YouTube AI disclosure: player label or description, by clip type
YouTube shows a prominent label for photorealistic AI and description-only for unrealistic or animated content. Sort your Sume clips into the right bucket.
YouTube says photorealistic content gets a label in the video player, while non-photorealistic content gets a label in the expanded description. Its May 27, 2026 blog adds that on Shorts the label is an overlay on the video, and on long-form it sits below the player. Sort each clip into a bucket before you upload.
What are the two buckets? (read 2026-10-09)
| Content | Where the label shows |
|---|---|
| Photorealistic, meaningfully altered or generated | Prominent label: in the player; on Shorts, an overlay on the video |
| Unrealistic, animated, or slightly altered | Description only |
| Long-form video, prominent label | Directly below the player, above the description |
| Made with YouTube's own AI tools, or C2PA says fully generative | Label is permanent; the creator cannot change it |
Which clips fall in which bucket?
The disclosure page asks you to disclose realistic cases: a real person made to say or do something they did not, altered footage of a real event or place, a realistic scene that did not occur, and music that is a main focus. It lists cases that need no disclosure, such as fantastical content, fully animated content, beauty filters, colour and lighting changes, caption creation, audio repair and own-voice cloning.
A Sume video model clip of a made-up scene that looks like camera footage belongs in the first bucket. A stylised animation belongs in the second.
How do you record the answer per file?
- In YouTube Studio the question sits under Attributes as AI use, Yes or No.
- Probe each file with video_inspect and write the bucket next to the job ID.
- If you cut a photorealistic clip into a longer video with Timeline 1.0 ($0.10 per output minute), treat the long video as photorealistic too.
- Trim variants for a Short with video_trim ($0.02 per job) and inherit the answer from the parent.
What if you get it wrong?
The disclosure page says YouTube can apply a label itself when it detects significant photorealistic AI use, and creators who consistently decline to disclose may face penalties. See what YouTube can do. The blog also states: a disclosure label alone does not change how a video is recommended or whether it is eligible to earn money.
If a clip sits between the buckets, answer on the side of disclosure. The page treats consistent non-disclosure as the problem, and a prominent label alone does not change recommendations or earning, per YouTube's blog. Over-disclosing a stylised clip costs little; hiding a realistic one costs more.
Sources
Related posts
More in Use cases
- YouTube AI label: player if photoreal, description if animated
YouTube shows the AI label in the player for photorealistic content and in the expanded description for animated content. Which Sume clips fall where.
- YouTube can add an AI label you cannot remove: C2PA 2.1 and Sume jobs
YouTube auto-labels files with Content Credentials 2.1+ and can add a label creators cannot remove. What to record per Sume job, and what the docs do not claim.
- Does a lip-synced dub in your own voice need YouTube's AI label?
YouTube says cloning your own voice for voice-overs or dubs needs no disclosure; a real person made to say what they never said does. How Sume lip sync fits.
- YouTube clip compilations: the reused content rule and narrative
YouTube's reused content policy flags clips edited with little narrative and allows storyline plus commentary. Add your text with caption cues.
Written by Sume