Regulating artificial intelligence: Japan between pragmatism and international ambition.

Japan occupies a singular position in the global landscape of artificial intelligence regulation. A major economic and technological power, a founding member of the G7, and the architect of the Hiroshima AI Process, the country has made a deliberate choice to adopt a flexible regulatory approach based on sector-specific guidelines rather than binding legislation. This choice, often misunderstood in Europe, deserves careful examination. It is not a regulatory lag, nor an assumed permissiveness, but a coherent strategy rooted in Japanese institutional culture and in a precise reading of the economic and geopolitical stakes of AI.

For France, understanding this approach is all the more important given that the two countries share converging interests on AI governance, while adopting different methods. It is precisely in this productive gap that the most promising opportunities for cooperation lie.

The Japanese approach: guidelines rather than a framework law

Unlike the European Union, which adopted the Artificial Intelligence Act (AI Act) in 2024, Japan has no horizontal legislation dedicated to AI. The Japanese framework rests on three main pillars: the Social Principles for Human-Centred AI, published in 2019 by the Prime Minister's Cabinet; sector-specific guidelines issued by the relevant ministries, in particular the Ministry of Economy, Trade and Industry (METI) and the Ministry of Internal Affairs and Communications (MIC); and self-regulatory mechanisms adopted by companies, often in consultation with government agencies.

This architecture reflects a deep conviction: in a rapidly evolving technological field, overly rigid regulation risks locking in categories that will become obsolete before they are even applied. Japan therefore favours an adaptive framework, capable of evolving at the pace of technology without requiring formal legislative revision.

This approach is not unique to AI. It is part of a Japanese tradition of governance by consensus and administrative guidance (gyousei shidou), in which ministries play a coordinating role between public and private actors. The AI governance guidelines, consolidated by METI and MIC into a unified document published in April 2024, function as a reference framework that companies are strongly incentivised to follow, without being legally bound to do so. In practice, failure to comply with these guidelines exposes companies to informal regulatory pressure, difficulties in obtaining public contracts, and reputational risks in a market where institutional trust is a strategic asset.

The role of the AI Strategy Council, created in May 2023 within the Prime Minister's Cabinet in response to the rise of generative AI, deserves emphasis. This body brings together representatives from academia, industry, and the administration to formulate recommendations on national AI policy. As early as December 2023, the Council published draft guidelines aligned with the Hiroshima Process, then oversaw the publication of the unified guidelines for companies in April 2024. The Council plays an inter-ministerial coordination role that is all the more important given that Japanese AI governance is distributed across several ministries with sometimes overlapping competences.

It should be noted that this flexibility does not exclude targeted legislative interventions. Japan has a robust personal data protection law (APPI), strengthened in 2022, whose scope covers many AI applications involving personal data. The Telecommunications Act, the Act on Specified Commercial Transactions, and the provisions of the Civil Code on tortious liability also apply to AI systems within their respective domains. The absence of a framework law on AI therefore does not mean the absence of a legal framework. It means that Japan has chosen to mobilise existing instruments rather than create new ones.

The Hiroshima Process: an international ambition

Japan's G7 presidency in 2023 gave rise to the Hiroshima AI Process, which to date represents Japan's most visible contribution to international AI governance. This process produced two main outcomes: an international code of conduct for developers of advanced AI systems, and a set of guiding principles for organisations that develop, deploy, or use such systems.

The choice of Hiroshima as a symbolic setting was not incidental. By associating AI governance with a place that embodies the consequences of a poorly governed technology, Japan framed AI regulation within a perspective of preventing major risks, while maintaining a pragmatic and action-oriented tone. The Hiroshima Process also allowed Japan to position itself as a mediator between the divergent regulatory approaches of G7 members, proposing a framework flexible enough to accommodate both European regulation and the American preference for self-regulation.

This mediating role reflects a diplomatic expertise that Japan has cultivated for decades. In trade, environmental, and security negotiations, Japan has often occupied an intermediary position between blocs with diverging interests. The Hiroshima Process extends this tradition into the digital domain, drawing on the credibility Japan has earned as a technologically advanced democracy, a G7 member, and a reliable partner for both the United States and European countries.

However, the process has structural limitations. The code of conduct is voluntary. It has no verification or enforcement mechanism. Its implementation depends entirely on the goodwill of the signatory companies. Since 2024, several observers have noted that major technology companies' commitment to these principles remains uneven, in the absence of concrete incentives for compliance. Japan is working to strengthen monitoring mechanisms, notably through the framework of the Global Partnership on Artificial Intelligence (GPAI), but the question of enforceability remains open. This is, moreover, a challenge shared by most international initiatives in this field: the difficulty of reconciling the flexibility needed to achieve broad consensus with the rigour needed to produce concrete effects.

Society 5.0 and the Japanese perception of AI

To understand Japan's regulatory posture, it is essential to situate it within the broader framework of the Society 5.0 vision, which the government has been advancing since 2016. This vision imagines a society in which digital technologies, and AI in particular, are integrated into all aspects of social life in order to address the country's demographic, economic, and environmental challenges.

Japan faces unprecedented demographic ageing. The working-age population decreases every year. Labour shortages affect sectors as varied as healthcare, transport, agriculture, and construction. In this context, AI is not perceived as a threat to employment, but as an economic and social necessity. This reading of the stakes colours the regulatory posture: the Japanese government considers that excessive regulation of AI could slow the adoption of technologies that the country structurally needs. The regulatory framework is therefore designed to accompany the deployment of AI rather than to constrain it. This is an explicit calculation: the cost of too slow an adoption of AI is judged to be greater than the cost of the risks that such adoption entails.

The concrete applications of AI within the Society 5.0 framework illustrate this orientation. In the healthcare sector, AI systems are already being used for the early detection of cancers, the analysis of medical images, and the management of hospital patient flows. In transport, experiments with autonomous driving are multiplying in rural areas where public transport services are insufficient for an ageing population. In agriculture, AI systems help farmers optimise irrigation, fertilisation, and harvesting in a context of continuous reduction in agricultural labour. These deployments take place within a framework of trust that rests more on the relationship between regulator and company than on formal legal obligations.

It is worth noting that public perception of AI in Japan differs markedly from what is observed in France and more broadly in Europe. Japanese popular culture — through manga, animation, and literature — has for decades maintained a relationship with technology and machines that is less marked by suspicion than by curiosity and familiarity. Robots and artificial intelligences are not systematically portrayed as threats in the Japanese imagination. This cultural factor, while not determinative on its own, contributes to a social climate more favourable to AI deployment and less conducive to defensive regulatory mobilisation.

This orientation does not mean that Japan ignores the risks. METI's guidelines explicitly cover issues of transparency, fairness, privacy protection, security, and accountability. But the approach remains fundamentally different from that of the European Union: whereas the European regulation classifies AI systems by risk levels and imposes proportionate obligations, Japan prefers to allow each sector to define its own standards according to its operational realities.

Tensions and recent developments

The Japanese approach is facing growing pressures, both internal and external.

On the domestic front, the rise of generative AI has exposed gaps in the existing framework, particularly on questions of copyright. The Japanese Copyright Act, which permits the use of protected works for training AI models without the explicit consent of rights holders (Article 30-4), has drawn fierce criticism from Japanese creative industries. Manga, animation, video games, and music represent a considerable economic sector and a major vehicle for international cultural influence. Japanese creators and publishers fear that insufficient protection against generative AI models will devalue their work and erode the economic models that sustain these industries. The government has initiated consultations on this matter but has not yet amended the legislative framework. This is a case that illustrates the fundamental tension inherent in the Japanese approach: the regulatory flexibility that facilitates innovation can also leave legitimate interests insufficiently protected.

On the external front, the adoption of the European AI regulation creates indirect but real pressure. Japanese companies operating in the European market will need to comply with the requirements of the AI Act, which could progressively align certain Japanese practices with European standards. This regulatory convergence effect is comparable to what has been observed in the area of data protection since the GDPR came into force. Several major Japanese companies, including NEC, Fujitsu, and NTT, have already begun adapting their internal processes in anticipation of these requirements — not only for their European compliance but also for their domestic operations, out of a desire for organisational consistency.

Furthermore, the geopolitical competition around AI, in particular between the United States and China, places Japan in a position that requires delicate balancing. A strategic ally of the United States and a partner in restrictions on the export of advanced semiconductors to China, Japan must at the same time contend with the fact that China remains a leading trade partner and an inescapable actor in the Asian AI ecosystem. Japan seeks to preserve its own technological autonomy and industrial interests while honouring its commitments within existing alliances. Navigating this course requires diplomatic finesse whose regulatory implications are often underestimated in Europe.

These converging pressures have led the Japanese government to take a significant step forward. In May 2025, the Diet adopted the Act on the Promotion of Research, Development, and Utilisation of Technologies Related to Artificial Intelligence. This text, often referred to as the Basic AI Act, is not an equivalent of the European regulation: it contains neither detailed obligations nor financial penalties. It is a principles-based law that establishes a national strategic seat for AI within the Cabinet, coordinates inter-ministerial action, and affirms Japan's commitment to transparency, the protection of rights, and international cooperation. The choice of a basic law rather than a prescriptive regulation confirms Japan's orientation: to frame AI through promotion and coordination rather than prohibition. This development opens new opportunities for dialogue with Europe, and with France in particular.

Prospects for Franco-Japanese cooperation

France and Japan possess complementary assets for cooperating on AI governance. France, through its leading role in developing the European regulation and in organising the AI Action Summit in February 2025, has acquired recognised expertise in normative regulation and diplomatic mobilisation. Japan, thanks to the Hiroshima Process and its experience of flexible governance, brings a different but equally legitimate perspective, grounded in sectoral consultation and continuous adaptation.

Three axes of cooperation deserve particular attention.

The first concerns regulatory interoperability. As the European regulation comes into force and Japanese guidelines are strengthened, it becomes crucial to ensure that companies can comply with both frameworks without excessive costs or practical contradictions. A technical dialogue between French and Japanese regulators on mutual recognition of conformity assessments and certifications would help reduce regulatory friction and facilitate trade in the digital sector. Japan and the European Union have already established a mutual recognition framework for data protection (the adequacy decision of 2019). An analogous arrangement for AI, even a partial one, would constitute a useful precedent.

The second axis concerns the governance of computing resources. The computing capacities required for the development of advanced AI models are concentrated among a small number of actors, predominantly American. France and Japan, both of which are investing in the development of their national computing infrastructures, share a common interest in diversifying semiconductor supply chains and in developing governance frameworks for the allocation and oversight of computing resources used for AI. Recent work on market mechanisms for compute governance, inspired by carbon emissions trading markets, opens avenues that both countries could explore jointly. Japan, which hosts key actors in the semiconductor value chain and has launched an ambitious programme to revive its advanced chip manufacturing capacity, is a natural partner for France on these questions.

The third axis concerns the cultural and creative dimension of AI regulation. France and Japan are two major cultural powers, with creative industries that contribute significantly to their economies and their international influence. The questions raised by generative AI with regard to copyright, the protection of creative works, and the remuneration of creators affect both countries in comparable ways. A common position on these issues, advanced in international forums, would carry considerable weight and could influence the development of norms that protect creators without hindering technological development.

Beyond these three axes, mention should be made of the potential for Franco-Japanese cooperation in multilateral settings. France and Japan are both members of the G7, the G20, and the OECD. They actively participate in the Global Partnership on Artificial Intelligence (GPAI) and in the work of the International Organisation for Standardisation (ISO) on technical standards relating to AI. In these forums, close coordination between Paris and Tokyo would allow them to advance positions that reflect a balance between regulatory ambition and operational pragmatism — a combination that is often lacking in international debates where European and American positions tend to occupy most of the spectrum.

The France-Asia Foundation, through its programmes and networks, is well placed to contribute to building this dialogue. Relations between France and Japan in the digital domain do not lack institutional relays, but they sometimes suffer from a lack of continuity and follow-through over time. The networks of young professionals and decision-makers being built around the Foundation can serve as a transmission belt between high-level diplomatic exchanges and concrete cooperation on the ground — whether in the form of joint research projects, business missions, or cross-training programmes.

Conclusion

Regulating artificial intelligence cannot be reduced to a binary choice between binding regulation and laissez-faire. The Japanese experience shows that intermediate paths exist, founded on institutional trust, sectoral consultation, and continuous adaptation. These approaches are not without weaknesses, particularly on questions of enforceability and the protection of fundamental rights. But they offer valuable lessons for the design of durable governance frameworks in a rapidly evolving field.

For France and Japan, the complementarity of their approaches is not an obstacle to cooperation. On the contrary, it is what makes cooperation both necessary and fruitful. Both countries have the opportunity to build a regulatory dialogue that aims not at uniformisation, but at interoperability and mutual learning. In a domain where governance failures can have lasting consequences, this ambition is commensurate with the responsibilities that France and Japan bear.

This dialogue cannot be confined to official forums. It must also draw on exchanges between researchers, engineers, legal experts, entrepreneurs, and creators from both countries. AI regulation is too important and too complex to be left to governments alone. It is a collective undertaking — and one in which the relationship between France and Japan has much to offer.

This publication reflects the views and opinions of the individual authors. As a platform dedicated to the sharing of information and ideas, our objective is to highlight a diversity of perspectives. Accordingly, the opinions expressed herein should not be interpreted as those of the Fondation France-Asie or its affiliates.