Upload a supported image
Sign in and add a JPEG, PNG, WebP, AVIF, HEIC, or HEIF file up to 20 MB that carries Content Credentials.
C2PA Content Credentials are a signed provenance manifest that labels a file as AI-made or edited. We strip the JUMBF/APP11 segment and “Made with AI” XMP/EXIF fields from images you own. Learn where these live in our metadata primer, or read the full walkthrough.
The signed manifest and the provenance labels, cleared in one pass.
Sign in and add a JPEG, PNG, WebP, AVIF, HEIC, or HEIF file up to 20 MB that carries Content Credentials.
We remove the C2PA manifest (JUMBF/APP11 or caBX) and the “Made with AI” XMP/EXIF fields, in isolation.
Retrieve the cleaned image from your signed-in account and confirm at contentcredentials.org/verify. First image free.
C2PA is provenance metadata, not a pixel watermark. It is invisible in the picture but fully readable by any verifier, and removable from files you own.
A cryptographically signed record embedded as a JUMBF/APP11 segment in JPEG or a caBX chunk in PNG, listing the tool and edit history.
XMP and EXIF entries that flag the file as AI-generated or edited are stripped alongside the manifest.
C2PA is separate from embedded signals like SynthID. If the file also carries SynthID, that needs regeneration. See our SynthID page.
C2PA (the Coalition for Content Provenance and Authenticity) defines Content Credentials, a signed manifest that records how an image was made and edited. In a JPEG it is stored as a JUMBF box inside an APP11 marker segment; in a PNG it lives in a caBX chunk. Tools like ChatGPT, DALL·E, Adobe Firefly, and many cameras attach it. Our metadata and Content Credentials primer maps it out.
Content Credentials do not change how the picture looks, but anyone can inspect them at contentcredentials.org/verify and see the “Made with AI” label and edit trail. That is often what triggers AI labeling on social platforms, even when nothing is visible in the image itself.
For images you own, we strip the C2PA manifest and the associated XMP/EXIF provenance fields, so a verifier no longer finds Content Credentials. This is a metadata operation and does not degrade the picture. It pairs naturally with the ChatGPT & DALL·E remover and the general AI watermark remover.
Stripping Content Credentials does not touch embedded pixel watermarks such as SynthID, which survive a metadata wipe and need local regeneration instead. If your file has both, handle both. Read what SynthID is and pick an output with the format guide.
They are a signed provenance manifest that records how an image was created and edited, including a “Made with AI” label. In JPEG they sit in a JUMBF/APP11 segment; in PNG they sit in a caBX chunk.
Upload an image you own and we strip the C2PA manifest and the “Made with AI” XMP/EXIF fields so verifiers no longer find Content Credentials. The first image is free.
No. C2PA is metadata alongside the pixels, while SynthID is embedded in the pixels. Stripping Content Credentials leaves SynthID intact, so an embedded signal needs regeneration.
No. Removing the manifest and metadata is a container-level operation and does not alter the visible pixels of your image.
Upload the processed file to contentcredentials.org/verify. If no Content Credentials are found, the manifest was removed. Treat it as one signal, since platforms may run their own checks.
Remove Content Credentials only from images you created or have the right to edit, and do not use removal to misrepresent authorship or provenance. Delete SynthID is an independent service. See responsible editing and whether removal is legal.