YouTube's in-Shorts AI clips are 10 s and auto-labelled; yours are not

YouTube's Shorts AI tools make clips up to 10 seconds, auto-disclosed. A clip made on Sume and uploaded is not auto-labelled; you set the answer yourself.

4 min readSume
All posts

Two routes to an AI clip in a Short

YouTube's help page on creating Shorts with AI-generated features (read 2026-10-09) says clips made with its in-app tools run up to 10 seconds and are disclosed automatically. It also says availability is limited and mostly excludes the EEA, the UK and Switzerland. A clip you generate elsewhere and upload goes through the ordinary upload disclosure question, which you answer yourself.

In-app AI clip versus uploaded clip, YouTube Help pages read 2026-10-09
QuestionMade with YouTube's Shorts AI toolsGenerated elsewhere, uploaded
Maximum clip lengthUp to 10 secondsUp to the Shorts limit of 3 minutes
DisclosureApplied automaticallySet by you in the upload flow
AvailabilityMostly outside the EEA, UK and SwitzerlandAny channel that can upload Shorts
Label removalNot removable by you for labels YouTube appliesDepends on your answer and on metadata

What that means for a Sume clip

A video from Sume's video models arrives as an MP4 URL. YouTube's disclosure page asks creators to disclose realistic altered or synthetic content, with a photoreal label in the player and an animated label in the description, and it lists exemptions such as beauty filters, color adjustment, captions and audio repair. The docs for Sume do not describe embedding provenance metadata in the file, so do not expect the upload to be labelled for you; treat the question as yours to answer.

The practical step is to write the answer down when the job finishes, next to the job id and model, so the person uploading does not have to guess weeks later.

Length differences that matter

The 10-second ceiling of the in-app tool matches the longest clip of gemini-omni-flash-1.1 on Sume (3 to 10 seconds), but the two are unrelated products. Other Sume models go further: seedance-2.5 and wan-3.0 to 30 seconds, and seedance-2 and the minimax models to 15. A 3-minute Short assembled from clips is a Timeline job, and the label question applies to the finished upload, not to each clip.

  • Omni Flash 1.1 at 360p is $0.0375 per second, so a 10-second clip is 10 x $0.0375 = $0.375 (Sume prices read 2026-10-08).
  • A 30-second wan-3.0 clip at 480p is 30 x $0.0625 = $1.875.
  • Rendering clips into one Short in Timeline costs $0.10 per output minute, rounded up.

A short decision rule

Use the in-app tool when you want a quick 10-second clip, are in a supported region, and are happy with the automatic label. Use an external generator when you need longer clips, a specific ratio or model, or a reproducible pipeline. In the second case, decide the disclosure answer per upload, and keep the page's exemptions in mind: a script written with AI, or a thumbnail or outline, does not need one, while a realistic synthetic scene does.

Sources

Related posts

More in Use cases

All Use cases posts

Written by Sume