arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2603.28166cs.CRcs.AI
收藏

评估具有现实工具的代理的特权使用

Evaluating Privilege Usage of Agents with Real-World Tools

Xinjiang University(新疆大学);Tsinghua University(清华大学)

Quan Zhang, Lianhang Fu, Lvsi Lian, Gwihwan Go, Yujue Wang, Chijin Zhou, Yu Jiang, Geguang Pu

AI总结:

本文提出GrantBox,一个用于评估代理特权使用的安全沙盒,通过集成真实工具和允许LLM代理调用真实特权,评估代理在提示注入攻击下的安全能力,发现LLM在面对复杂攻击时平均攻击成功率高达84.80%。

摘要:

为LLM代理配备现实世界工具可显著提升生产力。然而,授予代理自主使用工具的权限也同时将相关特权转移给代理和底层LLM。不当使用特权可能导致严重后果,包括信息泄露和基础设施损坏。尽管已有多个基准测试用于研究代理的安全性,但它们通常依赖于预编码工具和受限的交互模式。此类定制环境与现实世界差异显著,使评估代理在关键特权控制和使用方面的安全性变得困难。因此,我们提出GrantBox,一个用于分析代理特权使用的安全评估沙盒。GrantBox自动集成现实世界工具并允许LLM代理调用真实特权,从而在提示注入攻击下评估特权使用。我们的结果表明,尽管LLM表现出基本的安全意识并能阻止一些直接攻击,但它们仍易受更复杂攻击的攻击,导致在精心设计的场景中平均攻击成功率为84.80%。

英文摘要:

Equipping LLM agents with real-world tools can substantially improve productivity. However, granting agents autonomy over tool use also transfers the associated privileges to both the agent and the underlying LLM. Improper privilege usage may lead to serious consequences, including information leakage and infrastructure damage. While several benchmarks have been built to study agents' security, they often rely on pre-coded tools and restricted interaction patterns. Such crafted environments differ substantially from the real-world, making it hard to assess agents' security capabilities in critical privilege control and usage. Therefore, we propose GrantBox, a security evaluation sandbox for analyzing agent privilege usage. GrantBox automatically integrates real-world tools and allows LLM agents to invoke genuine privileges, enabling the evaluation of privilege usage under prompt injection attacks. Our results indicate that while LLMs exhibit basic security awareness and can block some direct attacks, they remain vulnerable to more sophisticated attacks, resulting in an average attack success rate of 84.80% in carefully crafted scenarios.

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。

↑