VidForensics-M1: Meta-Detection Reinforcement Learning with Verifiable Temporal Grounding for AI-Generated Video Forensics
2026-08-11 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors address the challenge of detecting AI-generated fake videos, which are becoming more realistic and harder to spot. Instead of just training models with simple yes/no labels, they introduce a method that also considers evidence about where in the video the fake parts are, using a technique called meta-detection. They create fake videos by swapping specific time segments and then guide the model to pay attention to these changes, which helps it learn better and generalize to new types of fakes. Their approach improves detection by focusing on clear, verifiable clues rather than vague descriptions.
AI-generated videosVideo forgery detectionMeta-detectionReinforcement learningTemporal groundingEvidence-guided rewardVideo generation modelsForgery localizationBoundary-frame conditioningLabel-level optimization
Authors
Bowei Liu, Zheng Lu, Yuhan Bian, Xinchen Zhang, Xingming Shui, Yuesheng Huang, Xuhuan Li, Zihao Liu, Yifan Yang, Jun Zhou, Xiu Li
Abstract
Recent advances in video generation models have significantly improved the realism of synthetic videos, blurring the boundary between generated and authentic content and raising concerns about misinformation. Existing MLLM-based detectors mainly rely on supervised fine-tuning or label-level reinforcement learning, where coarse supervision limits generalization to unseen scenarios and emerging video generators. To overcome these limitations, we are the first to introduce \textbf{meta-detection} into AI-generated video detection, enabling reliable forgery detection by jointly optimizing predicted labels and supporting evidence within reinforcement learning. This paradigm requires reliable evidence signals and effective mechanisms to integrate them into label-level optimization. Textual rationales provide semantic descriptions of forgery artifacts, but their generation and verification depend on external models, making supervision vulnerable to hallucinations and semantic biases. In contrast, temporal grounding provides more objective and verifiable evidence, as manipulated intervals can be precisely controlled during forgery construction. Based on this insight, we propose an automated data construction pipeline that generates paired real-fake videos by replacing temporal segments with boundary-frame-conditioned video generation models. Furthermore, we introduce \textbf{Evidence-Guided Reward Redistribution}, which performs evidence-aware credit assignment by redistributing rewards among label-correct responses according to evidence quality. This preserves reliable label supervision while encouraging detectors to acquire fine-grained and verifiable forgery localization capabilities. Extensive experiments demonstrate that \textbf{VidForensics-M1} effectively leverages verifiable temporal evidence to achieve robust and generalizable AI-generated video detection.