StealthDiffusion: Towards Evading Diffusion Forensic Detection through Diffusion Model

TL;DR

StealthDiffusion introduces a novel diffusion-based method to create high-quality, imperceptible adversarial images that evade forensic detection and human inspection, highlighting vulnerabilities in current AI-generated content detection.

Contribution

The paper presents StealthDiffusion, a diffusion model-based framework that generates imperceptible adversarial images by manipulating latent space and spectral properties, improving evasion of forensic detectors.

Findings

Effective in both white-box and black-box scenarios

Produces images with spectral properties similar to genuine images

Successfully evades state-of-the-art forensic classifiers

Abstract

The rapid progress in generative models has given rise to the critical task of AI-Generated Content Stealth (AIGC-S), which aims to create AI-generated images that can evade both forensic detectors and human inspection. This task is crucial for understanding the vulnerabilities of existing detection methods and developing more robust techniques. However, current adversarial attacks often introduce visible noise, have poor transferability, and fail to address spectral differences between AI-generated and genuine images. To address this, we propose StealthDiffusion, a framework based on stable diffusion that modifies AI-generated images into high-quality, imperceptible adversarial examples capable of evading state-of-the-art forensic detectors. StealthDiffusion comprises two main components: Latent Adversarial Optimization, which generates adversarial perturbations in the latent space of stable diffusion, and Control-VAE, a module that reduces spectral differences between the generated adversarial images and genuine images without affecting the original diffusion model's generation process. Extensive experiments show that StealthDiffusion is effective in both white-box and black-box settings, transforming AI-generated images into high-quality adversarial forgeries with frequency spectra similar to genuine images. These forgeries are classified as genuine by advanced forensic classifiers and are difficult for humans to distinguish.