G2VD: Generalizable AI-Generated Video Detection via Counterfactual Intervention and Causal Disentanglement
2026-07-06 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionArtificial Intelligence
AI summaryⓘ
The authors address the problem that existing AI-generated video detectors often fail when tested on videos from new, unseen generators because they focus on surface details tied to specific generators. They propose a method called G2VD that creates special 'what-if' videos to help the detector learn the true common signs of fakeness, ignoring generator-specific differences. Their system also uses a clever classification design to separate important forgery clues from domain biases. Experiments show their method works well across different datasets, even with limited training data.
AI-generated videoscross-domain generalizationshortcut learningcounterfactual interventioncausal disentanglementvariational autoencodersfrequency-domain alignmentclassificationAUCGenVidBench
Authors
Meng Du, Hongchang Chen, Ran Li, Junjie Zhang, Qi Ouyang, Shuxin Liu
Abstract
The rapid advancement of AI-generated videos poses increasing security risks and calls for robust detectors with strong cross-domain generalization. Although existing methods achieve promising results under in-domain evaluation, their performance often degrades substantially when tested on unseen generators. A key reason is shortcut learning, where detectors rely on domain-specific spurious cues, such as generator-dependent fingerprints and generation styles, instead of intrinsic forgery traces. To address this issue, we propose G2VD, a Generalizable AI-Generated Video Detection framework based on counterfactual intervention and causal disentanglement. First, G2VD introduces a counterfactual intervention pipeline (CFIPipeline) that generates controlled counterfactual samples via variational autoencoders (VAEs), followed by frequency-domain and pixel-domain alignment, thereby encouraging the detector to focus on generator-intrinsic cues. Building on this intervention process, we further design a causal disentanglement classifier consisting of two domain-anchored branches with distinct classification objectives, combined with an HSIC-based independence constraint to encourage the separation of task-relevant cues from domain-specific bias. Across four public datasets, G2VD shows strong average cross-domain performance and consistent gains over matched backbones. On the challenging GenVidBench cross-domain setting, it exceeds 90% accuracy and reaches an AUC close to 0.95. Notably, this performance is obtained using only 10% of the original training data. The code is available at https://github.com/dumeng98/G2VD.