How to remove the AI-generated label on TikTok

How TikTok detects AI content

In May 2024 TikTok became the first major video platform to adopt C2PA Content Credentials. In its own words, Content Credentials “attach metadata to content, which we can use to instantly recognize and label AIGC”. The rollout covered images and videos first. TikTok also attaches Content Credentials to content made with its own AI effects, and those credentials stay on the file when it is downloaded.

On top of the metadata check, TikTok runs its own detection and offers a self-labeling toggle. Its rules require realistic AI-generated content that shows people or scenes to be labeled, and unlabeled realistic content can be removed.

Clean a still image

1. 1

   Check the file

   Drop the image into the tool below. It reads the C2PA manifest, IPTC digital source type and generator traces and tells you which signal TikTok would act on.

2. 2

   Strip or re-encode

   Strip metadata keeps the pixels untouched. Re-encode writes a fresh JPEG or WebP with an optional 3% shrink when you also want new pixels.

3. 3

   Upload the clean copy

   Download the cleaned file and post that one. Do not re-open it in an editor that writes Content Credentials, or the record comes back.

Photo posts and slideshows

TikTok photo carousels are uploaded as individual images, so each one is checked. Clean the whole set. If an image came from DALL-E, Firefly or another tool that writes Content Credentials, the lossless strip removes the manifest; re-encoding does the same and also produces new pixels.

What this does not cover

• Video. This tool handles still images only. Video files carry C2PA in a different container.
• TikTok’s own detection. Classifiers and watermarks from partner generators can still flag content.
• Policy. Realistic AI content must be labeled under TikTok’s rules regardless of metadata.

Questions people ask