Skip to content

Research governance

Data Sources & Citation Policy

How AI Governance Ratings selects, attributes, versions and qualifies primary sources, standards, regulatory material, datasets and related research.

Source integrity is part of research quality.

AIGovernanceRatings.com is designed around traceable research. Material claims should be supported by sources appropriate to the claim, and readers should be able to distinguish an external fact or requirement from internal methodology interpretation.

Source hierarchy

Source class Preferred use Examples
Law and regulator Legal obligations, effective dates, official interpretations and regulatory status. Official legislation, regulator guidance, government publications.
Standards body Published management, governance, risk and technical standards. ISO/IEC, NIST and other official standards or framework publishers.
Original methodology Definitions, rating architecture, governance domains and operative methodology claims. AIGR™ Global methodology and controlled methodology documentation.
Original research Framework development, benchmark design, sector analysis and Responsible AI research. AIGX Research™ and other identified research institutions.
Academic / institutional research Empirical findings, technical analysis and literature context. Peer-reviewed papers, university research, recognized institutional studies.
Secondary sources Context, terminology or discovery of primary material. Professional research and institutional publications, used with attribution and source-quality review.

Publication-level citations

Research papers include a Sources and research basis section identifying principal material relied upon. Where a claim depends on a particular version, date, jurisdiction or dataset period, the publication should state that context when it is material to interpretation.

Related ecosystem sources

Third-party data limitations

External datasets remain governed by the original publisher’s methodology, sampling, definitions, error corrections and update cycle. Citation does not independently validate every underlying observation. Where data is transformed, aggregated or interpreted, the publication should distinguish the transformation from the source data.

Original and benchmark data

Where future publications use original assessment or benchmark data, the publication should describe the cohort, period, inclusion criteria, aggregation approach, material exclusions and confidentiality controls at a level sufficient to interpret the findings. Client evidence should not be disclosed through public benchmark research without appropriate authorization and safeguards.

Corrections and source changes

If an authoritative source is materially revised or a cited factual claim is found to be incorrect, the associated publication should be reviewed and corrected where the change affects interpretation. Correction or source questions may be submitted through the Contact page.