Does a scripted AI presenter video need YouTube's synthetic label?

Short answer

It depends on whom the presenter looks like and what the video shows. YouTube's help page asks creators to disclose realistic altered or synthetic content, and its first example is content that makes a real person appear to say or do something they did not do. A presenter invented from a text prompt is not a real person, but a photorealistic one can still read as a generated scene that did not occur. We cannot give you a legal or policy ruling; this post sets out the page's own categories and how to test your video against them.

The four triggers on the YouTube page

The page lists what to disclose when content is realistic. Compare each against your video before you upload.

What the page says you can skip

The page lists production help as exempt: script generation, thumbnail creation, caption generation, and video enhancement such as upscaling. It also lists fantasy content and cosmetic edits. In a Sume workflow, writing the script with a model and adding captions with the captions feature fall in those exempt groups. The avatar render itself is the part to assess.

• Exempt on the page: scripts, thumbnails, captions, sharpening, upscaling.
• Fantasy or clearly unrealistic content needs no label.
• The avatar render is the step to judge.

How Sume inputs map to the question

Sume creates avatars from a prompt, structured profile traits, or a reference image. A prompt or profile avatar has no real-person source. An image avatar built from a photo of a real person is different: if that person did not say the script, the first YouTube trigger is in play. For the scene, a prompt scene is generated, while a photo scene uses your own image.

What happens if you choose no

The page warns that creators who consistently choose not to disclose may get a manual label or penalties, including content removal or suspension from the YouTube Partner Program. It also says that disclosing does not limit audience reach or affect monetization eligibility. When a case is unclear, the page's own wording makes disclosure the lower-risk choice.

A practical test before you upload

Ask three questions of the finished file. Would a viewer reasonably take the presenter for a real, identifiable person? Does the script put words in that person's mouth that they never said? Does the background pass as real footage of a real place? A yes to any of them is a reason to disclose. If all three are no, for example a clearly stylized presenter introducing your own product, the page's fantasy and unrealistic-content exemption may apply, though the safest habit remains to label.

Keep a short record next to the file: the avatar input type, the script text, who approved the first frame, and the label decision. If you ever have to explain a video, that record answers the question in a minute.