Home / Technology / YouTube Tackles AI Video Deepfakes
YouTube Tackles AI Video Deepfakes
27 May
Summary
- YouTube is enhancing AI video labeling for better origin verification.
- New internal signals will flag significant photorealistic AI use.
- C2PA metadata and watermarked Google tools trigger permanent labels.
YouTube is enhancing its system for identifying AI-generated videos, a crucial step as AI technology blurs the lines between reality and artificial creation. Previously, the platform relied heavily on creators to disclose their use of AI tools, a method found to be insufficient.
This month marks the introduction of a more robust labeling system. YouTube will now utilize "new internal signals" to detect and flag videos that demonstrate substantial photorealistic AI implementation. This automated approach aims to provide more reliable verification of video origins.
The platform has identified specific triggers for these labels. Content exhibiting C2PA metadata that indicates a purely AI source, or videos created using watermarked Google tools like Veo, will receive permanent AI designation. While creators can appeal incorrect AI labeling, appeals will not be permitted for these specific, verifiable triggers.
