A Unified Framework for Stealthy Adversarial Generation via Latent Optimization and Transferability Enhancement

TL;DR

This paper introduces a unified framework that enhances the transferability of diffusion-based adversarial examples across various tasks by integrating traditional transferability strategies, demonstrated by winning a major competition.

Contribution

The paper presents a novel framework that combines transferability enhancement techniques with diffusion model-based adversarial generation, broadening their applicability beyond classification tasks.

Findings

Achieved first place in ACM MM25 competition.

Effectively generalizes adversarial examples to multiple downstream tasks.

Validates the approach's effectiveness through competitive success.

Abstract

Due to their powerful image generation capabilities, diffusion-based adversarial example generation methods through image editing are rapidly gaining popularity. However, due to reliance on the discriminative capability of the diffusion model, these diffusion-based methods often struggle to generalize beyond conventional image classification tasks, such as in Deepfake detection. Moreover, traditional strategies for enhancing adversarial example transferability are challenging to adapt to these methods. To address these challenges, we propose a unified framework that seamlessly incorporates traditional transferability enhancement strategies into diffusion model-based adversarial example generation via image editing, enabling their application across a wider range of downstream tasks. Our method won first place in the "1st Adversarial Attacks on Deepfake Detectors: A Challenge in the Era of AI-Generated Media" competition at ACM MM25, which validates the effectiveness of our approach.