Tencent (Yipeng Du, Yuxi Li, Ying Zhang, Chong Sun, Chen Li, Jing Lyu, Ying Tai) · Agentic Video Forensics

WeVisForensics: Self-Evolving Agent for Omni AI-Generated Video Detection

WeVisForensics is an agentic framework for detecting AI-generated video, in which a main investigator agent coordinates a self-evolving library of visual and audio forensic inspector sub-agents. It is trained with DTPO, a dual-timescale policy optimisation that combines fast reinforcement learning of the investigator with slower refinement of the inspectors. The authors also release ForenCue-30K, a dataset of forensic annotations and interaction trajectories. They report 95.2% average detection accuracy across in-domain, out-of-domain and agentic benchmarks, 9.0% above the strongest prior detector.

Open the source ↗

Similar Research & Safety releases

Published by Quanta AI Labs, an AI agency in India. Browse every week in AI News Weekly.