ACIArena:迈向多智能体系统统一评估
ACIArena: Toward Unified Evaluation for Agent Cascading Injection
AI总结:
本文提出ACIArena框架,用于评估多智能体系统(MAS)的鲁棒性,通过多攻击面和目标的系统化测试,揭示单纯依赖拓扑结构评估不足,强调需通过角色设计和受控交互提升系统安全性。
摘要:
协作与信息共享增强了多智能体系统(MAS)的能力,但也引入了关键安全风险——智能体级联注入(ACI)。在此类攻击中,被篡改的智能体利用智能体间的信任传播恶意指令,导致系统级联失效。然而,现有研究仅考虑有限的攻击策略和简化的MAS设置,限制了其通用性和全面评估。为弥合这一差距,我们引入ACIArena,一个统一的框架用于评估MAS的鲁棒性。ACIArena提供覆盖多个攻击面(如外部输入、智能体配置、智能体间消息)和攻击目标(如指令劫持、任务中断、信息外泄)的系统化评估套件。具体而言,ACIArena建立了一个统一的规范,同时支持MAS构建和攻击防御模块。它涵盖了六个广泛使用的MAS实现,并提供1,356个测试用例的基准,以系统性地评估MAS的鲁棒性。我们的基准测试结果表明,仅通过拓扑结构评估MAS的鲁棒性是不够的;鲁棒的MAS需要精心设计的角色和受控的交互模式。此外,简化的环境中开发的防御措施往往无法转移到现实世界中;狭隘范围的防御甚至可能引入新的漏洞。ACIArena旨在为深入探索MAS设计原则提供坚实的基石。
英文摘要:
Collaboration and information sharing empower Multi-Agent Systems (MAS) but also introduce a critical security risk known as Agent Cascading Injection (ACI). In such attacks, a compromised agent exploits inter-agent trust to propagate malicious instructions, causing cascading failures across the system. However, existing studies consider only limited attack strategies and simplified MAS settings, limiting their generalizability and comprehensive evaluation. To bridge this gap, we introduce ACIArena, a unified framework for evaluating the robustness of MAS. ACIArena offers systematic evaluation suites spanning multiple attack surfaces (i.e., external inputs, agent profiles, inter-agent messages) and attack objectives (i.e., instruction hijacking, task disruption, information exfiltration). Specifically, ACIArena establishes a unified specification that jointly supports MAS construction and attack-defense modules. It covers six widely used MAS implementations and provides a benchmark of 1,356 test cases for systematically evaluating MAS robustness. Our benchmarking results show that evaluating MAS robustness solely through topology is insufficient; robust MAS require deliberate role design and controlled interaction patterns. Moreover, defenses developed in simplified environments often fail to transfer to real-world settings; narrowly scoped defenses may even introduce new vulnerabilities. ACIArena aims to provide a solid foundation for advancing deeper exploration of MAS design principles.
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。