Skip to content

Article

Known-source risk and phantom citations

A system may cite or reconstruct a source because it appears known, not because the current page legitimately supports the answer.

CollectionArticle
TypeArticle
Categoryrisque interpretatif
Published2026-05-13
Updated2026-05-13
Reading time3 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. 02Site context
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.

Context and versioning#02

Site context

/site-context.md

Notice that qualifies the nature of the site, its reference function, and its non-transactional limits.

Governs
Editorial framing, temporality, and the readability of explicit changes.
Bounds
Silent drifts and readings that assume stability without checking versions.

Does not guarantee: Versioning makes a gap auditable; it does not automatically correct outputs already in circulation.

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
    Weak observationQ-Ledger
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.
Observation ledger#02

Q-Ledger

/.well-known/q-ledger.json

Public ledger of inferred sessions that makes some observed consultations and sequences visible.

Makes provable
That a behavior was observed as weak, dated, contextualized trace evidence.
Does not prove
Neither actor identity, system obedience, nor strong proof of activation.
Use when
When it is necessary to distinguish descriptive observation from strong attestation.

A system may cite or reconstruct a source because it appears known, not because the current page legitimately supports the answer.

Known-source risk is one of the most important limits of citation analysis. A cited URL can look authoritative while the underlying source behavior is unstable, outdated or reconstructed.

The visible citation may not be the real proof path.

What known-source risk means

Known-source risk appears when a system treats a source, URL pattern, domain or page type as familiar and uses that familiarity to support an answer without enough current verification.

This can happen when an old URL remains known, when a site changed its structure, when a brand used to publish a claim it no longer supports, or when a model reconstructs a plausible source path from memory-like patterns.

Phantom citation

A phantom citation is a citation that appears to point to a source but cannot be verified as a current, supporting, accessible source for the claim being made.

It may be a broken URL, a reconstructed URL, an outdated page, a page that never contained the claim, or a citation whose displayed source does not support the answer.

Why this matters for AI citation readiness

Citation readiness is not only about getting cited. It is also about preventing the wrong source from becoming the apparent authority.

A corpus can be damaged by old pages, archived claims, inconsistent slugs, copied descriptions, weak redirects and abandoned content that still looks plausible to answer systems.

If those surfaces remain ungoverned, a model may cite the past version of the entity rather than the current one.

Typical symptoms

SymptomRisk
AI cites a retired URLThe answer may be governed by a stale source
AI names a page that does not existThe system may be reconstructing a plausible path
AI cites a page but uses another claimThe citation may be ornamental or mismatched
AI uses an old brand descriptionEntity memory may override current positioning
AI cites an aggregatorA derivative source may replace the canonical source

How to reduce the risk

The correction is not simply “publish more content”. The correction is to stabilize the current source environment.

Useful actions include:

  • mapping old URLs and redirect behavior;
  • clarifying which pages are current and canonical;
  • removing or marking obsolete claims;
  • creating self-contained current passages;
  • strengthening entity consistency;
  • linking market-facing pages back to stricter definitions;
  • monitoring citations by role, not only by count.

Governance implication

Known-source risk is a source hierarchy problem. The system may know a source, but knowing a source does not make it admissible authority for the present claim.

An audit must therefore ask whether the cited source is current, accessible, relevant, supporting and authorized to govern the answer. If not, the citation is not evidence of fidelity. It is a symptom requiring correction.