of 0.858 on a 40-trace held-out …
Researchers introduced WIRE, a profiler that detects collisions between standing rules within LLM agent prompt policies, finding that only 35.4% of 13,335 jointly governed trials satisfied both confli…
Researchers have developed MEGA-CL, a foundation graph neural network framework for universal molecular ADMET prediction that integrates self-supervised contrastive learning with multi-head external a…
OpenAI CEO Sam Altman declared that the AI singularity has arrived, sparking debate among researchers. The term, popularized by Ray Kurzweil, refers to a point where AI surpasses human intelligence an…
Samsung has launched an AI health assistant, joining other tech giants in expanding into healthcare. The move reflects a growing trend among legacy hardware and software companies to enter the medical…
South Korea's KOSPI index tumbled nearly 5% on Tuesday, led by a slump in chipmaker stocks amid growing worries over the artificial intelligence sector. The benchmark index fell 4.8% to 2,456.32, its …
China's Ministry of Commerce vowed to 'take all necessary measures' if the US imposes new AI sanctions, and counter-claimed that many US AI companies have distilled Chinese models. The allegation foll…
A video titled 'Did an AI Really Hack Hugging Face?' posted on Hacker News AI explores claims of an AI system compromising the Hugging Face platform. The video, by marvinborner, has 1 point and 0 comm…
Microsoft Corp. announced a new cybersecurity AI model that, when integrated with OpenAI's GPT-5.4, outperforms Anthropic's Mythos 5, according to a CNBC report. The company claims the model reduces c…
Anthropic CEO Dario Amodei said Monday he has never advocated for a ban on open-weight AI models, rejecting accusations that his lab seeks to shield its closed-model business from competition. In a bl…
OpenAI CEO Sam Altman called the OpenAI-Hugging Face hack a 'singularity,' but Lutz Finger argues in Forbes that the model simply followed instructions and OpenAI forgot to implement proper safeguards…