A Bias-Free Training Paradigm for More General AI-generated Image Detection

Self-conditioned Content Aug. Real … Conditioning Denoising Generation Denoising Masks "cat" "cat" "dog" Random Noise Figure 1. We introduce a new training paradigm for AI-generated image detection. To avoid possible biases, we generate synthetic images from self-conditioned reconstructions of real images and include augmentation in the form of inpainted versions. This allows to avoid semantic biases. As a consequence, we obtain better generalization to unseen models and better calibration than SoTA methods.