YouTube altered or synthetic content disclosure: what counts
YouTube asks for disclosure only when altered or synthetic content looks realistic. Sort your own edit steps into needs it, exempt, and not listed.
What counts is realistic content that is made or changed with AI in one of three ways: it makes a real person appear to say or do something they did not, it alters footage of a real event or place, or it generates a realistic scene that did not occur. Fantastical scenes and routine polish such as color adjustment do not count. That is YouTube's own list, and the table below sorts video-editing steps against it.
The rules are from YouTube Help's disclosure page, read 2026-09-29. The steps are ones you can take with Sume's video generation, video trim and avatar video endpoints. Where I place a step, that is my reading, not YouTube's ruling.
Which of my edit steps need the label?
Read the last column as a first guess; when a step is not listed, YouTube has not said.
| Step | Sume endpoint | Against YouTube's list |
|---|---|---|
| Generate a realistic scene from a prompt | POST /v1/videos | Matches "a realistic scene that didn't occur": disclose |
| Generate a clearly fantastical scene | POST /v1/videos | Unrealistic or fantastical scenarios are not required |
| Talking video of a generated person | POST /v1/avatar-1.0/talking-video | Realistic person made to speak: my reading is disclose |
| Cut a range out of your own footage | POST /v1/video-trim | Not listed; if the cut changes what a real event shows, treat it as altered |
| Sharpen, upscale or repair | video_upscale_create (MCP) | Listed as not required |
What does YouTube say does not need disclosure?
The same page names these as not needed: unrealistic or fantastical scenarios, beauty filters, color adjustment and special effects filters, voice or audio repair, video sharpening, upscaling or repair, and cloning your own voice for voiceovers or dubs.
Where does the label appear?
YouTube shows it in the video player for photorealistic content and in the expanded description otherwise. The page also says YouTube may apply an AI label on its own for content from its AI tools or with C2PA metadata, so a video can be labeled even when you did not tick the box.
What if my step is not on either list?
Ask the three questions from the top of this post about the finished video, not the step. If a viewer could take the result for a real person, event or place, disclose. Sume's docs do not describe any label added to a file, so the choice is yours in the upload flow.
Why does a trim matter for real footage?
Video trim takes a start and one of end or duration and writes a new MP4, leaving the source untouched. Keeping the source gives you the original to compare with if anyone asks whether a cut changed what a real event showed. YouTube's page does not list ordinary cuts, so the record you keep is your own evidence, not a rule.
Sources
Related posts
More in Use cases
- Why isn't YouTube dubbing my video? The reasons on its help page
YouTube skips auto dubbing for videos over 120 minutes, music-only ones, copyrighted content and unclear or fast speech. How to check yours before upload.
- YouTube auto dubbing languages: which get expressive speech
YouTube's February 2026 post says auto dubbing covers 27 languages and Expressive Speech eight. The eight languages and where the help page marks them.
- YouTube auto dubbing turn off: the setting and what to disclose
Turn YouTube automatic dubbing off in Studio settings or review dubs before they publish. What viewers see, and what YouTube's disclosure page says about dubs.
- YouTube dynamic thumbnails: make three for different audiences
YouTube lists dynamic thumbnails among new Studio tools. Make three thumbnail candidates for three audiences from one video frame with the Sume image API.
Written by Sume