tech
YouTube will now auto-label AI-generated videos
YouTube will automatically detect and label videos with significant photorealistic AI content, moving beyond its voluntary creator disclosure system. Labels are moving to a more prominent position and will be permanent for content made with YouTube’s own tools or verified by C2PA metadata.
TL;DR
- YouTube will automatically detect and label photorealistic AI-generated videos, starting May 2026.
- Labels will be moved to more prominent positions: below the player for long-form, on-screen for Shorts.
- All AI-labelled videos will now have visible markers, regardless of topic, unlike the previous distinction for sensitive subjects.
- Labels will be permanent if content is made with YouTube's AI tools (Veo, Gemini Omni, Dream Screen) or has C2PA metadata indicating full AI generation.
- YouTube's system will leverage C2PA metadata and Google's SynthID watermarks for detection.
- Deepfake protections are extended to all adults, allowing anyone to request removal of AI content depicting them.
- Labels are informational and will not affect recommendation algorithms or monetization.
- The changes position YouTube ahead of the EU's AI Act transparency requirements.
- The shift acknowledges that voluntary creator disclosure has been unreliable.
- YouTube is also enhancing AI features for content creation, creating a tension between easier creation and harder concealment of AI content.