English

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

Cryptography and Security 2026-04-10 v1

Abstract

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.

Keywords

Cite

@article{arxiv.2604.07403,
  title  = {RefineRAG: Word-Level Poisoning Attacks via Retriever-Guided Text Refinement},
  author = {Ziye Wang and Guanyu Wang and Kailong Wang},
  journal= {arXiv preprint arXiv:2604.07403},
  year   = {2026}
}
R2 v1 2026-07-01T11:59:49.566Z