Relational Verification for AI Indexing is a standards-track proposal by Lee Sharks for representing the evidentiary relationships surrounding identity and authorship claims.
The proposal begins from a boundary in existing provenance technology. A valid signature can prove that a key holder issued a packet, while a verifiable credential can prove that declared constraints were preserved. Neither mechanism alone proves that the packet’s description of a person, institution, history, or authorship is correct. The draft calls this the ontological-claimant problem.
Its response is not a new truth tribunal. Instead, it adapts the MPAI grammar so that claims expose their relation to evidence, neighboring entities, independent citation, revision history, and first-party assertion. Three implementation routes are proposed: a custom JSON-LD vocabulary, an external Schema.org extension, and an OAI-PMH profile.
The draft’s core principle is that protocols should expose rather than adjudicate. Human communities remain responsible for interpretation and dispute. Version 1.1 is explicitly a pre-RFC working draft, so its namespaces, classes, terms, and adoption timelines are proposals rather than established standards.