Structure Disruption: Subverting Malicious Diffusion-Based Inpainting via Self-Attention Query Perturbation

TL;DR

This paper introduces Structure Disruption Attack (SDA), a novel method that disrupts diffusion model inpainting by perturbing self-attention queries, effectively preventing the generation of coherent, misleading images.

Contribution

The paper proposes SDA, a new targeted perturbation technique that disrupts self-attention in diffusion models to protect sensitive image regions from malicious editing.

Findings

SDA achieves state-of-the-art protection performance.

SDA effectively disrupts the contour generation process.

SDA maintains robustness against various attacks.

Abstract

The rapid advancement of diffusion models has enhanced their image inpainting and editing capabilities but also introduced significant societal risks. Adversaries can exploit user images from social media to generate misleading or harmful content. While adversarial perturbations can disrupt inpainting, global perturbation-based methods fail in mask-guided editing tasks due to spatial constraints. To address these challenges, we propose Structure Disruption Attack (SDA), a powerful protection framework for safeguarding sensitive image regions against inpainting-based editing. Building upon the contour-focused nature of self-attention mechanisms of diffusion models, SDA optimizes perturbations by disrupting queries in self-attention during the initial denoising step to destroy the contour generation process. This targeted interference directly disrupts the structural generation capability of diffusion models, effectively preventing them from producing coherent images. We validate our motivation through visualization techniques and extensive experiments on public datasets, demonstrating that SDA achieves state-of-the-art (SOTA) protection performance while maintaining strong robustness.