EA-Swin: An Embedding-Agnostic Swin Transformer for AI-Generated Video Detection

TL;DR

EA-Swin is a novel embedding-agnostic transformer model that effectively detects AI-generated videos across diverse generators, outperforming existing methods and demonstrating strong generalization on a new large-scale benchmark dataset.

Contribution

We introduce EA-Swin, a new spatiotemporal transformer model compatible with generic video embeddings, and create EA-Video, a comprehensive dataset for evaluating AI-generated video detection.

Findings

EA-Swin achieves 0.97-0.99 accuracy across major generators.

Outperforms prior state-of-the-art methods by 5-20%.

Maintains strong generalization to unseen data.

Abstract

Recent advances in foundation video generators such as Sora2, Veo3, and other commercial systems have produced highly realistic synthetic videos, exposing the limitations of existing detection methods that rely on shallow embedding trajectories, image-based adaptation, or computationally heavy MLLMs. We propose EA-Swin, an Embedding-Agnostic Swin Transformer that models spatiotemporal dependencies directly on pretrained video embeddings via a factorized windowed attention design, making it compatible with generic ViT-style patch-based encoders. Moreover, we construct the EA-Video dataset, a benchmark dataset comprising 130K videos that integrates newly collected samples with curated existing datasets, covering diverse commercial and open-source generators and including unseen-generator splits for rigorous cross-distribution evaluation. Extensive experiments show that EA-Swin achieves 0.97-0.99 accuracy across major generators, outperforming prior SoTA methods (typically 0.8-0.9) by a margin of 5-20\%, while maintaining strong generalization to unseen distributions, establishing a scalable and robust solution for modern AI-generated video detection.