How we protect regulatory text across jurisdictions.
Every jurisdiction we support asserts copyright over its regulatory publications. Seif respects this. The MCP server never exposes raw regulatory text — it delivers structured compliance intelligence, with every response traceable to its source.
Regulatory copyright and data rights
Regulatory texts across every jurisdiction we support — rulebooks, guidance notes, and circulars — are published under the copyright of the relevant authority. Firms can read them. They cannot redistribute them. Any platform that serves regulatory content must respect this.
For Seif, this is not a limitation. It is a design principle. Raw regulatory text is not particularly useful to an AI agent. What is useful is structured, interpreted intelligence: the specific obligations that apply to a firm, the definitions that govern those obligations, the cross-references that connect them, and the confidence-scored analysis that makes them actionable.
The MCP server is designed from the ground up to deliver intelligence, not text. Every response is a structured object with typed fields, citation references, and metadata — never a block of copied regulatory prose.
Transformed compliance intelligence
- Structured obligations
Typed obligation objects with deontic modality, applicable firm types, conditions, and cross-references.
- Regulatory definitions
Precise definitions with scope, source reference, and related terms.
- Citations by reference
Rule numbers and section identifiers that let you locate the source, not verbatim copies of the source.
- LLM-synthesised answers
Readable compliance answers grounded in graph-retrieved knowledge, not generated from scratch.
- Confidence scores
Numerical confidence with answer tier classification: DEFINITIVE, INTERPRETIVE, or REQUIRES_REVIEW.
- Reasoning chains
Step-by-step inference paths showing which graph nodes were traversed and which retrieval strategies were used.
Raw regulatory text
- Verbatim regulatory text
No paragraph-level reproduction of regulatory publications.
- Raw paragraph content
No unstructured blocks of regulatory prose in any response.
- Exact quotes exceeding 15 words
Short phrases for precision; never extended passages.
- Bulk text extraction
No tool or endpoint that returns regulatory text at scale.
- Document downloads
No PDF, DOCX, or other file format exports of source regulatory documents.
Every response, fully annotated
Every MCP response is wrapped in a metadata envelope. The envelope declares what the response contains, how confident the system is, where the information came from, and whether any raw text is present. AI agents can read this metadata to decide how to present the answer — and compliance officers can audit it.
The data_rights block is present on every response. It
is a machine-readable declaration that the response respects the
applicable jurisdiction's copyright constraints. The audit block chains each
response to its predecessors, creating a tamper-evident log.
{
"content": "Your compliance answer...",
"metadata": {
"tier": "professional",
"confidence": 0.94,
"answer_type": "INTERPRETIVE",
"sources": [
"COBS 3.2.1",
"GEN 3.1.4"
],
"data_rights": {
"contains_raw_text": false,
"citation_style": "reference_only",
"license": "seif_intelligence_license"
},
"audit": {
"query_hash": "sha256:a1b2c3...",
"chain_hash": "sha256:d4e5f6..."
}
}
}Licensed full-text access
Some firms require access to the underlying regulatory text for their internal compliance workflows. Seif is working with regulatory authorities across supported jurisdictions to establish licensed data agreements that allow qualifying firms to access full-text regulatory content through the Seif platform and MCP server.
Licensed full-text access will be available as an Enterprise add-on. It will carry the same explainability envelope, audit trail, and access controls as structured intelligence responses — with the additional data right to include verbatim regulatory text where the license permits.
If your firm requires full-text access for a specific compliance use case, contact us to discuss your requirements.
Hash-chained audit trail
Every MCP query is logged with the same hash-chained audit trail
used by the Seif platform UI. Each response includes a
query_hash (SHA-256 of the input) and a
chain_hash (SHA-256 of the current response chained to
the previous hash). This creates an append-only, tamper-evident log.
If any record in the chain is modified after the fact, the chain breaks — making tampering detectable. The audit log is exportable at any time, in both machine-readable (JSON) and human-readable (PDF) formats.
For regulated firms, this means every AI-assisted compliance decision that passes through Seif is documented, traceable, and defensible. When a regulator asks how a conclusion was reached, the audit trail provides the answer.
Trust is the product.
See the explainability envelope, audit trail, and data rights in action. Book a demo and we'll walk through a live compliance query.