What Encypher does: Insights from the Co-Chair of the C2PA Text Provenance Task Force on AI content authentication, content attribution, and licensing infrastructure. Standard publishes January 8, 2026.
Who it's for: Publishers seeking licensing strategies, AI labs exploring compliance, legal professionals interested in content attribution, and developers building with our API and SDKs.
Key differentiator: Written by the team co-chairing C2PA (c2pa.org) with NYT, BBC, AP, Google, OpenAI, Adobe, Microsoft and others - insider perspective on standards development.
Primary value: Stay informed on market licensing frameworks, regulatory developments, and technical innovations in cryptographic watermarking.
Authors of C2PA Section A.7 (Text Provenance)Authors of C2PA Section A.7 (Text Provenance): Building Infrastructure for the AI Content Economy.
EU AI Act Article 50 compliance requires two marking layers minimum. Organizations building metadata-only or watermark-only solutions face a gap the Code of Practice makes clear.
AI licensing revenue now flows through marketplaces requiring machine-readable provenance. Publishers without this infrastructure are invisible to the systems that would pay them.
Machine-readable opt-outs are now the only valid AI training reservation. A German appellate ruling and EU enforcement confirm plain-text copyright notices protect nothing.
AI copyright settlements pay per book while AI systems extract value per sentence. Without sentence-level provenance, every future settlement will underpay rightsholders.
Robots.txt has no legal standing as a rights reservation mechanism. A federal ruling, EU consultations, and empirical data show publishers need in-content credentials.
RAG copyright liability exceeds training risk. The US Copyright Office, two federal rulings, and active litigation confirm retrieval creates fresh infringement at every query.