arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2604.07403cs.CR
收藏

RefineRAG:通过检索引导的文本细化实现词级污染攻击

RefineRAG: Word-Level Poisoning Attacks via Retriever-Guided Text Refinement

Huazhong University of Science and Technology(华中科技大学);Beihang University(北京航空航天大学)

Ziye Wang, Guanyu Wang, Kailong Wang

AI总结:

RefineRAG通过检索引导的文本细化方法,解决传统污染攻击检测不足的问题,实现高成功率的词级污染攻击,同时保持自然性和低语法错误。

摘要:

检索增强生成(RAG)显著增强了大型语言模型(LLMs),但同时暴露了对知识污染攻击的关键漏洞。现有攻击方法如PoisonedRAG由于粗粒度的分开-拼接策略仍可检测。为此,我们提出RefineRAG,一种将污染视为整体词级细化问题的新框架。它分两阶段:宏生成产生保证诱导目标答案的毒种,微细化采用检索循环优化以最大化检索优先级而不牺牲自然性。在NQ和MSMARCO上的评估显示,RefineRAG实现了最先进的有效性,实现了90%的NQ攻击成功率,同时在所有基线中记录最低的语法错误和重复率。关键的是,我们的代理优化攻击成功转移到黑盒受害者系统,突显了严重的实际威胁。

英文摘要:

Retrieval-Augmented Generation (RAG) significantly enhances Large Language Models (LLMs), but simultaneously exposes a critical vulnerability to knowledge poisoning attacks. Existing attack methods like PoisonedRAG remain detectable due to coarse-grained separate-and-concatenate strategies. To bridge this gap, we propose RefineRAG, a novel framework that treats poisoning as a holistic word-level refinement problem. It operates in two stages: Macro Generation produces toxic seeds guaranteed to induce target answers, while Micro Refinement employs a retriever-in-the-loop optimization to maximize retrieval priority without compromising naturalness. Evaluations on NQ and MSMARCO demonstrate that RefineRAG achieves state-of-the-art effectiveness, securing a 90% Attack Success Rate on NQ, while registering the lowest grammar errors and repetition rates among all baselines. Crucially, our proxy-optimized attacks successfully transfer to black-box victim systems, highlighting a severe practical threat.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑