Beyond Text: Unveiling Privacy Vulnerabilities in Multi-modal Retrieval-Augmented Generation
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912383053594624 |
|---|---|
| author | Zhang, Jiankun Zeng, Shenglai Ren, Jie Zheng, Tianqi Liu, Hui Tang, Xianfeng Liu, Hui Chang, Yi |
| author_facet | Zhang, Jiankun Zeng, Shenglai Ren, Jie Zheng, Tianqi Liu, Hui Tang, Xianfeng Liu, Hui Chang, Yi |
| contents | Multimodal Retrieval-Augmented Generation (MRAG) systems enhance LMMs by integrating external multimodal databases, but introduce unexplored privacy vulnerabilities. While text-based RAG privacy risks have been studied, multimodal data presents unique challenges. We provide the first systematic analysis of MRAG privacy vulnerabilities across vision-language and speech-language modalities. Using a novel compositional structured prompt attack in a black-box setting, we demonstrate how attackers can extract private information by manipulating queries. Our experiments reveal that LMMs can both directly generate outputs resembling retrieved content and produce descriptions that indirectly expose sensitive information, highlighting the urgent need for robust privacy-preserving MRAG techniques. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_13957 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Beyond Text: Unveiling Privacy Vulnerabilities in Multi-modal Retrieval-Augmented Generation Zhang, Jiankun Zeng, Shenglai Ren, Jie Zheng, Tianqi Liu, Hui Tang, Xianfeng Liu, Hui Chang, Yi Cryptography and Security Computation and Language Multimodal Retrieval-Augmented Generation (MRAG) systems enhance LMMs by integrating external multimodal databases, but introduce unexplored privacy vulnerabilities. While text-based RAG privacy risks have been studied, multimodal data presents unique challenges. We provide the first systematic analysis of MRAG privacy vulnerabilities across vision-language and speech-language modalities. Using a novel compositional structured prompt attack in a black-box setting, we demonstrate how attackers can extract private information by manipulating queries. Our experiments reveal that LMMs can both directly generate outputs resembling retrieved content and produce descriptions that indirectly expose sensitive information, highlighting the urgent need for robust privacy-preserving MRAG techniques. |
| title | Beyond Text: Unveiling Privacy Vulnerabilities in Multi-modal Retrieval-Augmented Generation |
| topic | Cryptography and Security Computation and Language |
| url | https://arxiv.org/abs/2505.13957 |