RA-Det: Towards Universal Detection of AI-Generated Images via Robustness Asymmetry

TL;DR

This paper introduces a universal detection method for AI-generated images based on the concept of robustness asymmetry, which distinguishes real images from generated ones by their response to controlled perturbations, outperforming existing detectors.

Contribution

The paper proposes a novel behavior-driven detection framework, RA-Det, leveraging robustness asymmetry as a universal cue, with theoretical analysis and extensive empirical validation across diverse models.

Findings

RA-Det improves detection accuracy by 7.81% on average.

Robustness asymmetry is consistent across different generative models.

The method is data- and model-agnostic, with strong transferability.

Abstract

Recent image generators produce photo-realistic content that undermines the reliability of downstream recognition systems. As visual appearance cues become less pronounced, appearance-driven detectors that rely on forensic cues or high-level representations lose stability. This motivates a shift from appearance to behavior, focusing on how images respond to controlled perturbations rather than how they look. In this work, we identify a simple and universal behavioral signal. Natural images preserve stable semantic representations under small, structured perturbations, whereas generated images exhibit markedly larger feature drift. We refer to this phenomenon as robustness asymmetry and provide a theoretical analysis that establishes a lower bound connecting this asymmetry to memorization tendencies in generative models, explaining its prevalence across architectures. Building on this insight, we introduce Robustness Asymmetry Detection (RA-Det), a behavior-driven detection framework that converts robustness asymmetry into a reliable decision signal. Evaluated across 14 diverse generative models and against more than 10 strong detectors, RA-Det achieves superior performance, improving the average performance by 7.81 percent. The method is data- and model-agnostic, requires no generator fingerprints, and transfers across unseen generators. Together, these results indicate that robustness asymmetry is a stable, general cue for synthetic-image detection and that carefully designed probing can turn this cue into a practical, universal detector. The source code is publicly available at Github.