Global AI regulation divergence: what multinationals need to know

The EU AI Act is the most comprehensive AI regulatory framework in force globally, but it is far from the only one. Multinational organizations now face a genuinely complex mosaic of AI-related legal requirements, guidance, and voluntary frameworks across different jurisdictions — with meaningful divergence in approach, scope, and enforcement.

The EU: horizontal and binding

The EU AI Act establishes a horizontal, risk-based framework applicable across sectors. It is directly applicable in all 27 member states. Its extraterritorial reach — applying to providers outside the EU whose systems are used within the EU — means organizations based in the US, UK, Asia, or elsewhere cannot treat it as irrelevant.

The UK: sector-led and principles-based

The UK has chosen a deliberately different approach. Rather than a single horizontal AI law, the UK government has asked sector regulators (FCA, ICO, CMA, Ofcom, MHRA) to apply existing regulatory principles to AI within their domains. The ICO has published AI and data protection guidance; the FCA has addressed AI in financial services.

This creates a less prescriptive but more fragmented landscape. For multinationals operating in both the EU and UK, the challenge is reconciling EU Act specificity with UK principles-based flexibility — and tracking sector-specific developments in each relevant domain.

The US: fragmented and evolving

The US has no federal AI law in force. Federal agencies have issued sector-specific guidance (EEOC on AI in employment, CFPB on AI in credit decisions, FDA on AI in medical devices). Executive orders have shaped federal agency behaviour. State-level legislation is active: Colorado, Illinois, Texas, and others have enacted or are considering AI-specific laws, particularly in employment and automated decision-making contexts.

For multinationals, the US landscape requires state-by-state monitoring alongside federal developments — a resource-intensive exercise that shows no sign of simplifying.

Japan: promotion with governance

Japan's approach has prioritized AI innovation and international interoperability alongside governance. The Basic Act on the Promotion of Responsible AI was enacted in 2025, establishing principles-based national policy. Japan participates actively in international standards development and has sought alignment with the OECD AI Principles.

China: sector-specific and expanding

China has enacted several AI-specific regulations: on algorithmic recommendation systems (2022), on deep synthesis (2022), and on generative AI (2023). Further regulation is expected. The regulatory model is more directive and enforcement-focused than the EU's rights-based approach, and the obligations differ substantively.

Implications for multinationals

Operating across these jurisdictions requires:

The EU AI Act's requirements — risk management, documentation, human oversight, transparency — provide a reasonable baseline that meets or exceeds many other jurisdictions' requirements in their current form. Building compliance programs anchored to the EU Act, with jurisdiction-specific supplements, is a practical approach for many multinationals.

The regulatory landscape will continue to evolve. Staying current requires systematic monitoring, not reactive response.