How TikTok applies its AI-generated label, and what the label proves
TikTok says an AI-generated label reaches a post in three ways: a creator switches it on, TikTok's own AI effects add it, or TikTok's systems add it after finding Content Credentials or by detection. In every case the label is a claim by the creator or by TikTok. It is not proof of how the content was made.
How does TikTok say the AI-generated label is applied?
TikTok's support page About AI-generated content describes two labels. We read it in a browser on October 30, 2026. The page shows no date, so we cannot say when its text was last changed.
| Label | How TikTok says it is applied | Who makes the claim |
|---|---|---|
| Creator label, which reads "creator labeled as AI-generated" | The creator turns on the AI-generated content setting in the TikTok app before posting. | The creator |
| Auto label, "AI-generated" | TikTok may apply it to content it identifies as completely generated or significantly edited with AI. TikTok gives two examples: a creator uses TikTok AI effects, or uploads AI-generated content that has Content Credentials attached. | TikTok |
The page adds that content made only with TikTok effects may not need a creator label, because TikTok says it labels content made with its effects automatically if they use AI. It also says that once a post carries an auto label, the creator cannot remove it.
What does TikTok ask creators to label?
The same page separates a request from a rule. TikTok says it encourages creators to label content that is completely generated or significantly edited by AI. It also says it requires creators to label all AI-generated content that contains realistic images, audio and video, as set out in its Community Guidelines.
TikTok says that misleadingly labeling unaltered content with the creator label is a violation of its Terms of Service and may result in removal of the content. It also says that turning the setting on does not affect distribution as long as the content follows its Community Guidelines. We do not go beyond that text on what a creator must label.
Which systems does TikTok say it uses?
TikTok's newsroom post More ways to spot, shape and understand AI-generated content is dated November 19, 2025. We read it on October 30, 2026. It says TikTok requires people to label realistic AI-generated content and layers several strategies to apply that rule:
- labeling tools it offers creators;
- its own detection models;
- C2PA Content Credentials, which TikTok says embed metadata that lets it, and other platforms that use C2PA, know when something is AI-generated.
TikTok says these efforts helped label over 1.3 billion videos to date. It also says that, over the last year, it added context to its labels about whether content was labeled because of its AI detection, creator labels or TikTok AI tools. The pages we read do not describe how the detection models decide.
The post also says such labels may be removed when content is reuploaded or edited elsewhere. TikTok says it was testing invisible watermarks as another layer; our guide on whether TikTok watermarks AI content covers that.
What does the label prove?
It proves that TikTok shows the label on that post. It does not prove more, because each route is a claim by someone:
- A creator label is a creator's own statement. TikTok's page itself warns against using it on unaltered content.
- An auto label from Content Credentials rests on a record someone attached to the file. A signed record shows who signed and that the bytes match; the C2PA explainer says provenance alone cannot show that content is true.
- An auto label from detection rests on TikTok's models, whose workings we have not seen.
A missing label proves less. TikTok's own Identify AI-Generated Content page says that when no signal is found, the file may not come from TikTok, CapCut, Dreamina, Hypic or Pippit, or may not be AI-generated at all, or the signals may have been lost. It also says that if other platforms later changed a file, the tool cannot identify those changes. We read the page on October 30, 2026. Its upload note says TikTok deletes the file right after showing the result. It opens another company's site, you may need to sign in, and your file goes to them, not to us.
Would our checker read a record signed by TikTok?
Our checker compares a record's signing certificate with the trust lists we load. On October 30, 2026 we read the subject of all 174 certificates in the three list files that hold signers. Our copies were fetched October 27, 2026. Our accuracy page shows the lists and their dates.
| List and file | Certificates | Naming TikTok |
|---|---|---|
| C2PA Trust List, signer anchors | 32 | 0 |
| Interim Trust List, anchors | 27 | 0 |
| Interim Trust List, allowed certificates | 115 | 1 |
The one match names "TikTok Inc." and an address at tiktok.com. It sits on the allowed list of the Interim Trust List, which the Content Authenticity Initiative's trust list page says was frozen on January 1, 2026. No certificate names ByteDance, and our copy of the current C2PA Trust List has no TikTok subject. A different company, TikBox.io Ltd, holds two certificates on the allowed list; it is not TikTok.
The Interim pass of our checker loads that allowed list, so a record signed with the TikTok certificate would be judged against it. We hold no TikTok-signed file and have not seen how one reads, so we cannot say which certificate TikTok uses or whether it signs every file. A name on a list does not show which products sign by default.
What this means when you check an image
Open the checker and choose an image file. If it carried a signed record whose signer is on the Interim Trust List only, you would see the headline "Content Credentials found; legacy trust-list match". It means a signed record is present and its signer is on the frozen list, not the current C2PA Trust List. It does not say the record's claims are true, and it says nothing about a TikTok label.
The checker reads image files, not posts. A video such as an MP4 or MOV reads "Unable to check this file", as the one-second test clips we made did on October 30, 2026. For an image with no record or label, "No supported provenance signals found" does not mean it is real or human-made. These checks cannot determine whether an image is AI-generated.
We do not check TikTok's invisible watermark. If you saw something on TikTok, TikTok's checker is the place to look for signals only TikTok can read. Our Privacy and method page explains what our checker does with your file.
Guides on this topic
- Visible AI watermarks: what they show and what they do not is the main page on this topic.
- Does TikTok watermark AI content? covers TikTok's invisible mark and its checker.
- YouTube's altered or synthetic content label shows how another platform applies its label.
- Platform labels compared sets the platforms side by side.
- Which platforms read or show Content Credentials? covers where records are read.
Frequently asked questions
- Can I remove an auto label from my own post?
- TikTok's support page says no: once content carries an auto label, the creator cannot remove it.
- Does the label mean the content was made with AI?
- It means TikTok shows the label. TikTok says the creator label is the creator's own disclosure, and that misleadingly applying it to unaltered content violates its terms.
- Does a missing label mean the content was not made with AI?
- No. TikTok says labels may be removed when content is reuploaded or edited elsewhere, and its checker page says missing signals have several causes.
- Does our checker show a TikTok label?
- No. It reads signals inside an image file. We do not check TikTok's watermark, and we hold no file signed by TikTok.