Capture SDK research
OpenARI Capture is a separate research repository for capture signing on independent cameras and embedded devices. The plan starts with a Raspberry Pi experiment and defines an integration path for camera manufacturers.
Current status: architecture-only scaffold. No capture, signing, verification, hardware attestation, or C2PA implementation is available. No SDK packages are published. The project cannot issue Apple ARI attestations.
Planned boundaries
The capture SDK will build evidence and use optional signing backends. A separate capture verifier will evaluate that evidence offline with explicit trust policy, without requiring camera drivers or private keys. Future CLI integration will use optional commands in the existing openari application.
Independent capture profiles, version identifiers, and private/manufacturer trust roots remain separate from Apple ARI. A self-signed credential only becomes trusted through explicit enrollment; it is not trusted because an image includes it.
What a signature establishes
A signature binds a key to bytes and assertions. Software can substitute pixels before signing. A secure element can protect a key while still signing bytes chosen by compromised host software. Sensor-origin claims therefore require evidence about the protected capture path, not just a valid signature.
The first proposed Pi example will be labeled software-asserted capture. Stronger manufacturer profiles require reviewed hardware and firmware assumptions, provisioning, revocation, and adversarial tests. Sensor origin alone cannot prove the truth of a scene or rule out photographing a screen.