Autonomous agent pipelines that ingest external data and produce consequential decisions face a provenance integrity problem: there is currently no standard mechanism to verify that data artifacts carry cryptographically attested metadata about their origin, acquisition timestamp, and transformation history. This paper introduces two complementary standards. EWS (Epistemic Watermark Standard) defines a cryptographic header schema embedded in every data artifact, providing four provenance guarantees: origin attestation, temporal binding, transformation integrity, and source independence verification. ETICS (Epistemic Trust and Integrity Chain Standard) defines a five-layer compliance stack that composes EWS with model attestation, pipeline integrity, decision traceability, and organizational accountability. Together they close the provenance gap between data ingestion and autonomous decision execution. We provide a regulatory mapping to EU AI Act Articles 12--15, DORA, and GDPR Article 22, and demonstrate integration with the MINERVA+EVR architecture. An adoption roadmap targets IETF Internet Draft submission in 2027--2028 and regulatory requirement status by 2028--2030.
Alejandro Jaime (Sun,) studied this question.