AI RESEARCH
The Regularizing Power of Language-Training Deepfake Detectors
arXiv CS.CV
•
ArXi:2605.31192v1 Announce Type: new Recently, thanks to the advent of Multimodal-LLMs, deepfake detectors are striving not only to be generalizable but also interpretable. We propose that these two challenges can effectively be tackled jointly, since describable artifacts typically generalize better, opening the possibility to use language as a regularization mechanism.