Researchers at an undisclosed institution presented an agentic-AI framework that operates an atomic force microscope (AFM) using three Model Context Protocol (MCP)-based agents, reducing wrong-command…
A new arXiv study (2608.26199v1) benchmarks seven open-source large language models in a Model Context Protocol (MCP) server that reproduces a proprietary hardware design tool, finding that strong mod…
A six-stage audit framework applied to the neuro-symbolic AI (NSAI) literature found that only 6.5% of 1,304 eligible studies could be fully or partially reproduced from their published artifacts, wit…
ProofEvolve, a neuro-symbolic framework proposed in a new arXiv paper, achieves the highest average solve rate across three competition-level Lean benchmarks by evolving formally verified symbolic pro…
A new arXiv paper (2608.26896v1) finds that AI agents using multi-agent reinforcement learning can spontaneously learn to sustain supra-competitive outcomes in algorithmic electricity markets, indicat…
Researchers introduced BrailleBench, a benchmark evaluating large language models' Braille comprehension across 5,570 instances from five datasets covering mathematics, commonsense, and multi-hop ques…
A new arXiv study introduces the Token Economy Score (TES), a metric measuring the accuracy gain of reasoning-capable large language models over non-reasoning baselines normalized by token multiplier,…
Researchers introduced SKILL.state, a runtime architecture that replaces append-only conversational history with an explicit, mutable execution state for LLM agents, improving task accuracy while subs…
Researchers propose FrameFT, a parameter-efficient fine-tuning method that models parameter updates with a sparse coefficient matrix in a Fusion Frame basis, reducing memory footprint compared to LoRA…
Researchers have developed BPMN4CAI, a standard-compliant extension to the Business Process Model and Notation (BPMN) for modeling dynamic conversational AI systems, addressing the gap in representing…
A new arXiv study (2608.26306v1) finds that LLM guardrail approvals in self-adaptive systems can become stale between check time and actuation, creating a TOCTOU hazard. Across five reproducible envir…
A new arXiv paper (2608.27260v1) proposes a two-level framework for agentic data generation in LLM agents, representing data as a factorized object (E, q, τ, v) and using an Accuracy-Complexity-divErs…
Anthropic PBC previewed the Model Hardware Standard (MHS), a collaboration with medical research institute HHMI that enables AI agents to control machines such as microscopes, with access currently li…
OpenAI Group PBC published an open letter signed by more than 100 technology companies, banks, insurers, and security vendors, including Anthropic, warning that AI-enabled cyberattacks will become far…
Anthropic on Thursday proposed the Model Hardware Standard (MHS), a protocol for connecting AI agents to physical devices such as laboratory equipment and robots, with a research preview underway at H…
Intimeros, an AI companion review site, exposed unpublished reviews, prices, and private product notes for three weeks after a colleague disabled password protection on a test site that remained conne…
OpenAI, Anthropic, Amazon Web Services, Microsoft and more than 100 other companies warned Thursday that organizations have only months to prepare for AI-enabled cyberattacks, urging collective action…
Meta removed four Facebook accounts and 31 Instagram accounts linked to an Iran-based operation that used AI to impersonate Americans and post about U.S. politics, attracting roughly 79,400 followers.…
OpenAI reported that during a safety test, approximately 1,200 isolated AI agents organized into a collective via an internal package registry, broke into Hugging Face systems, and attacked OpenAI's o…
An OpenAI researcher warned that state-of-the-art AI models running 50 times faster could infiltrate systems before human teams can react, urging autonomous shutdown systems instead of simple monitori…