Global AI Governance Knowledge Commons — Case study

What the Dataset Does

An open-source legislative intelligence dataset designed to make national AI governance approaches comparable across jurisdictions.

Compares AI governance approaches across countries
Distinguishes enacted laws from pending legislation and national strategies
Creates a common framework across different legal traditions
Supports researchers, policymakers, and practitioners studying regulatory trends

The Problem

Legislators and regulators are attempting to govern AI with limited visibility into what peer jurisdictions are doing.

Existing resources are useful, but they are largely designed for compliance or monitoring. They often catalog hundreds of documents without providing a consistent framework for comparison.

The question behind this project was straightforward:

What is each government's official national approach to AI?

Answering that question required a common methodology that could compare countries operating under different legal systems, institutions, and policy traditions.

The Approach

I began by reviewing existing AI governance databases and identifying gaps in their coverage and comparability.

Working from primary source documents, I selected one governing document for each country and built a hierarchy for determining policy posture:

Tier 1 Enacted national AI law

→

Tier 2 Pending legislation

→

Tier 3 National AI strategy

This hierarchy allowed countries with binding legislation, draft bills, and strategy documents to be compared without treating them as equivalent.

Using Python and the Claude API, I built a pipeline to extract and classify information from each source. The data schema accurately populated fields across core regulatory tracking pillars:

Legislative Status Issuing Authority Enforcement Mechanisms Risk Categories Binding Status

Confidence scores and review flags surfaced uncertain extractions for additional scrutiny. Every record underwent human review against the source documents before inclusion in the dataset.

The result is a structured governance dataset designed for comparison rather than document collection.

The Methodology

Tier 1 — Enacted law

Does the country have an enacted national AI law? If yes, this is the primary source.

Tier 2 — Pending legislation

If no enacted law exists, does the country have a pending bill? If yes, this anchors the classification.

Tier 3 — National strategy

If neither exists, what is the country's national AI strategy? This defines its policy direction.

The project began with a review of existing AI policy trackers, including OECD, IAPP, White & Case, DLA Piper, Quorum, and FiscalNote. These resources provide valuable visibility into AI governance activity, but most are designed for compliance, monitoring, or document discovery rather than comparative analysis.

Twenty-eight jurisdictions were selected based on regional diversity, economic significance, and relevance to the AI ecosystem. China was included because of its importance to global AI development, and both the European Union and individual member states were included to capture implementation dynamics around the EU AI Act.

Primary source documents were collected from official government websites, with translation used when necessary. Python and the Claude API were used to extract information into a structured schema. Confidence scoring and review flags surfaced uncertain records, and every entry was reviewed against the original source material before publication.

Key Findings

Countries analyzed across North America, Europe, Asia, and the Middle East

Levels of policy maturity: enacted law, pending legislation, and national strategy

Common methodology applied across every jurisdiction

The exercise reinforced that AI governance develops unevenly. Some jurisdictions govern primarily through binding legislation, while others rely on strategies, guidelines, or emerging proposals.

By separating those approaches rather than collapsing them together, the dataset makes meaningful cross-national comparison possible.

Who It's For

Policymakers benchmarking national approaches
Researchers studying comparative AI governance
Organizations navigating international AI regulation
AI governance practitioners tracking emerging norms

How It Fits the Larger Project

The governance dataset is one layer of the AI Governance Knowledge Commons, an open-source legislative intelligence effort aimed at reducing information asymmetries between democratic governments and the AI industry.

The broader project seeks to make AI governance interventions searchable and comparable across jurisdictions so that legislators, regulators, and researchers can learn from one another rather than operate in isolation.