Systematic comparison of AI policies across jurisdictions — the input to most claims about regulatory convergence, divergence, or first-mover advantage.
The comparability problem
The methodological difficulty is that the instruments being compared are not the same kind of thing. A binding regulation with penalties, a voluntary code of conduct, a professional-body resolution enforced by disciplinary process, a procurement clause binding only government vendors, and a national strategy document all appear as "AI policy" in a jurisdiction count, and none is substitutable for another.
The wiki's own material illustrates the range: the EU AI Act is a binding regulation with risk tiers; Singapore issues measures as codes of practice under a 2016 media-sector statute; Brazil's CFM resolution binds physicians through professional discipline rather than an administrative regulator; and the US draft GSAR clause imposes obligations no US statute imposes generally, reaching only federal vendors. A tracker that counts each as one jurisdiction-with-AI-policy obscures more than it reveals.
The principal trackers
CAIDP. The Center for AI and Digital Policy's annual index tracks AI policies, regulatory frameworks, and adherence to democratic values across countries. Its scoring against a normative baseline makes it a ranking as well as a survey, which is its strength for advocacy and its limitation for neutral comparison.
UNESCO / Inter-Parliamentary Union. The consultation paper on emerging AI regulation approaches was developed through literature review, discussion at the IPU Geneva Assembly in March 2024, and capacity-building workshops with parliamentarians — a method that captures what legislators understand themselves to be doing, which documentary review alone does not.
The two are complementary: CAIDP scores outcomes against a standard, UNESCO/IPU documents approaches and the reasoning behind them.
What the tracking is used for
Comparative claims about regulatory models rest on this work — including the comparative map sorting jurisdictions into governance models, arguments about the Brussels Effect, and assessments of whether the summit-series instruments produced convergence. The Bletchley Declaration is the clearest case where tracking matters: it obtained agreement on a shared evidence base while expressly permitting policy divergence, so counting signatories measures something different from measuring alignment.
Relationships
- related: AI Sovereignty — the comparative map of national regulatory approaches
- related: CAIDP AI & Democratic Values Index 2026, Emerging AI Regulation Approaches — UNESCO + IPU Consultation Paper (2024), Bletchley Declaration (AI Safety Summit, 2023), EU AI Act (Regulation 2024/1689), Frontier AI Governance, AI Governance (umbrella)