Send message to Author

Paper short abstract

This paper examines the evolution of Japan’s approach to artificial intelligence governance, tracing its development from an initial reliance on soft-law instruments toward a more structured hybrid framework.

Paper long abstract

Historically, Japan has addressed AI-related risks and opportunities primarily through self-regulation, ethical principles, and multistakeholder dialogue and coordination, rather than through comprehensive and prescriptive legislation. Notably, in pursuing its policy ambition of becoming the “world’s most AI-friendly country”, the Japanese regulatory stance has been conceived as developmental, innovation and market-friendly, emphasizing flexibility, adaptability, and attractiveness to foreign technology investors over regulation and binding constraints. This trajectory, however, has been challenged by growing public and business concerns about unregulated artificial intelligence, prompting calls for stronger and more structured rules and ultimately leading to the 2025 enactment of Japan’s first law on artificial intelligence. Accordingly, this paper examines the development of artificial intelligence governance in Japan, focusing on its gradual transition from a predominantly soft law-based approach to a more structured hybrid regulatory framework.

Situating the Japanese experience within the broader global debate on AI regulation, the work moves beyond the simplified opposition between regulation and innovation. It shows how Japan, drawing on its corporate and business culture, which traditionally values consensus-oriented decision-making, reputational incentives, and voluntary compliance, has sought to reconcile risk management, technological development, and social trust through an “Agile Governance” approach, reframing traditional state-centric regulatory paradigms and shifting toward an institutionalized, adaptive, agenda-setting supervisory role for public authorities in close coordination with private actors’ internal yet vertically oriented rule-making.

The paper focuses on recent developments, notably the 2025 Act on the Promotion of Artificial Intelligence, which, while not establishing a fully enforceable regime, represents a significant step toward institutionalizing AI governance by providing a statutory basis for coordination, information-gathering, and policy steering. It also examines the continued centrality of soft-law instruments, particularly in sectors where AI systems have legally or economically significant effects. Moreover, it will critically assesses the limits of this approach, including the selective scope of statutory intervention, the limited use of enforceable obligations, and ongoing challenges to accountability and legal certainty in a globalized AI market. Finally, the work will offers a comparative perspective on how regulatory culture and institutional design shape responses to shared challenges posed by rapidly evolving AI technologies.