What does it mean to verify an AI transparency signal? A pre-platform comparison of PTC/FRS and C2PA 2.4

This record preserves the research abstract “What does it mean to verify an AI transparency signal? A pre-platform comparison of PTC/FRS and C2PA 2.4”, prepared for submission to the ECAT Research Workshop 2026. The contribution presents a structured normative and architectural comparison of the Canal Transparency Protocol (PTC), complemented by FRS-01 for exact-file reference, and C2PA 2.4 at the pre-platform boundary. It examines “verification” through five analytical dimensions: semantic disclosure; object correspondence and integrity; responsibility and attribution; identity and trust; and provenance continuity. The record also preserves the exact PDF research artefact together with its PTC/FRS declaration and verification materials. The accompanying evidence enables a third party to reproduce the correspondence between the declaration and the exact bytes of the referenced PDF and to inspect the declared human–AI production process. It does not establish authorship, operator identity, material truth of the declaration, or correctness of the paper’s claims. A browser-based reference implementation capable of verifying the accompanying PTC/FRS evidence is publicly available at validator.es. The inclusion of the PTC/FRS evidence constitutes a reflexive instantiation of one of the architectures analysed in the paper: the research artefact itself exposes both the properties that can be verified and the limits of those properties.