YouTube implements AI disclosure labels for photorealistic content

YouTube has implemented a system for labeling photorealistic or meaningfully AI-altered content. Since May 2026, the platform has used automated systems to detect significant AI use and apply disclosure labels even when creators have not manually flagged the content.

The labels are categorized into two distinct types. Permanent labels are applied to content where the origin is verified through provenance, such as content created using YouTube’s own tools, Veo and Dream Screen, or files carrying C2PA metadata. These labels cannot be removed by creators.

In contrast, labels based on automated inference regarding photorealistic AI are considered soft labels. If a creator believes an automated signal has incorrectly identified their content, they can update the disclosure status within YouTube Studio. For long-form videos, the label is positioned below the player and above the description, while for Shorts, it overlays the video itself.