Metadata Editor

EXIF EditorEXIF ReaderRemove MetadataGuidesGlossaryAbout

TikTok AI generated label: clean cover images and stills before upload

By Alexis Guay

Files are processed on our servers and deleted immediately. Never stored.

TikTok can show notices such as contains AI-generated media when metadata in your upload still describes AI creation or AI-assisted editing — even if the video itself was shot on a phone. For many creators the fix is not in TikTok’s settings; it is in the cover image, Photo mode still, or carousel slide you attach, which may still carry C2PA or XMP from DALL-E, CapCut, Photoshop, or Firefly exports.

Where TikTok reads metadata

TikTok’s pipeline inspects the files you upload. Still images (covers, slides, PNG/JPEG attachments) can include:

  • C2PA manifests in JPEG APP11, PNG caBX, or WebP chunks
  • XMP generation parameters (Midjourney-style) or IPTC digital source type
  • Standard EXIF software tags naming AI tools

Video containers (MP4) can carry C2PA in binary boxes. This guide focuses on raster images you can run through the metadata remover. If provenance lives only inside the MP4, you may need a video re-encode outside this site — metadata-only tools cannot edit video atoms here.

Workflow before you publish

  1. Inspect — Upload the cover or still to the EXIF reader or run exiftool -G1 -a -C2PA:all -XMP:DigitalSourceType file.jpg.
  2. Strip — Download a cleaned copy from the metadata remover (JPEG, PNG, TIFF, WebP).
  3. Replace — Upload the cleaned still as the cover or slide in TikTok, not the original export from the AI app.

The same three steps apply across platforms; see remove AI label from photos for generator-by-generator notes and false positives on real photos edited with AI tools.

CapCut and Reel-style covers

Exports from CapCut, Canva AI backgrounds, or Photoshop Generative Fill often embed C2PA on the image even when the subject is real. TikTok does not distinguish “touched up with AI” vs “fully synthetic” at the metadata layer — if the manifest says AI participated, you can get the disclosure. Stripping metadata on a copy you own is the same upload-prep step creators use before Instagram; compare Instagram’s AI generated label.

Limits

  • Pixel watermarks (for example SynthID) are not removed by metadata stripping.
  • TikTok can change detection; treat platform help centers as authoritative.
  • Removing metadata does not bypass laws or platform terms that require AI disclosure where applicable.

FAQ

Related guides

Remove AI label from a photo: C2PA, XMP, and what platforms read at upload

How to remove AI Info and Made with AI triggers from JPG and PNG: inspect C2PA and IPTC with ExifTool, strip markers with a free online remover, and know what metadata cannot fix.

Instagram's AI generated label: what triggers it and how to remove C2PA first

Why Instagram shows an AI info label on some photos, what Meta reads from C2PA and IPTC metadata at upload, how to check your file, and how to strip those markers before posting.

C2PA and Content Credentials: how signed image provenance works

C2PA attaches a signed provenance record to an image. What a manifest holds, where JPEG, PNG and WebP store it, how to inspect it, and what removing it does.

Does social media remove metadata from your photos?

What Facebook, Instagram, X, WhatsApp, Telegram, Signal, Discord, iMessage, email and cloud storage do to EXIF and GPS, and how to test any platform yourself.

Other Metadata Tools

Remove AI Label

Strip C2PA and XMP before Instagram or TikTok upload.

AI Metadata Checker

Inspect C2PA and XMP before you clean a file.

EXIF Editor

Edit camera settings, dates, GPS, and author fields in EXIF.

EXIF Reader

Read EXIF, GPS, and IPTC/XMP rights from your photos.

Remove Metadata

Strip all metadata from your images to protect your privacy.