In a move to combat misinformation and promote transparency, YouTube is implementing new disclosure requirements and content labels for videos that have been altered or created synthetically using Generative AI (AI that creates new content).
This initiative reflects YouTube’s commitment to fostering responsible AI innovation. While acknowledging the potential of AI to unlock creative possibilities, YouTube recognizes the importance of balancing these opportunities with protecting its community.
Following this approach, all content uploaded to YouTube remains subject to its existing Community Guidelines, regardless of its creation method.
Building on these efforts, YouTube recently introduced a new feature in November 2023. This feature requires creators to disclose the use of realistic, altered, or synthetic content during the upload process. For example, creators would need to disclose if they used AI to:
| Doesn’t require disclosure by creators | Does require disclosure by creators |
| Applying beauty filters | Digitally generating or altering content to replace the face of one individual with another’s |
| Synthetically generating or extending a backdrop to simulate a moving car | Digitally altering a famous car chase scene to include a celebrity who wasn’t in the original movie |
| Using effects to enhance previously recorded audio | Simulating audio to make it sound as if a medical professional gave advice when the professional did not actually give that advice |
| Using an AI-generated animation of a missile in a video | Showing a realistic depiction of a missile fired toward a real city |
Helping Users Identify Altered Content
To further empower viewers, YouTube is implementing two methods for identifying potentially misleading content:
- Disclosure Labels: Videos containing undisclosed, realistic alterations will be flagged with a label on both the description panel and the video player itself.
- Content Removal: In the coming months, YouTube will allow users to request the removal of AI-generated content that impersonates an identifiable person through their privacy request process.
This new system empowers creators to leverage the power of AI responsibly while ensuring viewers are informed about the content they consume.
