This v1.1 bridge packet maps three independent dimensions of AI-era provenance: artifact authenticity, training-corpus licensing, and semantic lineage through synthesis. It treats C2PA and Content Credentials as artifact-level infrastructure; dataset audits, W3C PROV, and legal transparency regimes as corpus-level infrastructure; and semantic provenance as the still-underdeveloped question of whether synthesized meaning remains accountable to authors, frameworks, traditions, and communities. The packet presents semantic provenance as an extension rather than a replacement or criticism of existing fields. It acknowledges archival provenance, digital preservation, Indigenous data sovereignty, citation evaluation, RAG faithfulness, and data attribution as adjacent foundations. PER-M, PER-C, and PER-D are proposed as progressively deeper measurement variants. The body explicitly removes earlier unsupported production-system estimates and calls the metric provisional pending pilot studies and inter-rater reliability. Legal, standards, adoption, and institutional claims reflect a 2026 snapshot and require current primary-source verification.
Full packet — deposit #725 (canonical record) Plain text
A Metadata Packet for AI Indexing catalogued at metadatapacket.dev. The specification is deposit #656; the catalog as data is catalog.json.