本文目前仅提供英文版本。
AI Regulation by Country 2026: The Complete Global Map
The complete 2026 guide to AI regulation globally — which countries have laws, which have guidelines, which are developing frameworks, and what organisations operating internationally need to know about the patchwork of obligations they face.
Key Takeaways
As of 2026, the EU has the only comprehensive mandatory AI law (EU AI Act). Most other jurisdictions have sector-specific guidance, voluntary frameworks, or AI regulations embedded in existing laws (privacy, consumer protection, anti-discrimination).
The four jurisdictions where AI governance obligations are most concrete: EU (AI Act), Australia (Privacy Act AI obligations, sector regulators), UK (ICO AI guidance, FCA Consumer Duty, sector regulators), and Singapore (MAS FEAT for financial services, PDPC Model Framework).
Countries actively developing AI legislation in 2025-2026: South Korea (AI Basic Act enacted January 2026), Japan (AI Promotion Act enacted May 2025), India (DPDPA implementation rules expected 2026). Canada's Bill C-27 (AIDA) lapsed in January 2025 when Parliament was prorogued — no replacement bill yet introduced.
The US has no comprehensive federal AI law — governance comes from sector regulators (CFPB, FTC, EEOC enforcement actions), state legislation (Colorado AI Act stayed, Illinois AI hiring law, Texas AI business law), and executive orders that vary by administration.
For organisations operating internationally, the practical approach is: comply with the EU AI Act as the highest mandatory standard, align with NIST AI RMF as the operational methodology, and layer jurisdiction-specific requirements on top for your key markets.
"仅供参考。本文不构成法律、监管、财务或专业建议。如需具体指导，请咨询合格专家。"
The global AI regulation landscape in 2026
The defining feature of global AI regulation in 2026 is fragmentation. There is no international AI governance treaty. There is no single global standard with binding force. What exists is an increasingly complex patchwork of national laws, sector-specific regulations, voluntary frameworks, and enforcement actions that organisations operating across borders must navigate simultaneously. The EU AI Act is the only comprehensive mandatory AI law in force, and it is setting the de facto global standard for multinational organisations — not because it applies everywhere, but because organisations with EU exposure must comply with it, and compliance with the EU AI Act substantially addresses the requirements of other jurisdictions.
Europe: the EU AI Act and its neighbours
The EU AI Act is in force and its major compliance deadlines are approaching. Prohibited AI practices (social scoring, certain biometric surveillance) were prohibited from 2 August 2024. GPAI model obligations for general purpose AI model providers apply from 2 August 2025. High-risk AI obligations under Annex III apply from 2 December 2027 (delayed from the original 2 August 2026 deadline under the AI Act Omnibus of May 2026). All EU member states have designated their national market surveillance authorities and competent authorities for AI Act enforcement.
Outside the EU, European AI regulation reflects the AI Act's influence. Switzerland is aligning its AI guidance with EU AI Act principles. Norway and Iceland, as EEA members, will adopt the AI Act through the EEA agreement. The UK has maintained its pro-innovation sector-led approach post-Brexit, but UK regulators are closely monitoring EU AI Act implementation and their approaches are converging in practice.
Asia-Pacific: a diverse regulatory landscape
Australia has the most developed regulatory framework outside Europe — not through AI-specific legislation, but through active sector regulator engagement. APRA, ASIC, the OAIC, the ACCC, and the Fair Work Commission all have established AI governance expectations. The Privacy Act reforms of 2024 created new obligations relevant to AI. And the AI6 framework (October 2025) provides voluntary guidance that is becoming a de facto standard for enterprise AI deployment. Singapore's MAS FEAT Principles and the Veritas Assessment Methodology create a financial services AI governance standard that is among the most technically sophisticated in the world. Japan enacted the AI Promotion Act in April 2024 — a deliberately non-binding framework that establishes principles without mandatory requirements. South Korea enacted the AI Basic Act in December 2024, which will come into force in 2026 and creates a risk-based framework with sector-specific requirements. China has multiple AI-specific regulations in force: the Generative AI Regulation (2023), the Deep Synthesis Regulation (2022), and the Recommendation Algorithm Regulation (2022) — collectively creating one of the most prescriptive AI regulatory frameworks in the world for specific use cases.
Americas: enforcement without comprehensive law
The United States has no comprehensive federal AI law, but AI governance obligations are real and enforced. The FTC has enforcement authority over unfair or deceptive AI practices. The CFPB has been the most active federal regulator on AI in financial services, pursuing enforcement actions on algorithmic credit decisions and AI adverse action notices. The EEOC has issued guidance on AI in employment that signals enforcement intent. State legislation is creating a patchwork: Colorado's AI Act (partially stayed pending amendment), Illinois's Artificial Intelligence Video Interview Act (employment AI disclosure), Texas's Business Use of AI Act, and California's executive orders and proposed legislation. Brazil's LGPD (Lei Geral de Proteção de Dados) creates data protection obligations relevant to AI. The Brazilian Senate passed a comprehensive AI Act in 2024, pending House approval. Canada's Bill C-27, which included the AI and Data Act (AIDA), died on the order paper in January 2025 when Parliament was prorogued. No successor bill has yet been introduced, leaving Canada without comprehensive federal AI legislation.