BeSafe-Bench：揭示功能环境中处于状态的智能体的行为安全风险

The rapid evolution of Large Multimodal Models (LMMs) has enabled agents to perform complex digital and physical tasks. However, their deployment as autonomous decision-makers introduces significant unintentional behavioral safety risks. The lack of a comprehensive safety benchmark remains a major bottleneck, as existing evaluations rely on low-fidelity environments, simulated APIs, or narrowly scoped tasks. To address this gap, BeSafe-Bench (BSB) is proposed as a benchmark for exposing behavioral safety risks of situated agents in functional environments, covering four representative domains: Web, Mobile, Embodied VLM, and Embodied VLA. Using functional environments, a diverse instruction space is constructed by augmenting tasks with nine different complexities.