What Are C2PA Content Credentials? A Plain-English Guide
Content Credentials are signed records of how an image was made and edited. Here's what they contain, who adds them, and how to check a file.
If you've used an AI image generator or a recent version of a major photo editor, your images may carry a C2PA manifest, also known as Content Credentials. It's a type of metadata designed to record where an image came from and what happened to it along the way.
This guide explains what's inside, who writes it, and what it means for you.
What is C2PA?
C2PA stands for the Coalition for Content Provenance and Authenticity, an industry group formed in 2021. Its founding members included Adobe, Microsoft, Intel, Arm, the BBC and Truepic. The coalition publishes an open technical standard for attaching provenance (a history of origin and edits) to media files.
"Content Credentials" is the consumer-facing name for the same thing. You may see a small "CR" icon on sites that display them.
What's inside a manifest?
A C2PA manifest is a bundle of statements, called assertions, that is cryptographically signed. Typical assertions include:
| Assertion | What it records |
|---|---|
| Actions | What was done: created, edited, cropped, filtered, or generated |
| Digital source type | How the content was made, e.g. captured by a camera or generated by a trained AI model |
| Claim generator | The software that wrote the manifest |
| Ingredients | Other files used to make this one, with their own manifests |
| Signature | A certificate identifying who signed the claim |
The signature is the important part. If someone edits the image without updating the manifest, verification tools can tell the record no longer matches the file.
Who adds Content Credentials?
- AI image generators. Several major generators attach C2PA manifests that mark images as AI-generated, including OpenAI's image tools and Adobe Firefly.
- Editing software. Some editors add manifests when AI features such as generative fill are used, even if the rest of the photo is real.
- Cameras and phones. A growing number of cameras and phones can sign photos at the moment of capture, to show they came from a real sensor.
Where is it stored in the file?
The manifest lives in a container format called JUMBF:
- In JPEG files, it sits in one or more
APP11segments. - In PNG files, it sits in a chunk named
caBX. - In WebP, AVIF and video formats, it uses equivalent boxes or chunks.
Related information often appears in XMP metadata too, for example the IPTC "digital source type" field, which can say trainedAlgorithmicMedia for AI-generated images.
How to check if an image has Content Credentials
- The Content Authenticity Initiative runs a public verification site, contentcredentials.org/verify, that reads and displays manifests.
- The MetaZero cleaner flags C2PA manifests, the software that signed them, and any IPTC "AI generated" tags, all locally in your browser.
Does removing metadata remove Content Credentials?
Yes, the manifest is metadata, so a metadata cleaner removes it, along with the XMP fields that reference it.
That's a known limitation of the standard. The C2PA specification also describes soft bindings, such as invisible watermarks or content fingerprints, that let a manifest be recovered from a cloud registry even after it has been stripped from the file. Some providers use separate pixel-level watermarks as well. Those live in the image itself, not in the metadata, so no metadata tool removes them.
Should you keep them or remove them?
It depends on why you're sharing:
- Keep them if you want to prove a photo is authentic, show your editing history, or meet a publication's disclosure rules.
- Remove them if a manifest exposes more than you'd like, such as the names of the tools you used, your editing steps or identifying certificate details, and you're sharing your own work.
Whatever you decide, follow the disclosure rules of the platforms and publications you post to. Removing metadata gives you control over what a file says about you. It doesn't change what the image is.