safety security
The Videos That Fooled People Also Fooled the Detectors
A 17,886-video crisis benchmark found no detector family that generalized consistently across generators and social dissemination.
Summary
A 17,886-video crisis benchmark found no detector family that generalized consistently across generators and social dissemination.
RA-Bench anchors 16,056 generated clips to 1,830 real videos across ten crisis-risk categories, then tests traditional detectors, zero-shot multimodal models and fine-tuned systems. None of the three detector families generalized consistently. The clips that misled human viewers were also hard for automated detectors, and dissemination through social platforms made detection harder, underscoring that a single detector score is not a durable authenticity guarantee.
Why it matters
A 17,886-video crisis benchmark found no detector family that generalized consistently across generators and social dissemination.
Limits and context
- The clips that misled human viewers were also hard for automated detectors, and dissemination through social platforms made detection harder, underscoring that a single detector score is not a durable authenticity guarantee.
Key claims
A 17,886-video crisis benchmark found no detector family that generalized consistently across generators and social dissemination.
Qualification: The clips that misled human viewers were also hard for automated detectors, and dissemination through social platforms made detection harder, underscoring that a single detector score is not a durable authenticity guarantee.
Evidence: source-2026-08-17-005
Sources
- arXiv preprint 2608.14391arXiv · primary research
Corrections
No corrections have been recorded for this story.