C2PA Metadata Remover

C2PA stands for the Coalition for Content Provenance and Authenticity. It defines a standard format for recording how a piece of digital content was created, edited and handled.

Information is stored in a manifest — a structured record attached to the file. Signed manifests can be verified, which is what makes them useful for provenance. What C2PA is is covered in more detail in the guides.

Content Credentials is the consumer-facing name for C2PA provenance data. In an image file, you will usually see them as a manifest with information about the producing or editing tool.

Camera makers, editing software and AI image tools can all write these manifests, which is why C2PA often comes up alongside AI metadata.

Because a tool that handled the image chose to attach it. That could be a camera that signs photos at capture, an editor that records an edit history, or a generator that records its own output.

CleanImage checks two things: whether a metadata reader reports C2PA tags, and whether the file structure contains the fingerprints of a C2PA manifest — the JUMBF box, the C2PA segment identifier, or an XMP manifest reference.

If either check finds something, Content Credentials are reported on the analysis screen.

When the manifest is stored as metadata that the encoder can drop, re-encoding the image leaves it out of the cleaned file. Because the cleaned copy is analysed again, that result is verified rather than assumed.

Some manifests are cryptographically bound to the content or are written back by the encoder. In those cases the manifest survives re-encoding, and the report shows Content Credentials as not removed.

This is stated plainly on purpose. A tool that always claims success on C2PA would be misleading.

Cleaning a file removes the manifest from that file when it can. It does not change how a platform classifies an image, and it does not remove pixel-level watermarks. See the AI metadata page for the full picture.