attack arXiv Apr 8, 2026 · 6w ago
Ziye Wang, Guanyu Wang, Kailong Wang · Huazhong University of Science and Technology · Beihang University
Word-level poisoning attack on RAG systems that injects stealthy toxic documents to manipulate LLM outputs via retriever optimization
Data Poisoning Attack Prompt Injection Training Data Poisoning nlp
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.
llm transformer Huazhong University of Science and Technology · Beihang University
ML Security Papers