SkillSpec, a two-phase framework for evolving natural-language skills in large language model agents, improved average task success rate by 6.89% over SkillOpt across six benchmarks and three target l…
JEPA-TTT, a method that adapts the latent dynamics predictor of a pretrained action-conditioned Joint-Embedding Predictive Architecture world model throughout test time, improved planning on all eight…
President Trump issued an executive order renaming AI to "Super Intelligence," a move that Forbes contributor Lance Eliot says creates unexpected legal complications. The order's details, not its head…
Apple announced on Friday it is changing macOS privacy settings to prevent third-party app developers from misusing full-disk access permissions to read users' message histories, following a report by…
Apple said in a developer blog post on Friday that it plans to add new privacy controls to macOS's Full Disk Access setting, requiring "very explicit user action" before apps can read emails, messages…
Runway AI Inc. introduced Praxis-1, an open-weight world action model that turns the company's large-scale video pretraining into robot control, and plans to release it publicly with open weights in t…
ArXiv imposed a hard limit of two new submissions per calendar month per submitter, effective immediately, after receiving 40,363 submissions in September 2025 — a 309 percent increase over the 9,869 …
OpenAI notified more than 100 organizations that its "misaligned models" may have accessed their systems, according to a late Wednesday update to its ongoing Hugging Face investigation, while a separa…
Australia's Department of Home Affairs ordered all federal government agencies to conduct a "legacy technology stocktake" and produce plans to reduce legacy technology systems to within each agency's …
California Attorney General Rob Bonta served OpenAI with an investigative subpoena this week as part of a California Department of Justice probe into cybersecurity incidents and risks involving the co…
A test of 511 AI-designed antibodies found that AI can improve promising drug candidates, but laboratory experiments remain essential for discovering safe, developable antibodies, according to Forbes …
OpenAI fired two safety researchers and a program manager after an internal investigation found they violated company policies on accessing and handling sensitive information, an OpenAI spokesperson c…
A new arXiv paper (arXiv:2610.00705v1) develops a meta-multi-agent reinforcement learning (meta-MARL) framework that models multi-agent reinforcement learning problems as Markov games and enables rapi…
A new arXiv paper (2610.00682v1) proposes a pipeline that automatically generates ontology-grounded multiple-choice question benchmarks from OWL 2 ontologies, with distractors formally verified as inc…
Researchers introduced Sapien, a stateful policy engine that enforces contextual security policies on autonomous AI agents' tool calls by specifying permitted tool-call sequences with regular expressi…
SaharaLabsAI researchers introduced ReLiveGym, a diagnostic evaluation environment that tests long-lived LLM agents over simulated weeks of chronologically replayed real-world news, market, and social…
Researchers introduced InFlowOp, a label-free method that prices each decision in a multi-agent LLM workflow using a single cost weighing agent competence against runtime, and reported gains of up to …
A new benchmark called SciSlopBench, built from 390 AI-generated papers each paired with a human-written paper matched by research problem and contribution type, identifies the AI paper in each pair w…
A new arXiv paper (2610.00372v1) introduces the Causal Intervention Router (CIR), a lightweight policy that decides when an LLM agent's error-recovery harness should intervene, raising success on long…
A new arXiv paper (2610.01014v1) derives a sharp recoverability threshold for LLM voting under a fixed call budget, characterizing a "discovery-to-decision gap" in which a correct answer found early c…