YouTube to auto-label AI-generated videos
▼ Summary
– YouTube will now automatically label videos it detects as using significant photorealistic AI, rather than relying solely on creator disclosures.
– AI labels will be made more prominent, appearing directly below the video player for long-form videos and overlaid on YouTube Shorts.
– Creators can update misidentified content but cannot remove labels if the video was made with YouTube’s own AI tools like Veo or Dream Screen.
– The automatic detection follows YouTube’s expansion of AI deepfake detection, allowing any adult to scan for face matches.
– AI labels do not affect a video’s recommendations or ability to monetize.
As AI-generated video tools grow more sophisticated, YouTube is stepping up its oversight. No longer will the platform rely solely on creators to flag their own AI content. Starting now, the company will automatically apply labels when its internal systems detect the use of significant photorealistic AI.
These labels are also getting a major visibility upgrade, making them easier to spot across both long-form videos and YouTube Shorts.
The policy around AI labeling on YouTube isn’t new. It has been in place for over two years, following an update to the platform’s AI guidelines. Back then, YouTube introduced a tool in Creator Studio requiring creators to disclose if their videos contained AI content that could be mistaken for a real person, place, or event. Videos featuring obviously fantastical or animated scenes, such as a unicorn in a magical forest, were exempt.
What has changed is YouTube’s enforcement approach. The company says it will now take a more active role in policing AI content. This shift comes on the heels of Google’s release of Gemini Omni at its Google I/O developer conference last week. That new family of multimodal AI models can produce high-quality videos that demonstrate an understanding of physics, culture, history, and science.
Beginning in May, YouTube will use new internal signals to identify and label AI-generated content. Creators are still expected to disclose their AI use voluntarily, but if they fail to do so, YouTube will step in and apply the label automatically.
If a creator believes their content was misidentified, they can update the disclosure status. However, there is a catch: if the content was created using YouTube’s own AI tools, such as Veo or Dream Screen, the label cannot be removed. Similarly, videos containing C2PA metadata indicating full AI generation will have labels permanently attached. OpenAI recently committed to the C2PA standard, joining Nvidia, Kakao, and Eleven Labs.
This automatic detection feature follows the recent expansion of YouTube’s AI deepfake detection, which now allows any adult to scan the platform for face matches. Earlier tests focused on celebrities, public figures, politicians, and other creators.
YouTube is also making its AI labels more consistent and prominent. Previously, labels appeared in the expanded description unless the video dealt with sensitive topics like health or news; in those cases, a more prominent label appeared directly on the video itself.
Now, the labels will sit directly below the video player, above the description for long-form content, and directly on YouTube Shorts. The company believes this repositioning will make the labels far more noticeable to viewers encountering photorealistic, AI-altered, or AI-generated content.
For videos that are only slightly altered, animated, or clearly unrealistic, such as that prancing unicorn, the label will remain in the expanded description only.
Importantly, YouTube confirms that AI labels will not affect a video’s recommendations or its ability to monetize.
Beyond policing AI content, YouTube continues to invest heavily in AI for its own features. These include the interactive search tool Ask YouTube, a playlist generator for YouTube Music, AI video summaries, and other generative AI creation tools.
(Source: TechCrunch)
