Loading
No two countries govern artificial intelligence the same way. Some have a binding, comprehensive AI Act; others rely on frameworks and existing sector regulators; the United States layers many state laws over a voluntary federal framework. This page compares the main approaches and links to a dedicated, source-backed policy page for each country.
Last reviewed: 21 July 2026 · A factual snapshot; each country page carries the primary sources and dates
A quick orientation. The final column links to the full policy page for each jurisdiction, with the instruments, the dates, and what is binding versus voluntary.
| Jurisdiction | Approach | AI Act? | Read the full page |
|---|---|---|---|
| European Union | Binding, comprehensive AI Act The EU AI Act (Regulation 2024/1689), phased in across the bloc | Yes | EU AI Act |
| United States | State patchwork, plus a federal fight State laws (Texas, California, Colorado and more) over the voluntary NIST AI RMF; a federal push to override them | No federal Act | United States AI policy |
| United Kingdom | Pro-innovation, principles-based Five cross-sector principles applied by existing regulators (ICO, FCA, Ofcom, MHRA, CMA) | No | United Kingdom AI policy |
| Australia | Frameworks and sector regulators Voluntary guidance (Guidance for AI Adoption) plus binding sector rules (APRA, ASIC, the Privacy Act) | No standalone Act | Australia AI policy |
| Germany | EU AI Act, plus a national supervision law The KI-MIG, in force since 29 July 2026, making the Bundesnetzagentur the central AI Act supervisor | Via the EU Act | Germany AI policy |
| Netherlands | EU AI Act, coordinated supervision The data protection authority as coordinating AI supervisor; the algorithm register | Via the EU Act | Netherlands AI policy |
| Singapore | Pro-innovation, framework-driven The Model AI Governance Framework and AI Verify over the binding PDPA; MAS for finance | No | Singapore AI policy |
| Canada | No federal AI Act after AIDA lapsed Existing law applied to AI (PIPEDA, Quebec Law 25, human rights, OSFI) plus voluntary federal frameworks and the AI for All strategy | No | Canada AI policy |
| Japan | Innovation-first, promotion law The 2025 AI Promotion Act (no sanctions) plus the voluntary AI Guidelines for Business, over binding APPI and Copyright Act | Yes, no penalties | Japan AI policy |
| South Korea | Comprehensive binding AI Act The AI Basic Act (in force 22 Jan 2026), high-impact + generative-AI duties, fines from ~2027; over binding PIPA | Yes | South Korea AI policy |
| India | No AI Act; a hardening patchwork Existing law applied to AI: the DPDP Act and 2025 Rules, SEBI Regulation 16C, the IT deepfake rules, plus voluntary MeitY guidelines and the RBI FREE-AI framework | No | India AI policy |
| France | EU AI Act, decentralised supervision The EU AI Act applied directly over the GDPR and Loi Informatique et Libertes; the CNIL as reference authority, national designations still pending | Via the EU Act | France AI policy |
| Spain | EU AI Act, plus a national bill in Congress AESIA already operating (created 2023) as the EU's first national AI agency; a national AI bill before Congress since May 2026, not yet enacted | Bill pending | Spain AI policy |
| Mexico | No federal AI law; INAI dissolved Multiple federal AI bills pending; data protection now with the Secretaria Anticorrupcion y Buen Gobierno (INAI's successor); binding platform-worker algorithm-transparency rights | No | Mexico AI policy |
| Hong Kong | Pro-innovation, framework-driven The binding Personal Data (Privacy) Ordinance; non-statutory PCPD, HKMA and SFC guidance; a legislative review underway since March 2026 | No | Hong Kong AI policy |
Where obligations carry dates, we track them row by row and update them as laws commence, get repealed or are amended:
The European Union has the most comprehensive one, the EU AI Act, which applies across its member states including Germany and the Netherlands. Most other major jurisdictions do not have a standalone AI Act: the United Kingdom, Australia and Singapore govern AI through principles, frameworks and existing sector regulators, and the United States relies on a patchwork of state laws rather than a federal AI Act.
Three broad models are visible in 2026. A comprehensive statutory model, exemplified by the EU AI Act, sets binding, risk-based rules directly. A framework-and-regulator model, used by the UK, Australia and Singapore, applies principles and voluntary frameworks through existing sector regulators rather than a new AI law. A patchwork model, seen in the United States, layers many state laws over a voluntary federal framework, with the balance between state and federal power still being contested.
Yes. The EU AI Act can apply to an organisation based anywhere if it provides or deploys AI systems in the EU market. For many US, UK and other non-EU firms, the EU AI Act is the most concrete AI statute they face, which is why it appears on their country pages as well.
Whichever regime applies, the foundations are the same: a clear AI inventory, named ownership, risk assessment and documented controls. A short, free assessment shows where an organisation stands against a structured model that maps across jurisdictions.
This page is general information as at 21 July 2026, not legal or compliance advice. AI policy is evolving quickly; confirm the current position against the primary sources on each country page and obtain advice from your own qualified counsel before relying on it.