A comparative analysis of national AI strategies, policies, frameworks, and regulations.
Each country directory contains source documents and analysis covering the full scope of a nation's approach to AI governance — from high-level strategy through to specific regulatory frameworks, whether they address foundation models, agentic systems, sector-specific applications, or AI safety.
- Not legal advice. This repository is for research, comparison, and policy analysis only. It does not constitute legal advice, regulatory advice, or a substitute for jurisdiction-specific professional counsel.
- AI-assisted reviews. Many of the reviews in this repository were produced or assisted by AI systems using a structured methodology. They should be treated as analytical working documents, not authoritative legal opinions.
- Errors are possible. AI systems can misunderstand source material, overstate legal effects, miss updates, or make citation and interpretation mistakes. Important claims should be checked against the primary sources and, where needed, reviewed by a qualified lawyer or subject-matter expert.
Each country directory contains its own README with an overview, source inventory, and links to analysis files.
| Country | Status | Reviews | Focus areas |
|---|---|---|---|
| EU | Analysis complete | 10 | EU AI Act, GPAI Code of Practice, Commission Guidelines, Transparency Code, HLEG Ethics Guidelines |
| Singapore | Analysis complete | 9 | Model AI Governance Frameworks (base / GenAI / Agentic AI), NAIS 2.0, FEAT, PDPC, ASEAN Guides |
| Switzerland | Analysis complete | 2 | Federal Council approach (no standalone AI law), FINMA AI governance guidance |
| UK | Analysis complete | 14 | Pro-Innovation framework, AI Opportunities Action Plan, AISI, Playbook, DUAA 2025 |
| USA | Analysis complete | 5 | NIST AI RMF, Executive Orders (Biden/Trump), AI Action Plan, AI Bill of Rights, OMB governance |
| Comparison | Location |
|---|---|
| UK vs EU | Dimension-by-dimension table, convergences, strategic implications |
| USA vs EU | Two poles of global AI governance; Brussels effect; administration volatility vs regulatory permanence |
| Small Open Economies: UK, Singapore, Switzerland | Three-way comparison of jurisdictions that rejected comprehensive AI legislation |
| EFTA and AI Governance | Norway, Iceland, Liechtenstein (EEA — AI Act binding) vs Switzerland (non-EEA) |
| GDPR as an AI Overlay (UK + EU) | Article 3 territorial scope, lawful basis, Article 22, DPIA / FRIA — how GDPR / UK GDPR functions as the de-facto AI overlay |
| UK vs Scotland | Intra-UK divergence on regulatory philosophy |
- Glossary — how key AI governance terms (provider, deployer, GPAI, risk classification, etc.) differ across the five jurisdictions
Analysis follows a structured approach adapted from a multi-LLM review of Scotland's AI Strategy 2026-2031. That project established a methodology in which three LLMs independently review a strategy document using a shared checklist, and the results are reconciled to surface consensus, disagreements, and blind spots.
This repo extends that approach to multiple countries. Each review covers: summary, key points, and observations across analytical lenses (clarity, measurability, gaps, feasibility, internal consistency, coverage, international alignment), concluding with an overall assessment.
See methodology.md for the full assessment framework.
The analysis covers all aspects of a country's AI governance position, including but not limited to:
- National AI strategies — overarching vision, goals, and delivery plans
- Regulatory frameworks — binding legislation and regulatory mechanisms (e.g. EU AI Act)
- Governance frameworks — non-binding guidance, model frameworks, codes of practice (e.g. IMDA's Model AI Governance Framework)
- Sector-specific policies — AI in healthcare, finance, defence, public services, etc.
- Technical standards — alignment with ISO/IEC 42001, NIST AI RMF, and other international standards
- Infrastructure and compute — national compute strategies, data centre policy, energy considerations
- Skills and workforce — AI literacy, upskilling, Fair Work, and labour market impacts
- Safety and ethics — AI safety institutes, responsible AI principles, ethical frameworks
- International positioning — participation in multilateral AI governance (OECD, G7, GPAI, ASEAN)
The goal is to understand not just what a country has published, but how its various policy instruments fit together, where the gaps are, and how its approach compares to other nations.
methodology.md # Shared assessment framework
README.md # This file
comparisons/ # Cross-jurisdictional comparisons
UK-EU-comparison.md
small-open-economies-comparison.md
efta-ai-governance.md
EU/
README.md # Country overview + 10-review index
source/
inventory.md
analysis/
review-claude/ # 10 section-level reviews
Singapore/
README.md # Country overview + review index
source/
inventory.md
analysis/
01-agentic-ai-governance-framework.md
02-agent-adoption-guide.md
review-claude/ # 3 section-level reviews
Switzerland/
README.md # Country overview + 2-review index
source/
inventory.md
analysis/
review-claude/ # 2 section-level reviews
UK/
README.md # Country overview + 14-review index
source/
inventory.md
analysis/
UK-Scotland-comparison.md
review-claude/ # 14 section-level reviews
USA/
README.md # Country overview (analysis not yet started)
source/
inventory.md
analysis/
- Scotland's AI Strategy 2026-2031 — Review — The original multi-LLM review that established the methodology used here. Scotland has its own dedicated repository.