一作ACL 2026Poster· Natural Language Processing
Explainable and Fine-Grained Safeguarding of LLM Multi-Agent Systems via Bi-Level Graph Anomaly Detection
Junjun Pan · Yixin Liu · Rui Miao · Kaize Ding · Yu Zheng · Quoc Viet Hung Nguyen · Alan Wee-Chung Liew · Shirui Pan
摘要
Large language model (LLM)-based multi-agent systems (MAS) have shown strong capabilities in solving complex tasks. As MAS become increasingly autonomous in various safety-critical tasks, detecting malicious agents has become a critical security concern. Although existing graph anomaly detection...