Divide and Conquer: Reliable Multi-View Evidential Learning for Deepfake Detection

With the evolution of generative models, deepfakes have achieved near-perfect semantic realism, leaving forensic traces only in subtle structural anomalies. However, existing single-view paradigms often fail to generalize, as dominant semantic features overwhelm subtle artifact cues within entangled representations. This imbalance leads to overconfident yet brittle predictions—a phenomenon we term the Semantic Masking Effect. To address this challenge, we propose a reliable framework called Divide-and-Conquer Multi-View Evidential Learning (DiCoME) for Deepfake Detection. In the "Divide'' phase, we employ Geometric View Purification to decompose the entangled representation space through principled geometric projection. This process suppresses semantic interference within artifact-sensitive representations, forming the foundation for decorrelated yet complementary semantic and artifact views. In the "Conquer'' phase, we leverage Uncertainty-Aware Evidential Learning to synthesize these distinct views. By explicitly modeling the "epistemic conflict'' between semantic and artifact cues, this mechanism provides calibrated uncertainty estimates instead of forcing rigid deterministic decisions. Extensive experiments across multiple benchmarks demonstrate that our method consistently outperforms existing approaches in generalization performance, while providing reliable uncertainty estimation for trustworthy deepfake detection. Code is available at https://github.com/kxl0825/DiCoME.git.

AI-generated fake images and videos are becoming so realistic that it is increasingly difficult to tell them apart from authentic content. Existing detectors often focus on what an image depicts, such as a person’s face or expression, rather than on the tiny inconsistencies that reveal manipulation. As a result, they can appear confident but fail when faced with new types of fakes. We introduce DiCoME, a method that examines an image from two complementary perspectives: what the image shows and whether it contains subtle signs of tampering. Our method first separates these signals so that the visible content does not hide the manipulation clues. It then combines them while estimating how trustworthy its own decision is, especially when the two sources of evidence disagree. Across several test datasets, DiCoME detects previously unseen deepfakes more accurately than existing methods and provides more reliable warnings when its predictions are uncertain. This could support safer use of deepfake detectors in situations where incorrect decisions may have serious consequences.