Contents I. Introduction
Existing generative models such as Diffusion models [1]–[3] have excelled in generating high-quality visual contents, driving advancements in video generation [4], [5] that produce highly realistic videos capable of deceiving the human eye. Unfortunately, these advancements have raised significant security concerns, such as privacy violations and the erosion of trust on social media [6], [7], underscoring the urgent need for effective and reliable detection of AI-generated videos.
