arXiv 2602.031002026-02-03cs.AI

Risky-Bench: 探索现实部署中智能体安全风险

Risky-Bench: Probing Agentic Safety Risks under Real-World Deployment

National University of Singapore（国立新加坡大学）；University of Science and Technology of China（中国科学技术大学）；Southern University of Science and Technology（南方科技大学）；University of Electronic Science and Technology of China（电子科技大学）；Nanyang Technological University（南洋理工大学）；Shanghai Artificial Intelligence Laboratory（上海人工智能实验室）

Jingnan Zheng, Yanzhen Luo, Jingjun Xu, Bingnan Liu, Yuxin Chen, Chenhang Cui, Gelei Deng, Chaochao Lu, Xiang Wang, An Zhang, Tat-Seng Chua

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

Risky-Bench通过基于现实部署的安全原则，系统评估智能体在复杂任务中的安全风险，适用于多种部署场景。

摘要：

大型语言模型（LLMs）正越来越多地被部署为在现实环境中操作的智能体，这引入了超越语言伤害的安全风险。现有的智能体安全评估依赖于针对特定智能体设置的风险导向任务，导致对安全风险空间的覆盖有限，并且无法评估在复杂现实部署中长周期、交互式任务执行中的智能体安全行为。此外，它们对特定智能体设置的专一性限制了在不同智能体配置中的适应性。为了解决这些限制，我们提出了Risky-Bench，一个能够系统评估智能体安全性的框架，基于现实部署进行评估。Risky-Bench围绕领域无关的安全原则组织评估，推导出具有上下文感知的安全标准，界定安全空间，并通过在不同威胁假设下进行真实任务执行，系统地评估此空间中的安全风险。当应用于生命辅助智能体设置时，Risky-Bench揭示了在现实执行条件下，最先进的智能体中存在显著的安全风险。此外，作为一项结构良好的评估流程，Risky-Bench不仅限于生命辅助场景，可以适应其他部署设置，构建环境特定的安全评估，提供了一种可扩展的方法论用于智能体安全评估。

英文摘要：

Large Language Models (LLMs) are increasingly deployed as agents that operate in real-world environments, introducing safety risks beyond linguistic harm. Existing agent safety evaluations rely on risk-oriented tasks tailored to specific agent settings, resulting in limited coverage of safety risk space and failing to assess agent safety behavior during long-horizon, interactive task execution in complex real-world deployments. Moreover, their specialization to particular agent settings limits adaptability across diverse agent configurations. To address these limitations, we propose Risky-Bench, a framework that enables systematic agent safety evaluation grounded in real-world deployment. Risky-Bench organizes evaluation around domain-agnostic safety principles to derive context-aware safety rubrics that delineate safety space, and systematically evaluates safety risks across this space through realistic task execution under varying threat assumptions. When applied to life-assist agent settings, Risky-Bench uncovers substantial safety risks in state-of-the-art agents under realistic execution conditions. Moreover, as a well-structured evaluation pipeline, Risky-Bench is not confined to life-assist scenarios and can be adapted to other deployment settings to construct environment-specific safety evaluations, providing an extensible methodology for agent safety assessment.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。