#agent safety

Seven Papers Define Agent Benchmarking, Safety, and Efficiency

Latest arxiv papers tackle creative reasoning, smaller models, clinical tracking, safety judgment, workspace dependencies, plan quality, and visual exploration.