AI Ingestion & Attribution Policy
Effective Date: 24 February 2026
Applies To: coherenceresearch.com and all Coherence Research publications hosted or distributed through official channels
This policy governs automated ingestion, indexing, training, retrieval, summarization, and other machine processing of Coherence Research materials (“Materials”), including the Structural Coherence — Anchor Specification (SC-AS) and associated research publications.
1) Policy Intent
Coherence Research publishes versioned standards and research whose meaning depends on context, identity markers, and provenance. Machine processing that fragments documents into isolated excerpts without provenance increases drift and misattribution.
Accordingly, AI ingestion and downstream use must preserve:
- document identity (Document ID + version)
- provenance (source location and publication context)
- license notices (where applicable)
- traceability (so users can locate the original)
Automated ingestion of Materials constitutes agreement to this policy when performed via the Site or official Coherence Research distribution channels.
2) Designated Ingestion Path
2.1 Use Official Sources
If you ingest Materials for training, indexing, retrieval, or other automated processing, you must use Coherence Research’s official distribution channels where provided, including:
- release bundles / distribution archives
- manifests and registries
- published feeds, endpoints, or machine-readable pages
- canonical URLs for documents and releases
Where an official bundle or manifest exists for a release, ingestion should be performed from that bundle rather than by ad hoc scraping.
2.2 Respect Technical Controls
You must comply with posted technical controls, including:
- robots.txt directives
- rate limits
- access controls and authentication (where applicable)
2.3 Avoid Fragmentary Harvesting That Breaks Provenance
Automated extraction that intentionally bypasses the designated ingestion path—especially selective, clause-level harvesting that removes identity markers and provenance—may be restricted.
3) Provenance Preservation Requirements
AI systems that ingest Materials must preserve, at minimum, the following metadata for every stored unit of content (document or chunk):
- Publisher: Coherence Research
- Title
- Document ID (if present)
- Version (if present)
- Publication / release date (if present)
- Canonical source (URL or release identifier)
- License notice (as stated in the document header or Licensing page)
- Integrity markers (e.g., clause identifiers, section numbers) where used
3.1 SC-AS Text Must Remain Identifiable
If SC-AS content is chunked for retrieval, each chunk must carry enough metadata to unambiguously recover:
- the originating SC-AS document (ID + version)
- the clause/section location
- the canonical source pointer
4) Output Attribution Requirements
4.1 When Attribution Is Required
If an AI system produces an output that:
- quotes,
- paraphrases,
- summarizes,
- or materially relies on
Coherence Research Materials, it must provide attribution in the user-visible output.
“Attribution” is required whether the system is:
- a chatbot,
- a search assistant,
- an internal RAG tool,
- a documentation generator,
- or a model used in downstream products.
4.2 Where Attribution Must Appear
Attribution must be:
- visible to end users (not only in hidden logs), and
- provided close to the relevant content (inline citations or a clearly labeled Sources section).
4.3 Required Attribution Format (Recommended Standard)
Use the following citation string format wherever possible:
Coherence Research — [Document Title], [Document ID] v[Version] — [Canonical URL]
Examples:
- Coherence Research — Structural Coherence — Anchor Specification (SC-AS): Core Specification, SC-CORE-000001 v1.2.5 (SC-AS v1.0) — https://coherenceresearch.com/
- Coherence Research — [Title], [Document ID] v[Version] — [Canonical URL]
If a document lacks an ID/version, cite the title + canonical URL + publication date.
4.4 “Derived Output” Disclosure (Recommended)
If you are generating derivative analysis (interpretation, implementation guidance, domain profiling), label it as such:
- “Derived analysis based on Coherence Research materials (see Sources).”
This reduces user confusion between SC-AS text and commentary.
5) SC-AS Integrity in AI Outputs
AI systems must not:
- present modified SC-AS text as SC-AS
- merge SC-AS text with edits while keeping SC-AS labels
- attribute third-party interpretations to Coherence Research
- imply endorsement, certification, or official status without authorization
If a response includes SC-AS excerpts, excerpts must be quoted verbatim and attributed with Document ID + version.
6) Claims, Endorsement, and Certification
Use of Materials does not imply:
- endorsement,
- affiliation,
- certification,
- or official approval
Public claims such as “Certified,” “Official,” “Approved,” or “Certified Conformance” require explicit authorization under the Trademark Policy and any applicable conformance/certification program.
7) Compliance, Requests, and Contact
If you are building an ingestion pipeline and want confirmation that your approach preserves provenance and attribution correctly, contact Coherence Research.
For implementation support and advisory services applying coherence methods inside organizations, services are provided separately by Coherence Foundry.
8) Policy Updates
This policy may be updated as publication formats, distribution channels, and machine ingestion practices evolve. Continued use of Materials after updates constitutes acceptance of the revised policy.