Physics based method detects fake AI generated audio video clips
PIVOT: Physics-Grounded Verification for AI-Generated Audio-Video Detection
Computer Vision and Pattern RecognitionArtificial IntelligenceMultimedia
Summary
AI-generated videos and sounds are getting very realistic, making it hard to tell if they're fake by usual methods. The authors created PIVOT, a new way to spot fake clips by checking if the events in them follow real-world physical rules, like how sound and motion should line up. PIVOT not only decides if a clip is real or fake but also shows its reasoning by highlighting which laws were tested and what it found. Tests show PIVOT works better than some existing tools at spotting AI-generated audio-video content.
What this means in practice
- •For content moderation teams: Detect AI-generated fake audio-video clips in uploaded media by verifying physical consistency to improve detection reliability.
- •For forensic analysts: Provide structured physical evidence to verify authenticity of audio-video clips during investigations of manipulated media.
Authors
Bo Zheng, Kangran Zhao, Xiaoyu Zhang, Weinan Guan, Zhiheng Li, Yize Chen, Haizhou Li, Qingshan Liu, Siwei Lyu, Baoyuan Wu
Abstract
As generative models continue to advance, AI-generated content (AIGC) is becoming increasingly realistic, weakening the artifact cues commonly exploited by existing detectors. Nevertheless, faithfully reproducing the physical behavior of real-world events remains challenging for current generators. We therefore explore detecting AIGC by assessing whether the depicted event satisfies measurable constraints derived from physical laws. We introduce PIVOT, a physics-grounded AIGC detector, instantiated here for audio-video clips, that estimates physical quantities from video and audio, selects physical laws relevant to each clip, and verifies their measurable constraints. Beyond a real/fake decision, PIVOT returns supporting evidence that records the verification outcome, relevant time window, and supporting quantities for each applicable law. Although instantiated and evaluated here on audio-video data, the framework can, in principle, extend to other AIGC modalities whenever the physical quantities required for verification can be estimated reliably. We also introduce PhysForensics-Bench, comprising paired real and generated audio-video clips from nine event-centric scene families and two recent audio-video generators. On PhysForensics-Bench, PIVOT achieves 70.30% accuracy and 64.29% F1 score on Real+Seedance, and 72.16% accuracy and 65.82% F1 on Real+VEO. In comparison, direct inspection with Gemini 3.1 Pro obtains 53.96% accuracy and 60.09% F1 on Real+Seedance, and 57.22% accuracy and 63.44% F1 on Real+Veo. These results demonstrate the practical promise of physical-consistency verification as a structured and inspectable source of evidence that complements artifact-based AIGC detection.