Crate c2pa_ml 

C2PA manifest embedding for AI/ML model container formats.

A C2PA Manifest Store is associated with a model by writing it into the format’s own metadata slot, so the model stays loadable by its usual runtime:

• GGUF (llama.cpp) — a c2pa.manifest key/value metadata entry (gguf).
• SafeTensors — a c2pa.manifest entry in the JSON header’s __metadata__ (safetensors).
• ONNX — a c2pa.manifest entry in the protobuf metadata_props (onnx).

A remote (or side-car) manifest can instead be referenced by URI under c2pa.manifest.uri, or both an embedded store and a URI can be written together (see ManifestSource).

The top-level functions auto-detect the format:

use c2pa_ml::{embed_manifest, read_manifest, ManifestSource};
let signed = embed_manifest(model, &ManifestSource::embedded(store))?;
let manifest = read_manifest(&signed)?;

Asset type

The C2PA core specification defines no dedicated embedding method for model containers; a manifest embedded in a model declares what the asset is with the asset type assertion. Format::model_type and asset_type::ModelType provide the canonical c2pa.types.model.* strings for that assertion.

Scope

This crate implements embedding and extraction only. Manifest construction, signing, and content (hard/soft) binding are out of scope; use the official C2PA SDK to build and sign manifests. The c2pa.hash.data assertion should exclude the metadata region carrying the Manifest Store.

Zero dependencies on native targets; the WebAssembly/npm build uses only wasm-bindgen.

Re-exports

pub use asset_type::ModelType;

Modules

asset_type

The C2PA asset type values for AI/ML assets, from Table 11 (“Asset type values”) of the C2PA Technical Specification.

gguf

GGUF (the llama.cpp model container) embedding.

onnx

ONNX embedding.

safetensors

SafeTensors embedding.

Structs

ManifestSource

What to associate with a model: an embedded Manifest Store, a remote manifest URI, or both.

Report

A report on a model’s C2PA embedding.

Enums

Error

Errors returned by the crate.

Format

A supported ML model container format.

Functions

embed_manifest

Embed a C2PA manifest into a model, auto-detecting the container format.

embed_manifest_as

Embed a C2PA manifest into a model of a known format.

read_manifest

Read the embedded C2PA Manifest Store from a model, auto-detecting the format.

read_manifest_uri

Read the remote manifest URI from a model, if present, auto-detecting the format.

remove_manifest

Remove any C2PA metadata from a model, auto-detecting the format.

verify

Report on a model’s C2PA embedding, auto-detecting the format.