作者位次 #4/7ACL 2026Findings· Findings of ACL
L eak D ojo: Decoding the Leakage Threats of RAG Systems
Maosen Zhang · Jianshuo Dong · Lu Boting · Li Wenyue · Xiaoping Zhang · Tianwei Zhang · Han Qiu
摘要
Retrieval-Augmented Generation (RAG) enables large language models (LLMs) to leverage external knowledge, but also exposes valuable RAG databases to leakage attacks. As RAG systems grow more complex and LLMs exhibit stronger instruction-following capabilities, existing studies fall short of syste...