Decodes AI/ML provenance assertions off a C2PA manifest's assertion array: every entry in every c2pa.actions/c2pa.actions.v2 assertion (a manifest can carry a created→edited→published history, so this reads the whole chain, not just the first action), the full IPTC digitalsourcetype vocabulary (a raw code is always surfaced, even one this tool doesn't recognize), and c2pa.ai_training/c2pa.ai_generative_training training-and-data-mining opt-out assertions. Generalizes art-361's single-action, IDV-scoped read into a general-purpose decoder. Client-side, zero-egress; emits a hash-anchored OpenChainGraph v0.4 artifact.
Reports assertions only: this tool never adjudicates whether an assertion is true. Absence of an AI/ML assertion means nothing: it is never evidence of human authorship, and its presence is never proof of AI generation. Not a trust-chain or signature claim (see art-123/art-124/art-125 for that).
🔒 All inputs are processed locally in your browser. No data is transmitted. Do not enter real personal data — use synthetic or anonymised inputs only.