What we know
This paper proposes GroupGuard, a training-free defense framework against group collusive attacks in large language model-based multi-agent systems. The approach combines continuous graph-based monitoring, active honeypot inducement, and structural pruning to identify and isolate collusive agents. The authors report high detection accuracy across experiments with five datasets and four topologies.
Evidence limits
This entry is based solely on the publicly available title and abstract; the full text has not been reviewed and the reported results have not been independently verified.
Reviewed: author abstract. Full-text results were not analyzed.
Sources and provenance
- GroupGuard: A Framework for Modeling and Defending Collusive Attacks in Multi-Agent Systems
abstract · 2026-10-04
Reviewed: 2026-10-04. This record may change when new evidence is found.