YouTube to Automatically Label Photorealistic AI Videos

YouTube will no longer rely solely on uploaders to disclose when they use AI tools to create a video. [1][2] Starting in May, the company will use new internal signals to automatically flag content that shows significant photorealistic AI use. [1][2] YouTube is also making these AI labels more prominent across both long-form videos and Shorts. [1] Previously, voluntary AI labels were only visible in an expanded video description section titled "How this content was made." [2]

Labels will be permanently attached to videos if they contain C2PA metadata indicating a purely AI source or if they were created using YouTube's own AI tools, such as Veo or Dream Screen. [1][2] Creators will be able to appeal or update the disclosure status if their content is misidentified, but they cannot remove the labels triggered by C2PA metadata or Google's tools. [1][2] This automated detection functionality follows Google's recent release of Gemini Omni, a family of multimodal models capable of outputting high-quality videos. [1]