Who Made This? Fake Detection and Source Attribution with Diffusion Features

TL;DR

FRIDA is a lightweight, data-efficient framework that uses diffusion features from a pre-trained model to detect and attribute AI-generated images, achieving state-of-the-art results across unseen generators.

Contribution

Introduces FRIDA, a novel diffusion feature-based method for robust fake image detection and source attribution, capable of generalizing to unseen generators with limited data.

Findings

State-of-the-art cross-generator detection accuracy

Effective source model attribution with a neural classifier

Robust detection with minimal training data

Abstract

The rapid rise of generative models has yielded synthetic images of striking realism, blurring the line between real and fake content. As novel models proliferate, detectors must go beyond mere fake identification to robustly generalise across unseen generators and synthetic content. We introduce FRIDA (Fake image Recognition and source Identification via Diffusion features Analysis), a lightweight, data-efficient framework that uses features from a pre-trained Stable Diffusion Model to detect and attribute AI-generated images. Through an in-depth analysis of how data from different generators are encoded across diffusion U-Net layers, we propose a method that (i) detects synthetic images using a training-free $k$-Nearest Neighbour approach and (ii) performs source model attribution via a compact neural classifier. On the GenImage benchmark, FRIDA achieves state-of-the-art cross-generator detection with limited data while maintaining robust source model attribution capabilities. These results establish diffusion features as a reliable framework for AI-generated image forensics.