YouTube's Shorts AI tools self-label; a clip you upload does not

Clips from YouTube's own Shorts AI tools are disclosed automatically and run up to 10 seconds. A clip generated elsewhere and uploaded needs your own toggle.

4 min readSume
All posts

The difference in one paragraph

YouTube's page on creating Shorts with AI features (read 2026-10-09) says creators using YouTube's generative tools "don't need to take extra steps to disclose the use of altered or synthetic content" because the tool discloses it automatically. A clip you generate with another service and upload as a Short is not covered by that: the GenAI disclosure page asks you to answer the "AI use" question yourself for realistic content, and says YouTube can add a label that cannot be removed if you do not.

What the built-in tools make

The page lists green screen backgrounds generated from text prompts, standalone video clips "up to 10 seconds long" from text prompts and reference images, photo to video, a feature that animates a photo of yourself with preset movements, and an audio-first conversion that is English only in select countries. It adds that most features are not available in the European Economic Area, United Kingdom or Switzerland, and that reference image uploads for video generation are unavailable in the UK.

YouTube also says it may block prompts that violate its policies and designs the tools to help creators avoid photorealistic images of identifiable people.

The same 10 seconds on Sume

Sume's video API lists each model's supported durations at GET /v1/videos/models, and prices are per output second inside that range. The table prices a 10-second clip at the lowest tier each model lists, using the public list prices read from the Sume catalog on 2026-10-08 (provider list times 1.25).

Cost of a 10-second clip by Sume model, lowest listed resolution, price per second times 10 (catalog read 2026-10-08)
ModelDuration rangeTierPer second10 seconds
gemini-omni-flash-1.13-10 s360p$0.0375$0.375
minimax-h35-15 s480p$0.0625$0.625
wan-3.02-30 s480p$0.0625$0.625
seedance-24-15 s480p$0.17577$1.7577
seedance-2.54-30 s480p$0.268677$2.68677

What you do after the upload

Generating elsewhere changes one thing: the disclosure decision moves to you. If the clip shows a realistic scene that did not occur, answer yes to the AI use question in Studio. If it is clearly animated, YouTube's page says disclosure is not required. If you used a Sume model with native audio, such as Gemini Omni Flash 1.1 whose audio is always on per the docs, remember the audio is generated too.

Keep the job id for each clip. The point of YouTube's built-in disclosure is that nobody has to remember; with an outside tool, a one-line record is the substitute.

Choosing between the two routes

The built-in tools are the simpler path when a 10-second background or a short animated photo is enough: there is no export step and no disclosure decision. They are limited by what the page lists, and by region.

An outside generator is the path when you need longer shots, a specific aspect ratio, or more control. In that case the cost is the toggle and the record-keeping. Sume's video catalog lists durations up to 30 seconds for seedance-2.5 and wan-3.0, so one shot can be three times the built-in length. Check the exact ranges with the models endpoint before you submit, since limits differ per model and the lists change.

Sources

Related posts

More in Use cases

All Use cases posts

Written by Sume