YouTube AI disclosure: four triggers and which Sume outputs hit them

YouTube asks you to disclose content that makes a real person appear to say or do something they did not, alters footage of a real event or place, generates a realistic scene that did not occur, or creates music that is the main focus of the video. A Sume text-to-video render of a realistic scene falls under the third trigger, while captions, trims and color changes do not.

The four triggers and the exemptions

YouTube's help page requires disclosure when you use AI to meaningfully alter or generate photorealistic content. The four categories are the ones above. You set it with the AI use setting in YouTube Studio, on a computer or a mobile device, and the content is then labeled as AI generated or altered for viewers.

The page also lists what you do not have to disclose: non-realistic content, such as a unicorn ride through a fantastical world, and minor edits, which it defines as primarily aesthetic changes that do not alter the content in a way that could mislead viewers. Its examples include beauty filters, color and lighting adjustment, background blur, production assistance like scripts and thumbnails, caption creation, video sharpening, upscaling or repair, cloning your own voice for voiceovers or dubs, and AI generating or extending a backdrop to simulate a moving car.

Mapping Sume outputs to the page

The table maps common Sume operations to the categories on YouTube's page. It is a reading of YouTube's examples, not a statement by YouTube about Sume, and edge cases are yours to judge.

The deciding question is the one YouTube uses: is the result realistic, and could a viewer be misled about what actually happened?

Where a mixed project lands

Many Shorts mix both: a real talking head with a generated insert. If the generated insert shows a realistic place or event that did not happen, the page's example of AI-generated extra footage of a real place, like a surfer in Maui for a travel video, applies and the whole video is disclosed. If the insert is a clearly stylized graphic, it is not.

A habit that helps: keep the job id of every generated insert in the project notes, so when you open Studio you know whether any generated realistic footage is in the file.

Make the setting part of the upload step

The AI use setting lives in YouTube Studio, so it is easy to forget when a pipeline uploads in bulk. Whether you upload by hand or from a script, add the check to your publish checklist.

For the nearby cases, read upscaled AI video needs no disclosure and AI music as the main focus.