Research note

What official documentation can and cannot prove about AI sources

A provenance-focused review of what platform documentation, standards and disclosure guidance can establish in an AI source profile.

15 November 2026Desk review · 9 primary documents and standardsMethod v0.1

This is a desk-research note. It summarizes primary documents and does not claim a live AI citation share.

Key findings

What this review establishes

  1. Official documentation is strong evidence for what an operator publishes about its own product.
  2. Standards such as PROV-O and Schema.org clarify provenance and entity structure, not citation frequency.
  3. Risk-management guidance helps frame limitations and accountability, but does not measure answer quality.
  4. Commercial disclosure sources should be cited when affiliate or sponsorship relationships could affect interpretation.
Research question

Scope and setup

A source record often combines several kinds of evidence: a product announcement, a standard, a risk framework and an observed answer. Each answers a different question.

The strongest defensible statement is usually narrow: this operator published this policy on this date, or this observed answer linked this URL. Broader claims about stable behavior require repeated observations.

This review documents the evidence boundaries used by SEOGEO. It does not replace live measurement, legal advice or product-specific testing.

Method

How observations are assembled

  • Record the full question, locale and product surface.
  • Start a fresh conversation and retain one answer per run.
  • Capture every visible source URL and its position.
  • Repeat the question and keep each outcome separate.
  • Publish sample size, method version and unresolved limits.
Read the method
Data

Data availability

The production report will include a data dictionary, file format, license and permanent download path when the first verified sample is published.