AIRiskAware
Asia
🇯🇵METIMICPPCFSA

AI governance in Japan.

Japan passed its first AI law in May 2025, deliberately innovation-first with no penalties and no prohibitions. But METI guidelines define the commercial standard of care, and the AI Strategic Headquarters chaired by the Prime Minister signals national seriousness.

Full Japan guide

Japan at a glance

Primary Law
AI Promotion Act (May 28, 2025, enforcement Sept 2025)
Primary Guidelines
METI/MIC AI Guidelines for Business v1.1 (March 2025)
Data Protection
Act on Protection of Personal Information (APPI)
Key Regulator
AI Strategic Headquarters (chaired by Prime Minister)
Penalties
None, reputational enforcement only ("name and shame")
Approach
Innovation-first, soft law, comply-or-explain

METI/MIC AI Guidelines: 10 principles

The operative compliance benchmark, treated as standard of care by courts, regulators, and enterprise buyers.

1
Human-centred, AI must respect fundamental human rights and dignity
2
Safety and security, AI systems must be safe and secure throughout lifecycle
3
Fairness, AI must treat people fairly and avoid unjust discrimination
4
Privacy protection, Personal information in AI must be handled lawfully (APPI)
5
Transparency, AI decision processes should be explainable to stakeholders
6
User control, People should maintain meaningful control over AI affecting them
7
Accountability, Organisations must be accountable for their AI systems
8
Innovation, Governance should enable, not restrict, beneficial AI innovation
9
International cooperation, Japan actively shapes global AI governance norms (Hiroshima Process)
10
Social benefit, AI should contribute to solving societal challenges

What makes Japan different from every other major AI jurisdiction

Japan is the only major economy to have passed dedicated national AI legislation with no mandatory compliance obligations whatsoever. No prohibited AI uses, no conformity assessments, no pre-deployment registration, no financial penalties.

Enforcement is entirely reputational: Article 25(2) of the AI Promotion Act authorises public naming of non-cooperating operators. In January 2026, the Prime Minister's Chief Cabinet Secretary used this power for the first time, reviewing AI-generated deepfake content.

But non-compliance has real commercial consequences. Enterprise buyers, government procurement, and export partners all treat the METI guidelines as the compliance standard. Being named as non-cooperative creates reputational and commercial damage that in practice functions as a significant incentive to comply.

Read the complete Japan AI governance guide