What are Content Credentials (C2PA)? A plain explanation

What they are

Think of Content Credentials as a label sewn into a file. The C2PA describes them as an open standard for showing the origin and edits of digital content, and compares them to a nutrition label for a picture. The name C2PA stands for the Coalition for Content Provenance and Authenticity.

Where a site or app supports them, you may see a small “CR” pin on an image. Content Credentials says the pin opens a panel with how the content was made and a record of its editing history.

What a credential records

A credential, which the standard also calls a manifest, can hold several kinds of statement. The C2PA’s FAQ names the main parts:

• Statements about the file, which the standard calls assertions. For example: taken with this camera, made with this AI tool, cropped, color changed.
• Ingredients, meaning the other files this one was built from, such as the two photos in a composite.
• Hashes, a fingerprint of the picture and of the record, so any change shows.
• A digital signature from the app or device that wrote the record.

How the signature works

The app or camera signs the record with a certificate. The same FAQ says those certificates come from certificate authorities on a C2PA trust list. When a reader checks the file, it recomputes the fingerprint and checks the signature. If someone changed a single pixel or edited the history after signing, the check fails and the reader says so.

When a supporting app edits a signed file, it adds a new record that points back to the old one. That chain is how a reader can show “taken on a camera, then edited in this app, then exported”.

Who reads them

Anyone can. The records are meant to be checked by ordinary people, not only by experts. You can:

• open a file in a Content Credentials reader, linked from contentcredentials.org;
• look for the pin on sites and apps that show it;
• use editing apps that display the history of a file you open.

Newsrooms use them to check where a picture came from before they publish it. Platforms can use them to label AI images.

What they can’t tell you

1. Whether the picture is true. A credential says which tool made the file and how it changed. A staged scene photographed with a real camera can carry a perfect record.
2. Anything, once removed. The C2PA says a record can be separated from the file. It adds that invisible watermarks or fingerprints can help find a lost record again, but only where a tool used them.
3. Anything about files with no record. Most images online today carry none, so a missing record is normal, not suspicious.

Our guide on telling whether an image is AI generated puts credentials next to the other checks.

Privacy: what you might share

The C2PA says its core standard does not attribute content to people or organizations, though some groups build add-ons that do. A record still describes your device and your edits, much as EXIF data does. If a tool you use adds credentials, look at what it writes before you post.

Apple has its own way to show a photo is unaltered. Apple’s guide describes Reference Image on iPhone 18 Pro models: the camera keeps reference data so you can verify what it originally captured, and you choose whether to include it each time you share. Apple’s page does not call it Content Credentials, so don’t assume a credential reader can open it.

Where Sealed helps, and where it doesn’t

It helps keep an original photo exactly as the camera made it, inside your vault and out of the shared library. Sealed encrypts the photo data it receives as it is, without shrinking or recompressing it first. Sealed imports only the photos and videos you choose, and deletes the originals in Photos only when you ask.

It doesn’t add or read Content Credentials. For checking a file, use a credential reader or the metadata viewer on this site.