C2PA Glossary

C2PA has a precise vocabulary because content provenance separates several different questions: what happened to an asset, who or what made a statement about it, how that statement is protected, how the record is connected to the content, and whether a verifier accepts the signer. This glossary defines the terms readers encounter most often across the C2PA 2.4 specification, Content Credentials interfaces, signing systems, validation reports, and trust infrastructure. [1]

The definitions below are written for practical understanding. They preserve the important distinctions in the specification, but they do not replace its normative requirements. Older C2PA 1.x material and vendor documentation may also use some terms differently, especially around identity and trust.

How to use this glossary

• Read a term in context. A signature can be valid while the signer is untrusted, and a trusted signer can still make an inaccurate statement.
• Treat code-like values such as parentOf, componentOf, and inputTo as exact relationship names, not loose synonyms.
• Use the linked Unfurled guides for the full explanation of major concepts; the glossary entries are intentionally compact.
• For implementation decisions, confirm the exact requirement in the current C2PA specification and the capability of the chosen SDK or product.

Six distinctions to learn first

Alphabetical glossary

A / B / C / D / E / F / G / H / I / J / M / N / O / P / R / S / T / U / V / W / X

A

Action

An operation performed by an actor on an asset, such as creation, cropping, compositing, transcoding, opening, publication, or AI-assisted modification. An actions assertion can record the action type, software agent, parameters, time, digital source type, and relevant ingredient references. The record shows what the signer asserted happened; it is not an independent observation of the event.

Active manifest

The last manifest in a C2PA Manifest Store and the starting point for validation. It represents the current provenance state of the asset. If the active manifest is an update manifest, a validator follows its parentOf chain until it reaches the standard manifest containing the applicable hard binding. Related guide: C2PA Manifests Explained.

Actor

A human or non-human participant in the C2PA ecosystem. An actor may be a person, organization, camera, editing application, cloud service, AI system, repository, or validator. More specific terms such as signer, claim generator, and manifest consumer describe particular technical roles.

AI Disclosure assertion

A standard assertion introduced in C2PA 2.4 for structured information about an AI system and declared levels of human oversight or intervention. It complements action records, digital source types, and ingredients; it does not replace them or function as a universal AI-detection flag. Related guide: C2PA for AI-Generated Content.

Asset

A file or stream of data containing digital content, asset metadata, and optionally a C2PA manifest. For C2PA purposes, an asset can also be cloud-native or dynamically generated rather than a conventional file stored on disk.

Asset metadata

Non-technical information about an asset and its digital content, such as a title, creator field, description, location, or rights information. Metadata that happens to be present in a file is not automatically attributed to the signer. To make a protected statement about it, a claim generator places selected values in an assertion.

Asset rendition

A representation produced through a non-editorial transformation such as resizing, re-encoding, or changing bandwidth. A rendition may preserve the meaning of the source while changing the protected representation, which ordinarily requires a new standard manifest if the rendition is issued with its own valid Content Credential.

Assertion

A labelled data structure representing a statement about an asset. Assertions can describe actions, ingredients, content bindings, metadata, thumbnails, AI disclosures, identity evidence, repository receipts, and many other subjects. The claim references each included assertion by hash so later modification can be detected. Related guide: C2PA Assertions Explained.

Assertion label

The identifier that names an assertion type and, where applicable, its version and instance. Standard labels use the c2pa namespace, while extension authors use their own namespaces. Labels allow claims and validators to locate and interpret the correct assertion structure.

Assertion store

The container inside a C2PA manifest that holds its assertion boxes. An assertion being physically present in the store is not enough: the claim must declare it through a hashed reference. Undeclared or malformed assertions can cause validation failure.

Attestation

Verifiable evidence about properties of a device, application, or execution environment, such as whether approved software or protected hardware produced a result. C2PA publishes a separate attestation specification. Attestation can strengthen confidence in a workflow, but it does not by itself prove factual truth or human authorship. [6]

Authenticity

In C2PA terminology, a property of digital content based on facts that can be cryptographically verified as not having been tampered with. It should not be read as a universal declaration that the content is truthful, accurate, unstaged, or complete. [8] Related guide: What C2PA Proves - and What It Does Not Prove.

B

BMFF-based hash assertion

A hard-binding assertion designed for assets based on the ISO Base Media File Format, including common MP4-family containers. It accounts for the box structure of the file and can support monolithic, fragmented, or piecewise validation patterns more efficiently than treating the asset as an undifferentiated byte sequence.

C

C2PA

The Coalition for Content Provenance and Authenticity and, by common usage, the open technical standard it develops. Context determines whether the name refers to the organization, the specification, or the broader technical ecosystem. The user-facing technology is called Content Credentials. [1] Related guide: What Is C2PA?.

C2PA Manifest

One verifiable provenance unit containing an assertion store, a claim, and a claim signature. The preferred non-technical term is Content Credential. A manifest can also reference earlier or contributing manifests and is stored inside a C2PA Manifest Store. Related guide: C2PA Manifests Explained.

C2PA Manifest Store

The container holding one or more C2PA manifests associated with an asset. It may be embedded in the asset or stored externally. The last manifest in the store is the active manifest, while other manifests can preserve earlier or ingredient provenance. Related guide: C2PA Manifests Explained.

C2PA Trust List

The C2PA-managed list of X.509 trust anchors used to evaluate signing credentials for conforming hardware and software products. A certificate chain reaching a listed anchor can support a trusted result when the remaining validation requirements also succeed. [7] Related guide: C2PA Certificates and Trust Lists.

C2PA TSA Trust List

A separate C2PA-managed list of trust anchors for Time Stamping Authorities. It is intentionally distinct from the signer trust list because claim signing and time-stamp signing are different certificate purposes and require separate trust evaluation. Related guide: C2PA Certificates and Trust Lists.

CBOR

Concise Binary Object Representation, a compact binary data format used by C2PA for claims and many assertions. C2PA also requires deterministic encoding in places where the exact serialized bytes must be hashed or signed consistently.

Certificate authority

An organization or service that issues X.509 certificates and vouches for the binding between a certificate and its subject under a certificate policy. In the current C2PA public trust ecosystem, participating certification authorities issue purpose-limited credentials to products that meet conformance requirements.

Certificate chain

The ordered path from a signer or TSA certificate through any intermediate certification authorities to a trust anchor. A validator checks the chain, certificate purposes, validity periods, constraints, and relevant status evidence before deciding whether the credential is acceptable.

Claim

The digitally signed, tamper-evident structure that identifies the claim generator and contains hashed references to the assertions included in the manifest. A standard claim also includes the information needed to represent the content binding. The claim is the object protected by the claim signature.

Claim generator

The non-human hardware or software actor that constructs the claim and produces the claim signature on behalf of the signer. It may be a camera, creative application, cloud service, mobile app, or signing workflow component. It is not necessarily the human creator or the certificate holder.

Claim generator information

Structured information in the claim that identifies the hardware or software actor that generated it, including a name and optionally version, icon, specification version, or namespaced fields. It describes the generator product; it should not be presented as verified human identity.

Claim signature

The digital signature over the serialized claim, created using the signer's private key. Because the claim contains hashes of the referenced assertions, and one assertion provides the hard binding, a successful signature check helps protect the integrity of the connected provenance structure. Related guide: How C2PA Signing Works.

Composed asset

An asset assembled from multiple pieces of digital content, such as a video built from clips and audio tracks or an advertisement containing a photograph, logo, and generated illustration. The contributing pieces are represented as ingredients, often using componentOf relationships.

componentOf

An ingredient relationship indicating that the ingredient is one component of a composed asset. For example, a logo placed into a poster or an audio track included in a video can be represented as componentOf ingredients.

Conformance Program

The C2PA governance program that evaluates generator products, validator products, and certification authorities against specification, security, and policy requirements. Conforming products are publicly listed, and the program supports the current official trust-list ecosystem. [7]

Content binding

Information associating digital content with a specific C2PA manifest and asset. C2PA recognizes hard bindings, which support exact cryptographic association, and soft bindings, which support identification or recovery after transformation or separation.

Content Credential

The preferred non-technical term for a C2PA manifest. Content Credentials also refers to the overall C2PA-based technology and user experience, so the phrase is commonly treated as plural when discussing the system as a whole. Related guide: What Are Content Credentials?.

COSE

CBOR Object Signing and Encryption, the IETF framework C2PA uses for claim signatures and related protected headers. A COSE signature can carry the signing certificate chain and the algorithm information needed for verification.

crJSON

Content Credentials JSON, a JSON-LD representation derived from validated C2PA data. It is intended for reporting, profile evaluation, and interoperability work. It does not replace the original claim, JUMBF, or COSE structures, is not an authoring format, and is not independently verifiable. [5]

Created assertion

An assertion made by the claim generator and included in the claim's created_assertions list. Created assertions are attributed to the signer under the C2PA trust model. This differs from a gathered assertion, which originated elsewhere in the workflow.

D

Data hash assertion

A hard-binding assertion that identifies which byte ranges of an asset are hashed and which ranges are excluded. It is useful for formats that do not have a more specialized box-based binding and for text formats whose embedded manifest must be excluded from the hash.

Derived asset

An asset created by starting with an existing asset and applying actions that modify its digital content. An edited photograph, shortened audio recording, or document with added pages is a derived asset. The source is commonly represented through a parentOf ingredient relationship.

Digital content

The portion of an asset that represents the content itself - such as image pixels, audio samples, video frames, document text, or code - together with technical metadata needed to interpret it, such as a color profile or encoding parameters.

Digital source type

A controlled vocabulary value describing the type of source associated with an action or asset, including captured, edited, algorithmically generated, or other origins. It provides semantic context but should be interpreted alongside actions, ingredients, AI disclosures, and the signer.

Durable Content Credential

A Content Credential supported by one or more soft bindings that enable its discovery in a manifest repository after the credential becomes separated from the asset. Durability means recoverability, not that metadata can never be removed or that every transformed copy will match unambiguously. [4] Related guide: Durable Content Credentials.

E

Editorial transformation

A transformation that changes the intent, meaning, or both of the digital content. Examples may include removing a person from an image or materially altering the narrative of a clip. The distinction is contextual and is different from a purely non-editorial rendition such as resizing.

Embedded Manifest Store

A C2PA Manifest Store carried inside the asset using a format-specific embedding mechanism. Embedded storage keeps provenance with the file, but downstream tools can strip it, relocate it, or invalidate the applicable hard binding if they transform the content without issuing new provenance.

External Manifest Store

A Manifest Store published separately from the asset and discovered through a direct reference, format-specific mechanism, adjacent file, or other supported method. External storage can preserve original asset bytes and handle large records, but it introduces retrieval, availability, and security dependencies.

F

Fingerprint

A set of inherent properties computed from digital content to identify the content or near duplicates. In a durability workflow, a resolver can use a fingerprint to search for candidate manifests after metadata is removed. Fingerprints can produce approximate or ambiguous matches and are not substitutes for hard-binding validation.

G

Gathered assertion

An assertion supplied to the claim generator by another component or actor and included in the gathered_assertions list. Its integrity is protected by the signed claim, but it is not attributed to the signer in the same way as a created assertion. Human-entered statements and external references are common examples.

General box hash assertion

A hard-binding assertion for non-BMFF formats that organize data into named boxes, chunks, or segments, including several image formats. It records ordered box identifiers and hashes so a validator can verify protected structure while excluding the manifest location where appropriate.

H

Hard binding

A cryptographic binding that associates a standard manifest with a specific asset representation or protected portion of it. A validator recomputes the relevant hashes to determine whether the manifest belongs with the asset and whether covered content changed. A standard manifest contains one hard-binding assertion. Related guide: C2PA Hard Bindings and Soft Bindings.

Hashed URI

A URI reference accompanied by a cryptographic hash, and where needed the hash algorithm, that identifies another structure such as an assertion or manifest. The reference lets a validator resolve the target and confirm that its bytes match the value protected by the claim.

I

Identity assertion

An ecosystem assertion used to express human or organizational identity, roles, and identity credentials separately from the product-oriented signing certificate. C2PA 2.4 recommends CAWG identity specifications for these claims. A verifier should explain exactly which identity was verified and what it is responsible for. [6]

Ingredient

An existing asset or piece of data that contributed to the creation or modification of the current asset. Ingredients can represent parent renditions, components, prompts, models, datasets, or other computational inputs. An ingredient assertion records the relationship and available provenance information.

Ingredient relationship

The field describing how an ingredient relates to the current asset. C2PA 2.4 uses parentOf for derivation, componentOf for composition, and inputTo for computational input. These relationships turn provenance into a graph rather than a single linear history.

inputTo

An ingredient relationship indicating that the ingredient was used as input to a computational process that created or modified the current asset. A prompt, reference image, dataset, or model can be represented this way when the workflow chooses to disclose it.

Invisible watermark

Information embedded into digital content in a substantially human-imperceptible form. It can carry an identifier or reference used as a soft binding to recover a manifest after embedded metadata is lost. Robustness, false matches, transfer attacks, and privacy depend on the watermarking system. Related guide: C2PA vs. Watermarking.

Interim Trust List

The temporary trust list used during early C2PA adoption before the official conformance-governed C2PA Trust List. It was frozen on January 1, 2026: existing entries can still support legacy interpretation, but no new entries or updates are added. [7] Related guide: C2PA Certificates and Trust Lists.

J

JUMBF

JPEG Universal Metadata Box Format, a general box-based container used by C2PA to serialize Manifest Stores, manifests, assertion stores, claims, signatures, and related data. The name includes JPEG, but JUMBF structures are used by C2PA across multiple asset formats.

M

Manifest consumer

An actor that obtains provenance data from an asset-associated C2PA manifest. A validator is a specialized manifest consumer that performs the technical validation process; a viewer may consume and present the results without implementing every validation function itself.

Manifest repository

A service that stores C2PA manifests or Manifest Stores and can be searched using content bindings. Repositories support external storage and durable recovery, but repository availability or a successful lookup does not by itself make a manifest valid or trusted.

Metadata assertion

A standard assertion for selected metadata that the claim generator chooses to include in the signed provenance record. It can carry values based on supported schemas such as XMP-related fields. This is different from unasserted metadata merely present elsewhere in the asset.

Multiple-step signing

A signing method used when the final hard binding cannot be calculated until the output layout and manifest placement are known. The workflow reserves space, finalizes the asset structure, calculates the binding, signs the deterministic claim, and inserts the completed manifest without changing protected content. Related guide: C2PA Implementation Guide.

N

Non-editorial transformation

A transformation that changes the representation of digital content without intending to alter its meaning, such as scaling, re-encoding, or transcoding. The new representation may still need a new standard manifest because its protected bytes or format structure differ from the source.

Normative language

The requirement words used by a specification - including MUST, SHALL, SHOULD, MAY, and their negative forms - with meanings defined by IETF standards. Normative clauses determine conformance. Explanatory examples, notes, and implementation recommendations do not automatically carry the same force.

O

OCSP

Online Certificate Status Protocol, a method for checking whether an X.509 certificate has been revoked. C2PA validation can use OCSP evidence obtained during verification or preserved through certificate-status information. A successful signature check does not override a revoked signing credential.

P

parentOf

An ingredient relationship indicating that the current asset is derived from, or is a rendition of, the ingredient. It is used for ordinary edit histories and is also required in a restricted form by update manifests to point to the manifest they extend.

Private key

The secret half of a cryptographic key pair used to create claim signatures. Anyone who gains unauthorized access to a production signing key may be able to issue apparently legitimate Content Credentials, so key custody, authorization, isolation, rotation, and incident response are central security concerns.

Private credential store

A validator-managed address book of individual signer credentials that a user has explicitly chosen to trust through an out-of-band relationship. Entries are trusted directly; they do not act as certificate authorities and are not used for Time Stamping Authority validation.

Provenance

The logical history of an asset and its interactions with actors and other assets, as represented by provenance data. Provenance can describe creation, edits, components, inputs, publication, and other recorded events. It may be verifiable without being complete.

Provenance data

The set of C2PA manifests associated with an asset and, for a composed asset, the relevant ingredient provenance. It is the recorded data from which a consumer reconstructs the asset's available history.

R

Redaction

The controlled removal of an assertion from inherited ingredient provenance, declared by a later claim rather than silently rewriting the earlier signed record. Redaction can support privacy or policy needs, but it also creates an explicit gap that a verifier should present accurately.

Region of interest

A defined spatial, temporal, textual, or other portion of an asset to which an assertion applies. A region can identify an area of an image, a range of video frames, an audio segment, or another localized scope rather than making the assertion about the entire asset.

Repository receipt

A C2PA 2.4 assertion carrying evidence that a manifest repository ingested a Manifest Store. It does not prove public publication, permanent availability, or the truth of the manifest's assertions; its claim is narrower: the repository acknowledged receipt. [4]

Revocation

The invalidation of a certificate before its scheduled expiration, commonly because a private key was compromised, a product lost authorization, or the certificate should no longer be trusted. Validators use status evidence such as OCSP to identify revoked signing or time-stamping credentials.

S

Signer

The credential holder whose private key is used to sign the claim. The signer is identified by the subject of the signing credential. It may represent a product, service, device, or organization and should not automatically be presented as the human creator. Related guide: C2PA Certificates and Trust Lists.

Signing credential

The X.509 certificate and associated private key authorized for C2PA claim signing. A validator checks the certificate chain, purpose, validity period, status, and applicable trust configuration separately from checking the mathematical claim signature. Related guide: C2PA Certificates and Trust Lists.

Soft binding

A content-derived identifier, such as a fingerprint or invisible watermark, used to identify related content or recover a candidate manifest after separation. Soft bindings can survive some transformations that break exact hashes, but they can be ambiguous and must not be used as hard bindings. [4] Related guide: C2PA Hard Bindings and Soft Bindings.

Standard assertion

An assertion type defined by the C2PA specification for interoperable use, such as actions, ingredients, metadata, content bindings, thumbnails, AI disclosure, or repository receipts. C2PA also permits namespaced custom assertions, which require their own definitions and handling.

Standard manifest

A manifest issued for a particular asset representation. It contains a hard-binding assertion connecting the provenance record to the protected content. When covered content changes, a new standard manifest is generally required rather than an update manifest. Related guide: C2PA Manifests Explained.

Status code

A structured success, informational, or failure result produced during C2PA validation. Status codes identify specific findings - such as a validated signature, hash mismatch, malformed assertion, untrusted credential, or inaccessible resource - and should not be collapsed into one generic valid or invalid badge.

T

Time Stamping Authority

A service that issues trusted time-stamp tokens under a certificate specifically authorized for time stamping. A trusted time-stamp can help show that a claim signature existed while the signing credential was valid, supporting long-term validation after the certificate later expires.

Trusted manifest

A manifest that is valid and whose signing credential is accepted by the validator's trust configuration. Trusted is therefore a stronger technical state than valid, but it still does not mean that the signer is honest or that every assertion is factually true. [3] Related guide: How C2PA Verification Works.

Trusted time-stamp

A time-stamp whose signature, certificate chain, purpose, and trust anchor validate under the validator's TSA trust configuration. It provides evidence about when the claim signature existed, not when the depicted event occurred unless other trustworthy evidence connects those facts.

Trust anchor

An X.509 certificate treated as the root of an accepted certificate path. Trust anchors can come from the official C2PA Trust List, the C2PA TSA Trust List, another configured list, or a private organizational policy. Trust is always evaluated for a particular purpose and context.

Trust configuration

The set of trust anchors, accepted certificate purposes, date constraints, private credentials, and policy choices a validator uses when evaluating a signer or TSA. The same cryptographically valid manifest can be trusted in one configuration and untrusted in another.

Trust signal

Information that helps a manifest consumer judge the trustworthiness or relevance of an asset, beyond the basic signer relationship. Examples can include trusted time, identity assertions, attestations, ingredient provenance, capture information, conformance status, and workflow-specific evidence.

U

Update manifest

A restricted manifest that adds signed provenance information without changing the digital content or other data covered by the existing binding. It contains no new hard binding and points through exactly one parentOf ingredient to the manifest being updated. Only permitted assertion types and non-content-changing actions may be included. Related guide: C2PA Manifests Explained.

V

Valid manifest

A well-formed manifest that has not been modified since signing and whose claim signature, signing-time conditions, and revocation checks satisfy the C2PA validity requirements. Validity does not require the signer to chain to a trust anchor accepted by the validator; that additional step produces a trusted result. [3] Related guide: How C2PA Verification Works.

Validation

The defined process for locating the Manifest Store, selecting the active manifest, parsing the claim, resolving assertions, checking hashes and content bindings, validating the signature and credential, processing ingredients, evaluating time and revocation evidence, and returning structured results. [2] Related guide: How C2PA Verification Works.

Validator

A manifest consumer that performs the C2PA validation process. A production validator must handle untrusted media and remote resources safely, preserve detailed status information, apply a clear trust configuration, and avoid presenting cryptographic success as a universal truth judgment.

Visible watermark

A perceptible mark placed in the digital content that communicates information to people, such as a creator name, source, or disclosure. It can remain useful when metadata is removed, but it has limited structured capacity and is not itself a substitute for a signed Content Credential. Related guide: C2PA vs. Watermarking.

W

Well-formed manifest

A manifest whose structure, allowed assertion set, assertion requirements, and ingredient rules satisfy the normative checks defined by C2PA validation. A well-formed manifest can still fail validity because of a broken signature, invalid timing, revocation, or another cryptographic problem. [3] Related guide: How C2PA Verification Works.

X

X.509 certificate

A standard public-key certificate used by C2PA for claim signing, time-stamp signing, and certificate-status signing. The certificate identifies its subject and authorized purpose, carries the public key, and participates in a chain to a configured trust anchor.

XMP

Extensible Metadata Platform, a widely used metadata framework that can carry descriptive information and, in some formats, a reference used to discover an external C2PA Manifest Store. XMP metadata is not automatically protected or attributed merely because it is present; relevant values must be covered or asserted appropriately.

The simplest mental model

• The asset is the content being described.
• Assertions are the individual provenance statements.
• The claim identifies and hashes the assertions.
• The claim signature protects the claim and associates it with a signing credential.
• The hard binding connects a standard manifest to the protected asset representation.
• Soft bindings help identify or recover related content after separation or transformation.
• The Manifest Store carries the active and related manifests.
• Validation checks structure, integrity, credentials, bindings, and relationships.
• Trust configuration determines whether a valid signer is accepted.
• People and organizations still decide what the verified evidence means in context.

A reliable C2PA implementation keeps these layers separate. The more a product collapses them into one word such as authentic, real, or verified, the more likely it is to mislead the reader about what the technology actually established.

Sources and notes

[1] C2PA Technical Specification 2.4, especially "Technical Overview" and "Glossary". Core terminology for actors, assets, manifests, Content Credentials, bindings, provenance, trust lists, and durability.

[2] C2PA Technical Specification 2.4, sections on assertions, claims, manifest structure, validation, and standard assertions. Normative relationships among claims, signatures, assertions, bindings, ingredients, and validation results.

[3] C2PA Technical Specification 2.4, "Validation states" and "Trust Lists". The hierarchy in which Trusted implies Valid and Valid implies Well-Formed, plus public and private trust configuration.

[4] C2PA Soft Binding Resolution API and the core specification's binding sections. Durable recovery, fingerprints, invisible watermarks, repositories, and repository receipts.

[5] Content Credentials JSON (crJSON) File Format Specification. crJSON as a derived JSON-LD view for reporting and interoperability rather than an independently verifiable authoring format.

[6] C2PA Human and Organizational Identity Recommendation and the C2PA 2.4 specification collection. Identity assertions and attestations as separate evidence layers beyond the product signing certificate.

[7] C2PA Conformance Program. Current governance for conforming products, certification authorities, the C2PA Trust List, the TSA Trust List, and the frozen Interim Trust List.

[8] C2PA Explainer 2.4, Security Considerations, and Harms Modelling. Limits of authenticity language, privacy, security, presentation, and the warning against treating provenance as a universal truth verdict.