ACIArena: Toward Unified Evaluation for Agent Cascading Injection
2026-04-09 • Artificial Intelligence
Artificial IntelligenceComputation and LanguageCryptography and Security
AI summaryⓘ
The authors focus on a security problem in Multi-Agent Systems (MAS), where one bad agent can trick others and cause big failures. They created a tool called ACIArena to test how strong these systems are against different types of attacks and defenses. Their tests show that just looking at the network layout isn't enough for safety; careful design of agent roles and how they interact is needed. Also, defenses made for simple setups often don't work well in real situations and can cause new problems. ACIArena helps researchers better understand and improve the safety of these agent networks.
Multi-Agent SystemsAgent Cascading InjectionMAS robustnessattack surfacesinter-agent trustsecurity benchmarkingrole designinteraction patternsinformation exfiltrationdefense mechanisms
Authors
Hengyu An, Minxi Li, Jinghuai Zhang, Naen Xu, Chunyi Zhou, Changjiang Li, Xiaogang Xu, Tianyu Du, Shouling Ji
Abstract
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.