Skip to content

Article

Source substitution in AI answers

Source substitution is one of the clearest ways a cited answer can become plausible but illegitimate.

CollectionArticle
TypeArticle
Categoryrisque interpretatif
Published2026-05-13
Updated2026-05-13
Reading time2 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
  3. 03Public AI manifest
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.

Entrypoint#03

Public AI manifest

/ai-manifest.json

Structured inventory of the surfaces, registries, and modules that extend the canonical entrypoint.

Governs
Access order across surfaces and initial precedence.
Bounds
Free readings that bypass the canon or the published order.

Does not guarantee: This surface publishes a reading order; it does not force execution or obedience.

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
    Weak observationQ-Ledger
Observation ledger#01

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.

Source substitution occurs when an AI answer uses a weaker, older or secondary source instead of the source that should govern the claim.

It is one of the most important failure modes in AI citation analysis because the answer may still look sourced. The problem is not the absence of citation. The problem is the wrong source carrying the wrong authority.

How substitution appears

A system may cite a directory instead of the official site, a media article instead of the canonical doctrine, a product page instead of the current service page, a cached description instead of the updated source, or a competitor comparison instead of the entity’s own perimeter.

The answer can remain fluent and useful. It can even contain true fragments. But if the wrong source structures the final claim, the answer becomes hard to defend.

Why it happens

Substitution usually appears when the canonical source is difficult to retrieve, too abstract, poorly linked, weakly structured, outdated in its metadata, or less explicit than a secondary source. It can also happen when a third-party source is more visible in the query cluster than the official page.

This is not only a ranking issue. It is a source-hierarchy issue. The system found something, but not necessarily the source with the right authority.

What to audit

A source-substitution audit should compare three layers: the displayed source, the probable structuring source and the source that should govern the claim.

If those layers diverge, the correction should not be limited to adding more content. The canonical source may need a clearer answer block, stronger internal routing, better entity language, a more explicit date, a more precise perimeter or a direct route from market-facing pages to the governing source.

Correction principle

The aim is not to suppress every secondary source. External and contextual sources can be useful. The aim is to prevent them from becoming governing sources when the claim requires a canonical one.

That is why source legitimacy matters. A source is not legitimate because it appears often. It is legitimate because it has the right role for the claim.