Benchmark RadarRA-Bench
RA-Bench evaluates AI-generated video detection using real crisis-event videos as anchors. It includes 17,886 videos: 1,830 real anchors across 10 social-risk categories and 16,056 generated clips from nine generators. Detection is scored per source via AUC and TPR@5%FPR, with separate tracks for human-deceptive and dissemination-processed videos.
- Released
- 2026-08-14
- Readiness
- Runnable
- Primary field
- General AI
Why it matters
Current detectors lack evidence on realistic AI-generated crisis videos. RA-Bench provides a matched real-generated evaluation to assess detector generalization, human perception, and robustness to social dissemination, supporting practical decisions on misinformation defense.
Motivation
Recent video generators can fabricate realistic depictions of wars, disasters, public emergencies, and other real-world crises, creating substantial risks of misinformation.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.