AI Governance Profiles

North America

United States

Model

Sectoral and standards-led governance

Regulatory maturity

Advanced

The United States remains a high-impact two-level governance jurisdiction: federal executive policy, agency authority, procurement, standards and enforcement operate alongside increasingly consequential state statutes and state attorney-general implementation. There is still no single comprehensive federal AI law, so enterprise exposure depends on federal/state/sector mapping.

Europe

European Union

Model

Comprehensive AI law

Regulatory maturity

Advanced

The European Union remains the anchor jurisdiction for binding, cross-sector AI governance. Its model is distinctive because it combines a directly applicable regulation, a tiered risk framework, centralized Commission and AI Office oversight for general-purpose AI models, and member-state market surveillance for most AI systems. Since the previous review, the Digital Omnibus on AI has become binding and extended some high-risk implementation dates, while AI Act enforcement and Article 50 transparency obligations began operating on 2 August 2026.

Asia-Pacific

China

Model

Public-sector-led governance

Regulatory maturity

Advanced

China’s AI governance system is not organized around a single horizontal AI Act. It is a layered governance stack built from cybersecurity, data security, personal information protection, internet information service regulation, algorithm recommendation rules, deep synthesis rules, generative AI service measures, AI-generated content labeling requirements, cross-border data controls, and technical standards. The result is a state-led, security-sensitive, platform-and-data governance model in which AI compliance is closely tied to content governance, data governance, filing or assessment mechanisms, technical labeling, and industrial policy.

North America

Canada

Model

Public-sector-led governance

Regulatory maturity

Developing

Canada is currently most distinctive as a public-sector automated-decision accountability jurisdiction rather than as a jurisdiction with a horizontal private-sector AI Act. The federal Directive on Automated Decision-Making and mandatory Algorithmic Impact Assessment create operational accountability for government systems, complemented by the 2025–2027 federal AI strategy, Government AI Register, agentic-AI guidance and the Canadian AI Safety Institute. The former AIDA proposal in Bill C-27 did not become law and remains historical context, not a current enterprise obligation.

Asia-Pacific

Singapore

Model

Practical governance / AI assurance / innovation-enabling / standards-aligned model

Regulatory maturity

Developing

Singapore is building AI governance through assurance infrastructure, not an AI Act. Its model is governance-by-guidance, testing, assurance, public-sector implementation, enterprise enablement, and workforce capability-building rather than a single comprehensive horizontal AI statute.

Asia-Pacific

South Korea

Model

Comprehensive AI law

Regulatory maturity

Advanced

South Korea is one of the jurisdictions operating a dedicated binding AI framework statute. Its regime combines industrial promotion with transparency, high-impact AI governance, advanced-model safety and institutional oversight, while a separate public-sector AI/data administration law creates inventory, ethics and impact-assessment mechanisms for government use. Korea should therefore be treated as a distinct statutory model, not as a simple copy of the EU AI Act.

Asia-Pacific

Japan

Model

Innovation-led governance

Regulatory maturity

Developing

Japan has a national AI statute, but its legal architecture should not be confused with an EU-style horizontal enterprise compliance regime. The AI Act creates binding national coordination, responsibilities, an AI Strategic Headquarters and a Basic Plan, while enterprise governance continues to rely heavily on official guidance, existing sector, data and intellectual-property law, and AI safety evaluation infrastructure. The July 2026 AI Basic Plan strengthens this promotion-plus-risk-management model.

Asia-Pacific

India

Model

Emerging / fragmented

Regulatory maturity

Developing

India is building AI governance through national principles, existing law, sector regulators, mission delivery and new inter-ministerial and safety institutions rather than a standalone horizontal AI Act. Institutional architecture accelerated materially in 2026 through the AI Governance and Economic Group and related expert and safety mechanisms. Enterprises should track cross-cutting data law and sectoral rules while treating the AI Governance Guidelines as an important policy benchmark rather than binding AI regulation.

Europe

United Kingdom

Model

Sectoral and standards-led governance

Regulatory maturity

Advanced

UK AI governance remains sectoral and regulator-led rather than organised around a single horizontal AI Act. Binding safeguards for significant solely automated decisions now sit in data law, while the AI Playbook, AI Security Institute and coordinated regulatory sandboxes operationalise public-sector governance, safety evaluation and sector-specific adoption. The key enterprise question is which existing regulator and legal regime governs a use case, not whether it falls into a single national AI Act category.

Asia-Pacific

Australia

Model

Innovation-led governance

Regulatory maturity

Developing

Australia is pursuing a distributed, sector-aware AI governance model: mandatory Commonwealth public-sector controls, a national AI strategy, privacy and consumer-law baselines, safety-institute capacity, and targeted transparency requirements. The government has not proceeded with a broad mandatory high-risk AI guardrail regime at this stage, so the enterprise picture remains a combination of existing law, sector regulation, voluntary guidance, and implementation signals.