What TikTok does with Content Credentials and AI labels
TikTok auto-labels uploads carrying Content Credentials and requires creators to label realistic AI content or risk removal. Read 12 Oct 2026.
- Published
- Last reviewed
TikTok reads C2PA Content Credentials on upload and automatically labels images and videos that carry them as AI-generated; it requires creators to label realistic AI content themselves, and it may remove, restrict or label what they do not. It is a label-and-disclose platform, not a reject one. Whether a manifest or a watermark survives in the file a viewer downloads is a question we are answering with our own uploads rather than taking anyone's word for it.
- Reads Content Credentials: yes, auto-labelling images and video since 9 May 2024; audio "coming soon" at the time.
- Attaches its own: TikTok said it would add Content Credentials to content made on TikTok, remaining on download.
- Creator duty: label realistic AI-generated or significantly edited content; unlabelled content "may be removed, restricted, or labeled".
- Rejects unmarked content: no.
- Display for untrusted signers and survival of marks: UNVERIFIED, being measured.
What TikTok says it does
From its newsroom post of 9 May 2024, read 12 October 2026: "TikTok is starting to automatically label AI-generated content (AIGC) when it's uploaded from certain other platforms", by "launching the ability to read Content Credentials", which "attach metadata to content, which we can use to instantly recognize and label AIGC. This capability started rolling out today on images and videos, and will be coming to audio-only content soon." It added: "Over the coming months, we'll also start attaching Content Credentials to TikTok content, which will remain on content when downloaded", so that "anyone will be able to use C2PA's Verify tool" on it. TikTok joined the Content Authenticity Initiative the same day and described itself as "the first video sharing platform to put Content Credentials into practice".
The Community Guidelines, read 12 October 2026, set the creator's duty: "we require creators to label AI-generated or significantly edited content that shows realistic-looking scenes or people", and "Unlabeled content may be removed, restricted, or labeled by our team, depending on the harm it could cause." Misleading content "about matters of public importance" is not allowed even with a label. Secondary reports put TikTok's labelled total at over three billion videos by August 2026 and describe an in-app hub on spotting AI content launched in July 2026 (UNVERIFIED against TikTok's own pages).
What this means for your output
A manifest on your output gets your content labelled on TikTok automatically, which is the law's intent and, under the Community Guidelines, the creator's obligation anyway. What the label says about who generated it depends on the manifest's signer: a supplier's manifest names the supplier. Whether TikTok shows anything for a signer that is not on the Trust List, and whether the manifest and watermark are still in the downloaded file after TikTok's re-encode, are the two things we are measuring in our platform survival suite; this page will carry the results with their dates.
Changes
- 2026-10-12 — First published, from TikTok's newsroom post and Community Guidelines.
Read from TikTok's own pages on the date shown. Platform policies change; tell us if this is out of date. SignUnder is not a law firm and this is not legal advice.
Questions
- Does TikTok read C2PA Content Credentials?
- Yes. Since 9 May 2024 it automatically labels uploaded images and videos that carry Content Credentials as AI-generated, and it said audio would follow. It also said it would attach Content Credentials to content made on TikTok so they remain when downloaded.
- Do I have to label AI content on TikTok?
- Creators must label AI-generated or significantly edited content that shows realistic-looking scenes or people, under the Community Guidelines. Unlabelled content may be removed, restricted or labelled by TikTok.
- Does a Content Credential survive a TikTok upload?
- TikTok reads it on upload to apply the label. Whether the manifest is still present in the file a viewer downloads, and whether an invisible watermark survives TikTok's re-encoding, is something we are measuring ourselves; until then it is UNVERIFIED.
- What does TikTok show for a manifest whose signer is not on the Trust List?
- Undocumented, as of 12 October 2026. We are testing it with uploads signed by trusted and untrusted certificates and will publish what displays.