TikTok AI label source: detection, creator label or TikTok's own tool

TikTok's AI-generated content label can come from three places, and TikTok says it now adds context about which. In its November 19, 2025 newsroom post, the company says that over the last year it enhanced AIGC labels by adding more context on whether content was labeled due to its AI detection, creator labels, or TikTok AI tools. It also says these efforts helped label over 1.3 billion videos to date.

For a clip made outside TikTok, for example with Sume, the first source you control is the creator label. The other two happen to you.

Three sources, three levers

The post describes a layered approach: labeling tools for creators, TikTok's own detection models, and C2PA Content Credentials.

What to do with a Sume-made clip

TikTok says it requires people to label realistic AI-generated content, and layers its own strategies on top. So the safe path is to apply the creator label when the clip is realistic AI video: Sume video generation output of a realistic person or place counts.

Do not expect a Sume file to carry Content Credentials; the Sume docs describe the media jobs and their fields, not C2PA signing. If the label still appears after you posted without one, the source shown tells you which route caught it, and that tells you what to fix next time.

Steps

• Decide per clip: realistic AI generation, significant AI edit, or neither.
• Apply TikTok's creator label before or at posting for the first two.
• Keep the generation job id and date in your own log.
• After posting, open the label details to see which source the label shows.
• If a label came from detection, add it to your pre-post checklist for that clip type.

Working notes

A practical habit: write down the label source after each post in a small log, with the clip, the date and the label you applied. When TikTok's label details disagree with your own record, you learn whether your disclosure step was skipped or detection added a label to a clip you considered unremarkable. That log is the cheapest way to see, over a few weeks, which clip types attract detection labels, and you can then move those clips to a stricter disclosure step.

Caveat

The newsroom post is a company announcement, not the full policy text. TikTok's advertising policy, which governs ads, has its own rules and says undisclosed AI-generated content in an ad will be rejected or restricted.