C2PA Standard

The Manifest Store is the core data structure that holds all provenance claims about a media asset. It is a CBOR (Concise Binary Object Representation) encoded container that cryptographically binds a series of assertions, signatures, and ingredient references. The store is designed to be embedded directly within the file format (e.g., JPEG, PNG, AVIF) or served as an external sidecar. It contains:

• A claim with the content's creation and edit history
• A signature block proving the identity of each actor
• An ingredient list referencing parent assets used in composition This structure ensures that the provenance data travels with the asset across the internet, enabling any viewer to validate its history.

To prevent the manifest from being separated from the media it describes, C2PA employs a hard binding mechanism. This is achieved by computing a cryptographic hash (typically SHA-256) of the final asset's binary data and embedding that hash directly into the manifest's claim signature. Any subsequent modification to a single pixel or byte of the asset will invalidate the hash, immediately signaling tampering. This creates an existential dependence between the asset and its provenance metadata, ensuring that a valid manifest cannot be transplanted onto a different or altered piece of content.

C2PA leverages the W3C Verifiable Credentials (VC) data model to standardize the identity assertions made by content creators and editors. Instead of relying on centralized certificate authorities, the standard supports Decentralized Identifiers (DIDs). This allows an organization or individual to cryptographically prove control over their identifier without a single point of failure. A camera manufacturer, for example, can issue a VC attesting to the GPS coordinates and timestamp of capture, signing it with a private key controlled by a DID registered on a distributed ledger. This provides a vendor-neutral trust framework for identity.

C2PA models content creation as a directed acyclic graph of ingredients. When a piece of media is created by compositing multiple source assets (e.g., a Photoshop document with multiple layers), the final manifest includes a reference to each source asset's own manifest. This creates a recursive provenance tree. A validator can traverse this tree to verify the authenticity of every component, not just the final output. This model directly addresses splicing attacks, where a small fake element is inserted into an otherwise authentic scene, by forcing the editor to disclose the lineage of all imported elements.

A critical privacy feature of the C2PA architecture is the ability to redact sensitive information without breaking the cryptographic chain of trust. Using a salted hash technique, a manifest can contain claims that are individually hashed. A publisher can later remove specific claims (e.g., the photographer's exact GPS coordinates) and provide only the hash of the removed data. A validator can still mathematically verify that the remaining claims haven't been altered, while the sensitive data remains hidden. This enables privacy-preserving provenance, balancing transparency with the need to protect personally identifiable information.

The C2PA specification does not dictate who is trustworthy; it provides the cryptographic evidence for a user to decide. The validation process checks:

• Signature integrity: Does the cryptographic signature match the signer's public key?
• Certificate chain: Does the signer's certificate chain terminate in a trusted root CA?
• Timestamping: Was the signature applied within the certificate's validity period, as proven by a trusted timestamp authority (TSA)?
• Trust list membership: Is the signer's identity present on a user-curated or industry-defined trust list? This separation of cryptographic validation from trust evaluation allows the standard to be used across different legal and cultural contexts.