AI RESEARCH
Degradation-Consistent Paired Training for Robust AI-Generated Image Detection
arXiv CS.AI
•
ArXi:2604.10102v1 Announce Type: cross AI-generated image detectors suffer significant performance degradation under real-world image corruptions such as JPEG compression, Gaussian blur, and resolution downsampling. We observe that state-of-the-art methods, including B-Free, treat degradation robustness as a byproduct of data augmentation rather than an explicit