paper-with-me

Papers

Agent-SafetyBench: Evaluating the Safety of LLM Agents

2024-12-19 · Zhexin Zhang, Shiyao Cui, Yida Lu, Jingzhuo Zhou, Junxiao Yang, Hongning Wang, Minlie Huang

As large language models (LLMs) are increasingly deployed as agents, their integration into interactive environments and tool use introduce new safety challenges beyond those associated with the models themselves. However, the absence of comprehensive benchmarks for evaluating agent safety presents a significant barrier to effective assessment and further improvement. In this paper, we introduce Agent-SafetyBench, a comprehensive benchmark designed to evaluate the safety of LLM agents. Agent-SafetyBench encompasses 349 interaction environments and 2,000 test cases, evaluating 8 categories of safety risks and covering 10 common failure modes frequently encountered in unsafe interactions. Our evaluation of 16 popular LLM agents reveals a concerning result: none of the agents achieves a safety score above 60%. This highlights significant safety challenges in LLM agents and underscores the considerable need for improvement. Through failure mode and helpfulness analysis, we summarize two fundamental safety defects in current LLM agents: lack of robustness and lack of risk awareness. Furthermore, our findings suggest that reliance on defense prompts alone may be insufficient to address these safety issues, emphasizing the need for more advanced and robust strategies. To drive progress in this area, Agent-SafetyBench has been released at https://github.com/thu-coai/Agent-SafetyBench/ to facilitate further research in agent safety evaluation and improvement.

📄 PDF Abstract BibTeX arXiv:2412.14470

Code (1)

thu-coai/agent-safetybench 공식 구현

Similar Papers 제목 키워드 기반

SkillSafetyBench: Evaluating Agent Safety under Skill-Facing Attack Surfaces

2026-05-12 · Chang Jin, An Wang, Zeming Wei, Kai Wang 외 arxiv

Reusable skills are becoming a common interface for extending large language model agents, packaging procedural guidance with access to files, tools, memory, and execution environments. However, this modularity introduce…

MobileSafetyBench: Evaluating Safety of Autonomous Agents in Mobile Device Control

2024-10-23 · Juyong Lee, Dongyoon Hahm, June Suk Choi, W. Bradley Knox 외

Autonomous agents powered by large language models (LLMs) show promising potential in assistive tasks across various domains, including mobile device control. As these agents interact directly with personal information a…

Automated Safety Benchmarking: A Multi-agent Pipeline for LVLMs

2026-01-27 · Xiangyang Zhu, Yuan Tian, Zicheng Zhang, Qi Jia 외 arxiv

Large vision-language models (LVLMs) exhibit remarkable capabilities in cross-modal tasks but face significant safety challenges, which undermine their reliability in real-world applications. Efforts have been made to bu…

SafetyBench: Evaluating the Safety of Large Language Models

2023-09-13 · Zhexin Zhang, Leqi Lei, Lindong Wu, Rui Sun 외

With the rapid development of Large Language Models (LLMs), increasing attention has been paid to their safety concerns. Consequently, evaluating the safety of LLMs has become an essential task for facilitating the broad…

Multiple-choice

MemEvoBench: Benchmarking Safety Risks from Memory Misevolution in LLM Agents

2026-04-17 · Weiwei Xie, Shaoxiong Guo, Fan Zhang, Tian Xia 외 arxiv

Equipping Large Language Models (LLMs) with persistent memory enhances interaction continuity and personalization but introduces new safety risks. Specifically, contaminated or biased memory accumulation can trigger abno…