Do I label an AI cartoon or fantasy clip on YouTube?
YouTube says non-realistic AI content needs no disclosure: a unicorn ride, a green-screen spacewalk, an animated missile. The limits to watch.

Usually no. YouTube's GenAI disclosure page says creators do not need to disclose non-realistic content made with AI, and it lists three examples: someone riding a unicorn through a fantastical world, green screen used to depict someone floating in space, and an AI-generated or altered animation of a missile in a fully animated video. The test is realism, not whether a model made the pixels.
There are two traps. Realistic scenes inside a cartoon frame still count, and AI music that is the main focus of the video is on the disclose list regardless of the visuals. A stylized clip that a viewer could mistake for real footage is a realistic clip.
Realistic or not: the page's own contrast
The same page pairs each exempt example with a disclose example. Keep the pairs side by side when you review a generated clip.
| Needs no disclosure | Needs disclosure |
|---|---|
| Someone riding a unicorn through a fantastical world | Realistic video of a tornado moving toward a real city |
| Green screen showing someone floating in space | Realistic footage of a match between two real professional tennis players |
| AI animation of a missile in a fully animated video | Making a real person look arrested or imprisoned |
| Beauty filters, color adjustment, background blur | AI music as the main focus of the video |
What YouTube may still label
Answering No does not stop YouTube from labeling. The page says YouTube may apply a label automatically to content made with its own GenAI tools, content that carries C2PA metadata, and content its systems detect as AI generated or altered. If you choose Yes, a label in the player may appear for photorealistic content, and a label in the expanded description may appear for non-photorealistic or animated content.
Make a clearly stylized clip with Sume
If you want to stay on the non-realistic side, steer the generation toward obvious illustration instead of photography. With Sume video generation the prompt is yours; the API does not add a realism switch, so the check is on you after the job returns. Pull still frames with video inspect (it returns stills as image artifacts) and look at them cold: would a stranger call this footage?
Steps
- Generate the clip and sample stills with video inspect.
- Ask of each still: could a viewer mistake this for something that happened?
- If it is obviously a cartoon, answer No under 'AI use' in YouTube Studio.
- If a realistic shot slipped in, or you used generated music as the focus, answer Yes.
- Re-check the expanded description after publishing to see whether YouTube added its own label.
Caveat
This is YouTube's reading as published on its help page today, and the page calls its own list non-exhaustive. If your clip includes a real person's face or voice, move to the realistic column first.
Sources
Related posts
More in Use cases
- Does dimming or cropping a clip need a YouTube AI label?
YouTube lists dimming as a colour adjustment needing no label; cropping is not listed either way. The label turns on realistic synthetic content.
- DOOH 1080x1920 10 s master: a render checklist
An example vertical DOOH master (1080x1920, H.264 MP4, 10 s, 25 to 30 fps). What a Sume video request controls and what you must check yourself.
- Dribbble shot image: 1600 px wide, 10 MB, PNG JPG or GIF
Dribbble recommends images at least 1600 px wide and takes PNG, JPG or GIF up to 10MB. Sizes Sume can generate at 1600 wide, and a script to check the file.
- Dribbble shot video: 20 MB limit and the bitrate budget
Dribbble allows MOV or MP4 video up to 20MB and recommends 1600 px width. What 20MB means in average Mbps by clip length, and which Sume Timeline fields help.
Written by Sume