Japan’s evolving AI and digital health regulations: legal developments and outlook

Yurika Inoue
Mori Hamada & Matsumoto, Tokyo
yurika.inoue@morihamada.com

Introduction

Artificial intelligence (AI) is being increasingly embedded into the healthcare and pharmaceutical landscape in Japan, from software that supports patient behavioural change to algorithms that can generate novel drug candidates. As these technologies advance, Japan’s regulatory regime is striving to keep pace with both the necessary safety imperatives and the opportunities for innovation. This article outlines the key legal and regulatory developments under Japanese law that companies and practitioners should monitor.

SaMD and digital therapeutics in Japan

In Japan, the Pharmaceuticals and Medical Devices Act (the ‘PMD Act’) governs medical devices and, since its amendment in 2014, it explicitly covers ‘programs’ as medical devices if they are intended for diagnosis, treatment or the prevention of disease. To guide developers in determining whether their software qualifies as a medical device, the Ministry of Health, Labour and Welfare (MHLW) has issued Guidelines on the Determination of Whether Programs Are Regarded as Medical Devices,^[1] which contains a decision-making flowchart. The Guidelines set out criteria based on (1) the software’s intended purpose, namely how much it contributes to diagnosis, treatment or disease prevention, and (2) the level of risk to the patient’s health if the software malfunctions. Software used merely for storing, displaying or transmitting data without a diagnostic or therapeutic purpose generally falls outside of the PMD Act’s remit.

A notable example of the provision of successful digital therapeutics using this framework is CureApp SC, the first therapeutic app officially approved in Japan. Classified as a Class II medical device program under the PMD Act, it was approved in 2020 to treat nicotine dependence. The app provides behavioural therapy content, physician-linked data management and an AI-driven chat function that offers interactive guidance on coping strategies when the patient’s nicotine cravings increase. It represents the first SaMD approved in Japan and is a good example of how digital interventions can be safely integrated into clinical practice. By supporting patient engagement between clinical visits, the app enhances the overall effectiveness of physician-led smoking cessation treatment.

To further support the development and practical implementation of such innovative software-based devices, Japan has introduced the Digital Transformation Action Strategies in Healthcare for SaMD or the ‘DASH for SaMD’ strategy. This government-led initiative promotes the rapid development and early application of SaMD through the provision of streamlined regulatory support, including faster reviews, development guidance and industry engagement. As part of this strategy, the Pharmaceuticals and Medical Devices Agency (PMDA) operates a dedicated SaMD One-Stop Consultation Desk (iryou-kiki program sougou soudan madoguchi). This consultation service allows developers to seek early regulatory advice on product classifications, clinical evaluations and application strategies. It aims to promote timely and transparent communication between regulators and innovators, reducing uncertainty during the pre-submission stage.

AI use in drug discovery: Japanese developments and regulatory perspective

Japan has experienced rapid growth in AI-driven drug discovery, combining public infrastructure with private sector innovation. A prominent example is the 2025 commercialisation of a federated AI model by Tokyo-based startup Elix and the Life Intelligence Consortium (LINC), a Japanese organisation dedicated to advancing the life sciences sector through the development and social implementation of applications using AI, Big Data and Internet of Things (IoT) technologies, supported by the Japan Agency for Medical Research and Development (AMED). This model was trained on proprietary data from 16 pharmaceutical companies and integrates multiple AI systems for target prediction, compound screening and optimisation, representing a milestone in secure industry-wide collaboration.^[2]

Although AI systems used solely in drug discovery are not regulated as medical devices, the PMDA has begun to articulate expectations around AI validation when such tools are used to support clinical or regulatory decisions. Core review points, borrowed the from SaMD regulation, emphasise transparency, data quality and reproducibility. These principles are becoming increasingly relevant as AI-generated results are incorporated into clinical trial design, companion diagnostics and submission packages. The PMDA has also released an internal AI Action Plan outlining how AI-based technologies should meet standards of security and reliability, with a view to evaluating their future use to support increasingly complex and specialised regulatory tasks.

AI regulations in Japan

Aside from the SaMD regulations, Japan is developing a broader framework for governing AI, while maintaining its emphasis on innovation. Rather than introducing a single, rigid AI law like the European Union’s AI Act, Japan continues to rely on principles-based guidance. The first version (Ver. 1.0) of the AI Business Operator Guidelines was released on 19 April 2024 (updated to Ver. 1.1 on 28 March 2025)^[3] by the Ministry of Internal Affairs and Communications and the Ministry of Economy, Trade and Industry, representing a consolidation and update to the prior AI-related guidance. The Guidelines were formulated to set out the fundamental principles and approaches necessary for the development, provision and utilisation of AI. Although the Guidelines are not legally binding, discussions on the future structure of Japan’s AI governance framework, including whether formal legislation should be introduced, have continued since their publication.

A key new development is the proposed AI Promotion Act, which was enacted in 2025. The Act aims to create a national framework for promoting safe and trustworthy AI use, supporting both innovation and risk management. It emphasises voluntary compliance by business and the establishment of a body to oversee ethical and technical guidelines issued by the government. Although the details remain under discussion, the Act signals a gradual move towards more formalised AI governance. Japan’s participation in the G7’s Hiroshima AI Process also demonstrates its commitment to global cooperation on the matter. As the government continues to develop implementing measures and detailed guidance, close attention should be paid to future policy developments.

Conclusion

Japan’s regulatory approach to AI and digital health is still evolving, but the direction is clear: to build an environment where innovation and safety can coexist. With initiatives such as DASH for SaMD, the AI Business Operator Guidelines and the AI Promotion Act, Japan is taking steady steps towards establishing a framework that encourages responsible AI use. Ultimately, the goal should be to foster a system in which the life sciences sector can further advance through the use of cutting-edge AI technologies, enabling new discoveries, improving patient outcomes and driving sustainable healthcare innovation.