Dekimu · anchored receipts docs

APR — Anchored Provenance Receipts

Cryptographic proof of what an AI agent produced — who issued it, when, and what artifact was anchored.

Purpose

An APR anchors the output of an AI agent (or human, or hybrid process) to a tamper-evident receipt. Verifiers can confirm the artifact hash, the issuing agent identity, and the daily audit root without accessing the original artifact.

APR uses a single schema discriminator ar.provenance.v1 — the event type is implicit (provenance is always a snapshot, not a lifecycle).

Artifact types

The body.artifact_type field classifies what was anchored:

textcodeimagedocumentstructuredbinaryemailmessagesite_snapshot

Agent kinds

The body.agent_kind field identifies who or what produced the artifact:

aihumanhybridtemplate

Key fields

artifact_hash — SHA-256 digest of the anchored artifact, hex-encoded. Enables artifact re-verification without storing the original content.

disclosure — disclosure policy applied at issuance time. Controls which fields are visible to verifiers without holder involvement.

Regulatory context

APR receipts provide evidence of transparency obligations under EU AI Act Art. 52, which requires that AI-generated content be disclosed as such to recipients. An APR receipt allows any downstream verifier to confirm the AI provenance of an artifact independently of the issuing system.