The TikTok AI-generated content label, via API

What TikTok asks for

TikTok's policy requires creators to label AI-generated content that contains realistic images, audio or video. It encourages, rather than requires, labels for content that is clearly unrealistic. TikTok also detects some AI content itself using Content Credentials metadata and may apply its own label automatically.

Unlabeled realistic AI content can be removed under TikTok's integrity and authenticity rules. For a UGC operation, that risk sits on the creator account, which is the asset you least want to lose.

The API field

TikTok's Content Posting API has an optional boolean, is_aigc. On video posts it sits in post_info, and TikTok's reference says the video is then labeled with the AI-generated tag. On photo posts, TikTok's reference says the photo gets an AI-generated label. In UGC API the same switch is tiktok.ai_generated.

{
  "accounts": ["acc_tt_01"],
  "format": "video",
  "media": ["https://cdn.example.com/avatar/a-77.mp4"],
  "caption": "I asked 3 dermatologists about retinol",
  "options": { "tiktok": { "ai_generated": true } }
}

When to set it

When in doubt, label. The cost of a label is small compared to the cost of a removed post or a strike on the account.

Does the label hurt reach?

TikTok hasn't said that it does, and we have no data to claim either way. What we can say: if TikTok detects AI content you didn't label, you lose control of how it's presented. Labeling it yourself keeps the decision with you.

Building it into your pipeline

1. Tag each asset with how it was made at generation time: real, AI-edited, AI-generated.
2. Map AI-generated and realistic to ai_generated: true automatically.
3. Let the creator or reviewer override it before scheduling.
4. Store the flag with the post, so you can show it in reports.

AI UGC tools that generate thousands of variations should make this default-on for avatar and voice content. See AI UGC tools.

Other platforms

Meta labels some AI content on Instagram and Facebook using industry signals and its own detection, and has its own disclosure rules for realistic AI. Meta's Facebook Reels publishing API lists an is_ai_generated field. YouTube asks creators to disclose realistic altered or synthetic content in YouTube Studio. Check each platform's current policy, since this area changes fast.

Related settings

• Commercial content disclosure is separate from the AI label. TikTok requires apps to offer it in their posting UX. See the TikTok audit guide.
• Drafts (tiktok.draft: true) let the creator review labels and disclosures in the TikTok app before posting.
• All TikTok options are listed in formats and options.

A policy you can hand to your team

1. If a person, voice or scene in the video was generated and looks real, label it.
2. If AI only edited real footage, such as captions, color or cleanup, don't label it unless the edit changes what happened.
3. If you aren't sure, label it.
4. Never remove a label TikTok applied automatically by reposting the same file.

Write this down once and apply it in code, so the decision doesn't depend on who schedules the post that day.