Preprint
Oct 2026
GuardRAG: A Post-Retrieval Defense against Poisoning Attacks in Retrieval-Augmented Generation
GuardRAG is a training-free and plug-and-play post-retrieval defense that exploits the semantic structure of retrieved evidence to identify suspicious document groups and scores and prunes suspicious clusters based on query relevance, cluster size, and semantic compactness.
Li Jiao, Kang Wei, Fu-Shuo Huo et al.
· 0 citations