Remove the AI content label on TikTok

Remove the AI label on TikTok

TikTok was among the first platforms to adopt C2PA Content Credentials to automatically label content created or retouched with AI. When an image you add to a post or a photo carousel carries those credentials, TikTok can show the AI-generated content notice.

Remove IA Label removes those credentials and the associated metadata without recompressing the image. This helps when your file carries provenance data that does not reflect how you actually made it, but it does not erase invisible watermarks or stop other forms of detection.

Drop images here or click to upload

JPEG, PNG, WebP, AVIF, TIFF, HEIC · up to 5 photos · 100 MB · no GIF · pixels untouched. iCloud or cloud? Download the full original; files under 4 KB are usually a proxy.

How TikTok shows the label

TikTok places a “Created with AI” mark (or similar) on the content, and it reads Content Credentials (C2PA) embedded in images and videos. It also applies its own self-labeling rules for material that goes through its AI effects. In photo posts, each image in the carousel can carry its own credentials.

How to remove it, step by step

Check the image first in the checker to see if it includes C2PA credentials or generation metadata.
Use the /c2pa mode to remove provenance credentials specifically.
Also review EXIF/XMP with /exif and /xmp if you want the file free of extra traces.
Download, check again and upload to TikTok without re-editing.

Nuance and false positives

Exports from editors with provenance turned on (including minor AI retouches on real photos) can make TikTok show the notice even when the content is authentic. Cleaning the metadata fixes that false positive, but TikTok can detect by other means, and pixel-level watermarks stay intact.

Final tip

If you work with carousels, clean every image in the batch: one file that still has credentials is enough for the notice to appear. You can process up to 5 photos at once.

Frequently asked questions

Does TikTok detect AI only from metadata?

Not only. It reads the file’s C2PA Content Credentials, but it also has its own detection and self-labeling. Cleaning metadata helps, but it is not a guarantee.

Can I clean several carousel photos at once?

Yes. We accept up to 5 photos and 100 MB per batch. Clean all of them, because a single image with credentials can trigger the notice.

Does video or photo quality change?

Nothing is recompressed and no pixels are changed; only credentials and metadata are removed. The image stays identical (verified by hash).

Is it legal to remove these credentials?

In general it is legal to clean metadata from your own images. It should not be used to deceive or to commit fraud by presenting content as something it is not.