Skip to content

Article

Authority must survive extraction

The real test of authority is not whether it is visible on the source page, but whether it remains attached to a statement once AI systems extract and reuse it.

CollectionArticle
TypeArticle
Categoryarchitecture semantique
Published2026-04-28
Updated2026-04-28
Reading time7 min

Governance artifacts

Governance files brought into scope by this page

This page is anchored to published surfaces that declare identity, precedence, limits, and the corpus reading conditions. Their order below gives the recommended reading sequence.

  1. 01Definitions canon
  2. 02Citations
  3. 03Interpretation policy
Canon and identity#01

Definitions canon

/canon.md

Canonical surface that fixes identity, roles, negations, and divergence rules.

Governs
Public identity, roles, and attributes that must not drift.
Bounds
Extrapolations, entity collisions, and abusive requalification.

Does not guarantee: A canonical surface reduces ambiguity; it does not guarantee faithful restitution on its own.

Policy and legitimacy#02

Citations

/citations.md

Surface that makes explicit the conditions of response, restraint, escalation, or non-response.

Governs
Response legitimacy and the constraints that modulate its form.
Bounds
Plausible but inadmissible responses, or unjustified scope extensions.

Does not guarantee: This layer bounds legitimate responses; it is not proof of runtime activation.

Policy and legitimacy#03

Interpretation policy

/.well-known/interpretation-policy.json

Published policy that explains interpretation, scope, and restraint constraints.

Governs
Response legitimacy and the constraints that modulate its form.
Bounds
Plausible but inadmissible responses, or unjustified scope extensions.

Does not guarantee: This layer bounds legitimate responses; it is not proof of runtime activation.

Evidence layer

Probative surfaces brought into scope by this page

This page does more than point to governance files. It is also anchored to surfaces that make observation, traceability, fidelity, and audit more reconstructible. Their order below makes the minimal evidence chain explicit.

  1. 01
    Canon and scopeDefinitions canon
  2. 02
    Response authorizationQ-Layer: response legitimacy
  3. 03
    External contextCitations
Canonical foundation#01

Definitions canon

/canon.md

Opposable base for identity, scope, roles, and negations that must survive synthesis.

Makes provable
The reference corpus against which fidelity can be evaluated.
Does not prove
Neither that a system already consults it nor that an observed response stays faithful to it.
Use when
Before any observation, test, audit, or correction.
Legitimacy layer#02

Q-Layer: response legitimacy

/response-legitimacy.md

Surface that explains when to answer, when to suspend, and when to switch to legitimate non-response.

Makes provable
The legitimacy regime to apply before treating an output as receivable.
Does not prove
Neither that a given response actually followed this regime nor that an agent applied it at runtime.
Use when
When a page deals with authority, non-response, execution, or restraint.
Citation surface#03

Citations

/citations.md

Minimal external reference surface used to contextualize some concepts without delegating canonical authority to them.

Makes provable
That an external reference can be cited as explicit context rather than silently inferred.
Does not prove
Neither endorsement, neutrality, nor the fidelity of a final answer.
Use when
When a page uses external sources, sector references, or vocabulary anchors.

The extraction problem

A web page can be clear for a human reader and still become unstable once it is processed by AI systems.

The reason is simple: AI systems do not always carry the page as a whole. They extract statements, compress them, rank them, cite them, combine them with other fragments, and then produce an answer whose final frame may no longer be governed by the original source.

This is where authority becomes fragile.

Authority is not naturally portable

Human publishing often relies on context. The domain, page title, layout, author, navigation, and surrounding paragraphs help the reader understand who is speaking and under what conditions.

Machine interpretation does not preserve all of that context by default.

Once a statement is extracted, the system must still retain:

  • issuer;
  • canonical source;
  • publication or update state;
  • perimeter;
  • modality;
  • exceptions;
  • governing source;
  • supersession state.

If those signals disappear, the statement may continue to look useful while losing its authority.

Statement-level authority

Statement-level authority names the capacity of a claim to carry its governing signals after extraction.

That capacity becomes critical for public information, policies, definitions, technical documentation, doctrinal claims, and any source whose meaning depends on scope.

A statement that cannot preserve its authority should not become the governing basis of an AI answer.

The GovLoop contribution

The GovLoop article on government information gives this problem an institutional version: when AI systems interpret public information, authority should be defined rather than inferred. The same logic applies beyond government.

It applies to every entity that publishes claims likely to be reused by AI systems outside their original document.

Closing rule

The real test of authority is not whether it is visible on the source page. The real test is whether it survives extraction.