AI regulation in Japan: the approach in brief

Short answer: In 2025 Japan passed its first dedicated AI statute, the AI Promotion Act, but chose a promotion-first, innovation-led approach. The law sets out basic principles and a coordination structure; it contains no prohibited AI practices, no fines and no mandatory conformity assessment. The practical norms live in non-binding guidelines issued by the Ministry of Economy, Trade and Industry (METI).

A framework law, not a product law

In May 2025 Japan's National Diet passed the Act on Promotion of Research and Development and Utilization of AI-Related Technologies — the AI Promotion Act. It is Japan's first law expressly directed at AI, but its design differs fundamentally from the EU AI Act. Where the EU builds a risk-based product law with prohibitions, high-risk duties and penalties, the Japanese statute mainly sets out basic principles and policy direction.

The Act tasks the state with promoting AI research, development and use, drawing up a national AI Basic Plan, and coordinating policy across ministries. To do so it establishes an AI Strategy Headquarters headed by the Prime Minister as the central coordinating body. The institutional core — the chapters on this headquarters and the Basic Plan — came into effect during 2025.

Promotion rather than prohibition

What distinguishes the Japanese approach is what the law does not do. It introduces no prohibited AI practices, no mandatory prior registration or conformity assessment, and no direct fines or criminal penalties on businesses. The toolkit is soft: principles, a policy plan, ministerial guidelines, and powers to gather information, conduct investigations and request cooperation from market participants.

This makes the Act an organising and steering instrument rather than an enforceable compliance regime in the European sense. The premise is innovation-first: the government seeks to accelerate the development and use of AI and to address risks through coordination and transparency rather than through compulsion.

Where the practical norms live

Because the Act itself carries few concrete obligations, the operational core sits in non-binding guidelines. The AI Guidelines for Business, drawn up by METI together with the Ministry of Internal Affairs and Communications (MIC), consolidate earlier guidance into a set of guiding principles for safe and trustworthy AI use. They ask organisations to recognise AI risks and voluntarily take appropriate countermeasures across the whole lifecycle. The guidelines are not law; adherence rests on self-regulation.

Where it sits on the global map

Japan illustrates a lighter variant within the international landscape. Where the EU regulates horizontally and bindingly, and South Korea adopted that form but front-loads promotion, Japan remains a step more restrained: a promotion-oriented framework law resting on guidelines rather than binding duties. For organisations operating across jurisdictions this means Japan's "rules" are largely expectations and guidance — authoritative, but not enforceable the way the EU obligations are. Read the official Japanese texts on their own terms, not as a mirror of the EU.

Read more: International AI governance. Take the scan.