Stay-Positive: A Case for Ignoring Real Image Features in Fake Image Detection

Detecting AI-generated images is a challenging yet essential task. A primary difficulty arises from the detector’s tendency to rely on spurious patterns, such as compression artifacts, which can influence its decisions. These issues often stem from specific patterns that the detector associates with the real data distribution, making it difficult to isolate the actual generative traces. We argue that an image should be classified as fake if and only if it contains artifacts introduced by the generative model. Based on this premise, we propose Stay-Positive, an algorithm designed to constrain the detector’s focus to generative artifacts while disregarding those associated with real data. Experimental results demonstrate that detectors trained with Stay-Positive exhibit reduced susceptibility to spurious correlations, leading to improved generalization and robustness to post-processing. Additionally, unlike detectors that associate artifacts with real images, those that focus purely on fake artifacts are better at detecting inpainted real images.

Detecting AI-generated images can be tricky because systems often rely on patterns that look like they belong to real images, but these patterns can be misleading. This causes the system to make mistakes, especially when dealing with edited or new images. We believe an image should only be considered real only if it doesn't show signs of being made by an AI generator. Our method, Stay-Positive, helps these systems focus only on the fake signs and ignore the misleading ones, making them better at spotting fake images, even if they've been changed or edited.