
Dispatch 03 / Written from 2035
Scenario confidence: high · Evidence cutoff: 12 August 2026
Provenance Is the New Publishing Layer
By 2035, a media file without inspectable lineage will feel as operationally incomplete as a page without a URL.
AI-generated conceptual scenario art, not documentary or forecast evidence.
The thesis
Provenance will mature from a defensive label into a publishing substrate. The valuable system will not claim to certify truth; it will preserve verifiable statements about origin, edits, ingredients, authority, and delivery so that people and machines can make better trust decisions across an asset’s life.
Evidence / 2026
Signals visible from here.
C2PA 2.2 specifies signed manifests, content bindings, ingredients, actions, and trust mechanisms across common media formats. Its own guidance carefully limits the claim: valid provenance indicates that assertions are bound and untampered, not that the depicted event is true.
The EU AI Act establishes disclosure duties for defined classes of generated or manipulated media while recognizing artistic and editorial contexts. The U.S. Copyright Office separately emphasizes identifiable human expression and warns about unauthorized digital replicas. Together these are signals that origin and responsibility are becoming part of media operations.
Many current workflows still strip metadata at export, sever derivatives from sources, and ask a final publisher to reconstruct lineage after dozens of transformations. That architecture cannot scale to abundant synthetic and hybrid media.
Inference / 2035
A future history.
Capture devices, creative tools, models, review systems, and distribution platforms contribute to a chain of signed assertions. A viewer can inspect a useful summary; a newsroom can validate a deeper chain; a rights holder can locate transformations; a governed collection can preserve the package for later interpretation.
Provenance becomes compositional. Directors choose which process details improve the audience’s understanding, artists can disclose synthetic intervention without flattening the experience into a warning label, and documentary teams can protect sensitive sources while still establishing institutional accountability.
The absence of credentials remains possible and sometimes necessary. Rather than treating unsigned work as automatically false, systems communicate what is known, unknown, redacted, broken, or unverifiable. Trust design becomes an editorial discipline of calibrated claims.
Interpretation
What changes for leaders and makers.
- Media supply chains need identity, signing, preservation, and verification capabilities at every handoff. Adding a badge at publication will not repair lost lineage.
- Organizations that preserve provenance can license, localize, retract, audit, and reuse work with lower ambiguity. The return is operational as much as reputational.
- Creative teams must help design disclosure. If provenance interfaces are built only by compliance teams, they may be technically correct and culturally unreadable.
Recommendation / Now
Moves to make before the future hardens.
- 01Map where metadata, identity, and source relationships disappear from capture through distribution; prioritize the first irreversible break.
- 02Pilot C2PA signing and validation on one bounded workflow, including derivatives, redactions, offline transfer, and a deliberately broken credential.
- 03Write a provenance language guide that distinguishes verified origin, declared process, editorial review, identity, and truth.
- 04Store original assets and manifests independently of any single vendor, and rehearse key rotation, revocation, and long-term validation.
Open questions
What could change this conclusion.
Platform support may remain uneven, credentials can be removed, signers can misrepresent events, and trust lists can reproduce institutional power. Provenance is evidence infrastructure, not an oracle.
Disclosure rules will vary across jurisdictions and genres. Privacy-preserving redaction and protection for vulnerable creators must develop alongside greater transparency.
Primary sources
The evidence beneath the scenario.
- C2PA Technical Specification 2.2Coalition for Content Provenance and Authenticity · Version 2.2
Defines signed, tamper-evident Content Credentials that bind provenance assertions to media while preserving the distinction between provenance and truth.
- Regulation (EU) 2024/1689: Artificial Intelligence ActEuropean Union · 2024-07
Establishes transparency duties for certain AI-generated or manipulated content, including deep fakes, with provisions for artistic works and editorial responsibility.
- Copyright and Artificial Intelligence, Part 2: CopyrightabilityUnited States Copyright Office · 2025-01
Concludes that generative-AI material can participate in copyrightable work when a human determines sufficient expressive elements, while prompts alone generally do not supply that authorship.
- Copyright and Artificial Intelligence, Part 1: Digital ReplicasUnited States Copyright Office · 2024-07
Finds gaps in protections against unauthorized digital replicas and recommends a federal right covering all individuals.