On the Multi-modal Vulnerability of Diffusion Models
Keywords: multi-modal alignment, adversarial attack, diffusion model
TL;DR: We observe a misalignment of text-to-image diffusion models between different modalities. Based on this observation, we propose MMP-Attack, a targeted method for diffusion models, manipulating generated content by appending a specific suffix.
Abstract: Diffusion models have been widely deployed in various image generation tasks, demonstrating an extraordinary connection between image and text modalities. Although prior studies have explored the vulnerability of diffusion models from the perspectives of text and image modalities separately, the current research landscape has not yet thoroughly investigated the vulnerabilities that arise from the integration of multiple modalities, specifically through the joint analysis of textual and visual features. In this paper, we first visualize both text and image feature space embedded by diffusion models and observe a significant difference, i.e., the prompts are embedded chaotically in the text feature space, while in the image feature space they are clustered according to their subjects. Based on this observation, we propose MMP-Attack, which leverages multi-modal priors (MMP) to manipulate the generation results of diffusion models by appending a specific suffix to the original prompt. Specifically, our goal is to induce diffusion models to generate a specific object while simultaneously eliminating the original object. Our MMP-Attack shows a notable advantage over existing studies with superior manipulation capability and efficiency. Our code is publicly available at \url{https://github.com/ydc123/MMP-Attack}.
Submission Number: 1
Loading