ImgTrojan: Jailbreaking Vision-Language Models with ONE Image

arXiv:2403.02910v1 Announce Type: new
Abstract: There has been an increasing interest in the alignment of large language models (LLMs) with human values. However, the safety issues of their integration with a vision module, or vision language models (VLMs), remain relatively underexplored. In this paper, we propose a novel jailbreaking attack against VLMs, aiming to bypass their safety barrier when a user inputs harmful instructions. A scenario where our poisoned (image, text) data pairs are included in the training data is …

abstract alignment arxiv cs.ai cs.cv human image integration jailbreaking language language models large language large language models llms novel paper safety type values vision vision-language models vlms