Where AI law binds, where it is drafted, where it is still a strategy
Each of the 109 jurisdictions is placed in one of six regulatory models and given a maturity score from 1 to 5 for binding force, scope, implementation and AI specificity. Select a country for its flagship instrument; recolour the map by model, maturity or status.
Rank by maturity
Top ten of 109 on the Atlas score, coloured by model
Soft law still outnumbers hard law
The tracker's instruments divide into six types. Strategies and guidelines outnumber binding acts, regulations and decrees, and roughly one instrument in three is written but not yet in force.
Instruments by type
Blue: soft law · grey: binding · light: bills
Status of the 1,170
In force, in the pipeline, or other (repealed, expired, unclassified)
Instruments by year
Dated instruments added to the corpus
Most active jurisdictions
By number of tracked instruments
The United States figure combines federal action with a large and fast-growing body of state law. NYU's Center on Technology Policy counted 109 new state AI laws in 29 states by 1 July 2026.
Also in the tracker: 4,500+ glossary terms, 296 enforcement actions and court decisions, and 581 research papers.
There is no single grammar of AI law
Brussels legislates horizontally; London distributes the work to existing regulators; Beijing writes targeted measures for each new application. Around those three sit promotion laws, pending bills and strategy-first governance. Select a model to filter the map.
Seven years from soft principles to binding deadlines
Milestones in cyan have happened; deadlines in gold are still ahead. On small screens, scroll the years sideways.
Law on paper is not governance in practice
The Global Index on Responsible AI (2026) scores 135 countries on 38 indicators and finds a global average near 35 out of 100. The IMF AI Preparedness Index scores 174 economies from 0 to 1. Both maps are densest in rich, digital, mostly European states.
GIRAI 2026: top ten
Score out of 100 · global average ≈ 35 · Europe 54.2 · Global South 27
IMF AI Preparedness Index: top ten
Score 0–1 · China 0.63 (31st) · India 0.49 (72nd) of 174
Maximum fines differ by a factor of more than a thousand
Ceilings for the most serious breach, converted to euros at approximate November 2026 rates and drawn on a logarithmic scale. Percentage-of-turnover caps apply when higher.
Which rules touch my AI system?
Choose where the system's outputs are used and what it does. The Atlas lists the instruments that apply or are coming, with their dates. It is an orientation tool, not legal advice.
All 109 jurisdictions, sortable
Search by name or instrument; filter by region and model; sort any column. Confidence reflects how directly the classification rests on primary or high-quality secondary sources.
| Jurisdiction | Region | Model | Maturity | Status | Year | Flagship instrument | Confidence |
|---|
Every number links back to its source
Instrument totals, types, status and the five largest jurisdiction counts come from the regulations.ai tracker. Responsibility scores come from GIRAI 2026; readiness scores from the IMF AIPI; investment, incidents and trust from the Stanford HAI AI Index 2026. Model, maturity and confidence are the Atlas's own editorial classification of each jurisdiction's flagship instruments as of 4 November 2026; they are judgements, shown with their sources so that you can check them.
- regulations.ai — the source for AI regulation
- Global Index on Responsible AI, 2026
- IMF AI Preparedness Index (World Bank Data360)
- Stanford HAI — AI Index Report 2026
- CASRAI — AI regulations around the world
- CASRAI — Brazil's PL 2338/2023 status
- Cooley — Digital Omnibus on AI, Reg. (EU) 2026/1744
- Council of Europe — EU ratifies the AI Convention
- Tech Policy Press — US state AI legislation, mid-2026
- PPC Land — Asia's AI laws compared
- ComplyJet — AI regulations by country, 2026
- Mondaq — India's AI Governance Guidelines
- Comparitech — AI legislation in 178 countries
- GKToday — IMF AIPI rankings