BlogProvenance

Content Credentials

How to remove C2PA / Content Credentials from an image

C2PA Content Credentials are a structured provenance record — not a pixel watermark. Here's what they are, how a structural strip works, and where the limits are.

C2PA — the standard behind Adobe's Content Credentials — attaches a signed manifest to a file describing how it was made and edited. Because it is structured data embedded in the container (JUMBF boxes in JPEG, caBX chunks in PNG, and similar), removing it is a parsing-and-editing problem, not a pixel problem. That makes it far more tractable than a watermark, with one important caveat we'll get to.

C2PA is metadata, not a watermark

This distinction matters more than any other point on this page. A C2PA manifest is a record bolted onto the file. A SynthID-class watermark is baked into the pixels. Stripping the manifest cleanly removes the provenance record but does nothing to an embedded signal — and vice versa. A file can carry both, one, or neither, so a real solution has to handle them as separate layers.

How a structural strip works

  1. Detect the containersWalk the file structure for C2PA/JUMBF markers, linked-manifest URLs, and PNG caBX chunks.
  2. Choose lossless where possibleFor supported JPEG/PNG, provenance segments can be excised without re-encoding, preserving image quality.
  3. Re-encode as fallbackFor other containers, a clean re-encode drops the manifest along with other metadata.
  4. Confirm and reportRe-scan to confirm the markers are gone and record what was removed.

The limits worth knowing

  • Platforms can re-attach or re-derive provenance after upload, so a locally clean file may be re-tagged.
  • Removing C2PA does not remove a pixel watermark that may also be present.
  • Provider terms may prohibit obscuring provenance; the technical ability to strip it does not grant the right to.
Try it on your own file

Best-effort, multi-layer, and reported honestly — originals always preserved.

Create an account