Methodology

Inclusion

Records concern laws, bills, executive instruments, regulator materials, standards, court records, institutional reports and other public materials materially related to AI governance. Inclusion does not mean that a record is binding, in force or comprehensive.

Official-source-first provenance

The public interface links to the official primary document whenever one has been verified. Aggregators and discovery indexes are never cited as the authority for a legal or policy instrument. When an official document URL cannot be verified, the interface identifies the available issuing institution or jurisdiction authority instead of promoting an aggregator link.

Source classes

• Primary legal source: record-level legal instrument on an authority host.
• Court record: case-level material on a recognised court host.
• Official non-legal source: first-party government, regulator, legislature, standards-body or IGO material.
• Publisher report: allowlisted first-party AI-provider publication.
• Secondary index: discovery evidence that must not be cited as primary law.

Processing

The pipeline normalises titles and dates, maps jurisdictions, classifies source hosts, deduplicates records and generates public indexes. Host classification does not independently prove every title, translation, date or editorial summary.

Release verification

A release manifest records the release identifier, generation time, source commit, metrics and SHA-256 checksums. Website and MCP deployment verification refuses mismatched release identifiers or entry counts.

Limitations

Coverage varies by jurisdiction and language. Public web availability, anti-bot controls and changing government URLs affect verification. Users should read the linked official document and obtain qualified legal advice where consequences matter.