isGenAI.com

Glossary

Manifest: the signed unit inside Content Credentials

A C2PA manifest is the signed data structure that records one set of provenance statements about an asset, consisting of assertions (what happened), a claim (a signed list of those assertions with content hashes) and a claim signature, stored with any earlier manifests in the file’s manifest store.

Definitions and isGenAI behaviour as of October 2026. Updated .

Inspect a manifestFrequently asked questions
Controlled example PNG file that carries this record: Content Credentials found

test-signed.pngPNG file · controlled example

  1. Signature + IHDR
  2. iTXt XML:com.adobe.xmp
  3. tEXt parameters
  4. tEXt prompt / workflow
  5. tEXt invokeai_metadata
  6. eXIf
  7. caBX C2PA manifestrecord
  8. IDAT pixels
  9. IEND

Content Credentials found

A signed manifest describes how the file was made.

Signer, recording tool, actions and the signature check result are reported separately, so you can weigh each one.

Open the Content Credentials checker →
Content Credentials found: where this record sits inside a PNG file and what the isGenAI report says about it. The picture is a code-drawn controlled example that carries this record; download it and check it yourself.

Parts of a manifest

Assertions describe actions (created, edited, placed), the creative work, the digital source type, ingredients (earlier versions) and hashes of the content. The claim lists assertion hashes and the claim generator, the software that made it. The claim signature is a COSE signature over the claim using the signer’s certificate. The most recent manifest is the active manifest; earlier ones are kept as ingredients, forming a chain of edits.

Manifests are stored in a JUMBF manifest store inside the file.

How isGenAI validates a manifest

The C2PA checker and AI image detector validate the active manifest’s signature, recompute the content hashes to confirm the binding, check the certificate against a trust list and read the actions and digital source types. The report prints the manifest labels present, the claim generator, the validation state and the trust state separately, and lists any validation codes so a reviewer can follow them.

Reading a chain

An image generated by one tool and edited in another may carry two manifests: the generator’s as an ingredient and the editor’s as active. isGenAI reports the active manifest’s actions and notes the ingredients, so an AI generation recorded earlier in the chain is not lost.

Manifests can also carry ingredient thumbnails, a creator identity assertion and training-and-mining preferences. isGenAI lists the assertion labels it finds so a reviewer can see what the signer chose to record, beyond the AI generation action itself.

How to use this term when checking a file

  1. Upload the original file to the AI image detector or the tool named on the page.
  2. Find the finding that uses this term under “How we checked” and read the source field it names.
  3. Weigh it by kind: a signed record, a declared field, a saved prompt and a matching copy are different strengths of evidence.
  4. Remember that absence of a record is not evidence of a human author.

Inspect a manifest

Frequently asked questions

What is the active manifest?

The most recent manifest in the store, describing the current state of the asset; earlier manifests are kept as ingredients.

What does the claim generator field show?

The software or service that created the manifest, such as a generator or editor name and version.

Why might a manifest validate but be untrusted?

The signature is intact but the signer’s certificate does not chain to the trust list. isGenAI reports both facts separately.

Can a manifest be edited?

Any change breaks the signature or the content hashes, which the validator detects and reports.

Sources

Definitions follow the published standards linked above; the isGenAI behaviour described is the deployed checker as of October 2026.