On 1 September 2025 Japan’s Act on the Promotion of Research, Development and Utilisation of AI-Related Technologies (commonly the “AI Promotion Act”) came fully into force. The statute was promulgated on 4 June 2025 and entered full operation after a preparatory period. It is Japan’s first comprehensive AI statute, and the first AI-specific law in East Asia built on a promotion-led footing.
Promotion-led versus regulation-led
The EU AI Act is built on risk tiers and compliance duties — setting the red lines first, then discussing development. Japan’s AI Promotion Act places innovation promotion first, defines the roles of the state, local authorities, research institutions and companies, and emphasises industrial competitiveness through development and utilisation.
The difference is not a clash of values but a choice about sequencing: the EU regulates high-risk uses first, while Japan leaves room for application and regulates gradually through soft guidance. For companies operating in both markets, this means preparing two different compliance narratives. See our notes on dual-track compliance for AI companies going global and the EU AI Act.
Three principles and the institutional design
The Act sets three principles: human-centric, meaning utilisation must not harm fundamental rights or the public interest; risk minimisation, meaning necessary measures must match the technology and the use context; and innovation promotion, meaning the environment for development and reasonable use should be created.
Institutionally, the Act requires the government to adopt a basic AI plan and establishes an “AI Strategy Headquarters” chaired by the Prime Minister. This raises AI policy to cabinet-level coordination and avoids fragmentation across ministries.
Notably, the statute itself contains no administrative penalty provisions; concrete compliance requirements come mainly through ministry guidance and existing laws. Risk management guidance for generative AI and AI agents is the area to watch.
What it means in practice
First, soft law is not unconstrained. A promotion-led statute still generates duties indirectly through data protection, consumer protection and copyright law. The compliance baseline in Japan still runs through the Act on the Protection of Personal Information and related guidance.
Second, policy benefits and compliance costs arrive together. The Strategy Headquarters is typically accompanied by industrial support for compute, data and R&D; companies seeking that support should also be ready to cooperate with risk assessment and information reporting.
Third, Japan and the EU stack. A product sold in both markets must satisfy “innovation-friendly” and “risk-tiered” requirements within one governance architecture, which makes design flexibility critical. See our note on coordinating multi-jurisdiction compliance.
Japan offers a useful reference: AI governance is not only about tighter control. The key is matching the institutional design to the stage of development and the risk profile of the application.
This article is based on Japan’s Act on the Promotion of Research, Development and Utilisation of AI-Related Technologies and on public information. It is for general reference only and does not constitute legal advice.