Vis-Poison: Poisoning Visual Knowledge in Multimodal Retrieval-Augmented Generation

Engineering notes

Engineering notes will be added by the aipentium editorial team.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为大语言模型、生成式AI、ChatGPT相关技术、计算机视觉、深度学习等高价值论文补充中文说明。

Original abstract

While multimodal retrieval-augmented generation (RAG) systems increasingly rely on images as external knowledge sources, the introduction of poisoned visual evidence can severely compromise multimodal large language model (MLLM) generation. Unlike prior attacks that rely on altering textual metadata, we introduce Vis-Poison, a novel visual knowledge poisoning attack where the poisoned image itself is the attacker-controlled payload, without manipulating captions, summaries, metadata, or other associated text. Specifically, this attack is instantiated through an automated multi-agent method that constructs visually plausible poisoned images. To assess its impact, we evaluate Vis-Poison across two representative multimodal RAG pipelines, four embedding models, and six generation models. Empirically, Vis-Poison achieves an end-to-end attack success rate of 40.16\% to 65.40\% against 30k-entry multimodal knowledge bases in \emph{black-box} settings. Moreover, Vis-Poison remains effective against various MLLMs that can answer correctly from parametric knowledge alone, with an average success rate above 60\%. Code and data are available at https://github.com/SWUFE-DB-Group/Vis-Poison.

Links and sources

• Official / arXiv page
• PDF from original source

Comments