Exposure against governance readiness

Two measures of the same population: the share of working-age people using AI tools, and how much machinery their jurisdiction has built to govern it. Each is converted to a percentile rank within the jurisdictions present in both, and one rank is subtracted from the other.

The gap

Rank in use, minus rank in Governance readiness A score compiled here from primary instruments, recording the machinery a jurisdiction has for governing AI. It records presence, not quality or enforcement. Glossary entry . 17 jurisdictions sit meaningfully ahead in use, 19 meaningfully ahead in governance.

Governance well ahead
Governance ahead
Roughly level
Use ahead
Use well ahead
Not in both datasets

Gap is the difference between percentile ranks within the 58 jurisdictions present in both datasets. AI Diffusion Report: AI User Share Microsoft AI for Good Lab Licence MIT Updated quarterly Retrieved 2026-10-23, 5 days ago Derived from anonymised Microsoft telemetry, then adjusted for OS/device market share, internet penetration and a modelled mobile-to-desktop ratio. It therefore measures Microsoft-visible AI use, scaled, not observed total AI use. Systematically less reliable where Microsoft platform share is low, and it cannot see AI embedded in products users do not recognise as AI. Cite Microsoft AI for Good Lab, "AI Diffusion Report: AI User Share". https://github.com/microsoft/ai-diffusion-report. Retrieved 23 October 2026 via AI Safety Tracker, https://aisafetytracker.org/map/. AI Governance Readiness Index This site Licence MIT Updated quarterly Retrieved 2026-10-23, 5 days ago Data JSON Our own editorial coding from primary legal sources, not an objective measurement. Every score cites a specific instrument and article. "Readiness" means the machinery exists to govern AI. It is emphatically not a judgement that the resulting rules are good, proportionate, or well enforced. Cite AI Safety Tracker, "AI Governance Readiness Index". https://aisafetytracker.org/methodology/governance-readiness/. Retrieved 23 October 2026. Natural Earth country boundaries, 1:110m Natural Earth Licence public-domain Updated irregular Retrieved 2026-10-23, 5 days ago Public domain, with no permission needed and no attribution required; we credit it anyway because a boundary is a datum like any other. Consumed through the world-atlas package at build time, at 1:110m, so small island states are absent or a single pixel. Boundaries follow Natural Earth's own editorial choices on disputed territory and are not a position of this site. Antarctica is dropped from our render. Cite Natural Earth, "Natural Earth country boundaries, 1:110m". https://www.naturalearthdata.com/about/terms-of-use/. Retrieved 23 October 2026 via AI Safety Tracker, https://aisafetytracker.org/map/.

New Zealand has the widest gap in the direction of use, Romania the widest in the direction of governance. A jurisdiction may score low on readiness by choice, and a high score does not indicate that its rules are proportionate or enforced. Read the map as a question about where to look, not as an answer.

Find a country

The two layers separately

The gap map hides the levels it is built from. These are those levels, each in five equal-count bands.

Share of working-age population using AI tools, 2026-Q2 AI Diffusion Report: AI User Share Microsoft AI for Good Lab Licence MIT Updated quarterly Retrieved 2026-10-28, today Derived from anonymised Microsoft telemetry, then adjusted for OS/device market share, internet penetration and a modelled mobile-to-desktop ratio. It therefore measures Microsoft-visible AI use, scaled, not observed total AI use. Systematically less reliable where Microsoft platform share is low, and it cannot see AI embedded in products users do not recognise as AI. Cite Microsoft AI for Good Lab, "AI Diffusion Report: AI User Share". https://github.com/microsoft/ai-diffusion-report. Retrieved 28 October 2026 via AI Safety Tracker, https://aisafetytracker.org/map/.

low high

Governance readiness, 0 to 100 AI Governance Readiness Index This site Licence MIT Updated quarterly Retrieved 2026-10-23, 5 days ago Data JSON Our own editorial coding from primary legal sources, not an objective measurement. Every score cites a specific instrument and article. "Readiness" means the machinery exists to govern AI. It is emphatically not a judgement that the resulting rules are good, proportionate, or well enforced. Cite AI Safety Tracker, "AI Governance Readiness Index". https://aisafetytracker.org/methodology/governance-readiness/. Retrieved 23 October 2026.

low high

Both axes at once

Each jurisdiction placed by its rank on each measure. The diagonal is where the two ranks match; distance from it is what the first map colours.

Ranks, not raw values. A use rate is a percentage of a population and a readiness score is a count against a rubric we wrote; subtracting one from the other directly would be meaningless. Ranking each within its own distribution first makes the comparison a statement about relative standing, which is all it can honestly be. It also means the picture moves when the set of jurisdictions changes. AI Diffusion Report: AI User Share Microsoft AI for Good Lab Licence MIT Updated quarterly Retrieved 2026-10-23, 5 days ago Derived from anonymised Microsoft telemetry, then adjusted for OS/device market share, internet penetration and a modelled mobile-to-desktop ratio. It therefore measures Microsoft-visible AI use, scaled, not observed total AI use. Systematically less reliable where Microsoft platform share is low, and it cannot see AI embedded in products users do not recognise as AI. Cite Microsoft AI for Good Lab, "AI Diffusion Report: AI User Share". https://github.com/microsoft/ai-diffusion-report. Retrieved 23 October 2026 via AI Safety Tracker, https://aisafetytracker.org/map/. AI Governance Readiness Index This site Licence MIT Updated quarterly Retrieved 2026-10-23, 5 days ago Data JSON Our own editorial coding from primary legal sources, not an objective measurement. Every score cites a specific instrument and article. "Readiness" means the machinery exists to govern AI. It is emphatically not a judgement that the resulting rules are good, proportionate, or well enforced. Cite AI Safety Tracker, "AI Governance Readiness Index". https://aisafetytracker.org/methodology/governance-readiness/. Retrieved 23 October 2026.

Every jurisdiction

Ordered by gap, widest use-ahead first. Country maps open an enlarged view with policy coverage.

AI use, governance readiness and the gap between their ranks, for every jurisdiction present in both datasets.
Jurisdiction	Using AI tools	Readiness, of 10	Gap in ranks	Basis
New Zealand	44.1%	3	+74	National
United Arab Emirates	73.3%	4	+66	National
Israel	38.7%	3	+60	National
Norway	49.4%	4	+58	National
Singapore	64.3%	6	+51	National
Switzerland	39.1%	4	+43	National
Saudi Arabia	31.4%	3	+42	National
United Kingdom	43.7%	6	+40	National
Australia	40.7%	6	+36	National
Canada	38.5%	5	+33	National
Colombia	25.9%	3	+25	National
United States	33.0%	6	+19	National
Chile	24.0%	3	+19	National
South Africa	23.9%	3	+17	National
Ireland	49.9%	10	+17	EU harmonised
France	49.6%	10	+15	EU harmonised
Argentina	23.4%	3	+15	National
Malaysia	22.6%	3	+13	National
Spain	45.1%	10	+11	EU harmonised
Mexico	21.4%	3	+9	National
Turkey	18.9%	3	+6	National
Netherlands	43.5%	10	+6	EU harmonised
South Korea	40.6%	10	+2	National
Belgium	40.3%	10	0	EU harmonised
Egypt	15.5%	3	-2	National
Indonesia	14.8%	3	-4	National
Vietnam	27.9%	7	-7	National
Nigeria	10.4%	3	-8	National
Sweden	37.3%	10	-8	EU harmonised
Austria	35.4%	10	-9	EU harmonised
Brazil	20.2%	4	-11	National
Ukraine	9.5%	3	-11	National
Hungary	33.3%	10	-11	EU harmonised
Kenya	8.9%	3	-13	National
Denmark	32.4%	10	-15	EU harmonised
Germany	32.3%	10	-18	EU harmonised
Poland	32.3%	10	-18	EU harmonised
Japan	24.7%	7	-18	National
Italy	31.4%	10	-22	EU harmonised
Thailand	13.4%	4	-25	National
Czechia	31.2%	10	-25	EU harmonised
Bulgaria	30.7%	10	-26	EU harmonised
India	18.5%	6	-28	National
Finland	30.6%	10	-28	EU harmonised
Slovenia	30.1%	10	-30	EU harmonised
Portugal	27.4%	10	-34	EU harmonised
Russia	9.9%	5	-35	National
Croatia	27.2%	10	-36	EU harmonised
Slovakia	27.1%	10	-38	EU harmonised
China	17.5%	9	-42	National
Lithuania	25.1%	10	-42	EU harmonised
Greece	21.5%	10	-53	EU harmonised
Romania	18.0%	10	-62	EU harmonised

22 of these are EU member states scored once at bloc level, because the AI Act is a regulation applying directly in each of them. Their own national measures are not separately scored, so treat an EU member's number as the floor its membership guarantees rather than a complete account. 5 coded jurisdictions are missing from every view on this page because the exposure data does not cover them: Cyprus, Estonia, Latvia, Luxembourg, Malta.

How readiness is scored

Coded here from primary instruments. The rubric and the instruments behind each score are published in the methodology.

Binding law
Rules actually addressed to AI systems. A horizontal instrument in force scores highest; sectoral or partial rules score lower; guidance scores nothing.
0 to 3
Oversight body
Whether anyone supervises. A regulator holding statutory powers over AI scores above a coordinating or advisory body.
0 to 2
Evaluation capacity
Whether the jurisdiction can test models or systems itself. An operating safety institute scores above one that has only been announced.
0 to 2
Transparency duty
Whether people have to be told. An explicit AI disclosure or labelling duty scores above a general data-protection route.
0 to 2
Strategy
A published national AI strategy or equivalent. Present or absent.
0 or 1

This measures how much apparatus exists, not how strict it is. South Korea and the EU tie at the top while imposing very different burdens, and that is the index working as designed. Every jurisdiction carries the instruments it was coded from, and a build check probes each of those links, because a citation that returns a 404 looks like evidence and is not. It is not legal advice.