YouTube to spot and flag AI-generated videos

YouTube announced on Wednesday that it will now automatically detect and label AI-generated videos on its platform, taking a more active role in identifying content that uses significant photorealistic artificial intelligence. The move comes as AI video generation technology has become increasingly sophisticated, following Google's recent unveiling of its Gemini Omni multimodal AI models at the Google I/O developer conference. Starting in May, YouTube will deploy new internal systems to identify AI content when creators fail to disclose its use, addressing concerns about deepfakes and manipulated media. The platform is also relocating its AI labels to make them far more visible to viewers. Previously, these disclosures were hidden in the expanded description under the "How this content was made" section, requiring users to actively seek out the information. Now, for long-form videos, the AI label will appear directly below the video player above the description, while YouTube Shorts will display the label as an overlay on the video itself. Content that is animated, unrealistic, or only slightly altered, such as fantasy scenes, will continue to have labels in the expanded description only. YouTube emphasized that the new labeling system will not affect how videos are recommended or monetized. Creators who believe their content was incorrectly flagged can update the disclosure status in YouTube Studio. However, the company stated that labels will be permanently attached to videos created using YouTube's own AI tools like Veo or Dream Screen, or when content contains C2PA metadata indicating it was fully AI-generated. The C2PA standard, which OpenAI recently committed to alongside Nvidia, Kakao, and Eleven Labs, provides a technical framework for certifying AI-generated content.