Table of Contents
Available Downloads
This report is part of Strategic Japan, a CSIS Japan Chair initiative featuring analysis by Japan’s leading foreign policy scholars on key regional and global challenges and the implications for the U.S.-Japan alliance.
Introduction
Artificial intelligence (AI) governance systems vary based on countries’ cultural and social attributes. Influenced by its culture and social background, Japan’s AI governance policy focuses on promoting innovation, taking an agile, “soft-law approach” to governance, and emphasizing international interoperability of AI systems.
The Japanese government recently strengthened its legal efforts regarding AI by introducing legislation in the Diet (parliament) at a time of dramatically shifting debate on global AI governance policies in 2025. With the start of the second Trump administration in the United States and the AI Action Summit held in Paris in February 2025, the direction of global AI policies is shifting rapidly in favor of innovation, and Global South countries are beginning to participate in discussions about global AI policies.
Japan’s AI governance policy focuses on promoting innovation, taking an agile, “soft-law approach” to governance, and emphasizing international interoperability of AI systems.
This white paper presents the issues that Japan should address in its AI governance policies and the direction of future collaboration between Japan and the United States. The paper begins with a review of the evolution of Japan’s AI governance strategy and a summary of recent developments in Japan’s AI policy debate. It then identifies current trends in the debate on global AI governance policies and offers recommendations on the future direction of Japan’s AI governance policy and cooperation with the United States.
An Overview of Japan’s AI Governance Policy
A History of Japan’s AI Policy Before Generative AI
From AI Technology Strategy to AI Strategy 2022
Historically, Japan has been a country with a strong orientation toward technology and industrial policy—particularly innovation policy—led by the Ministry of Economy, Trade, and Industry (METI). This innovation focus applies in the field of AI technologies as well. Taking the global lead, Japan has already started to formulate AI strategies aimed at promoting innovation since the mid-2010s.
Structural problems within the Japanese government explain the multiple revisions of the country’s AI strategy.
As public interest in AI increased around 2015, the Japanese government began to implement some measures regarding AI. The first steps included developing an R&D (research and development) system, studying the impact of AI on Japan’s industrial structure, and promoting the concept of “Society 5.0.,” a Japanese vision of future society leveraged by new technologies. Instructed by Prime Minister Shinzo Abe, the Japanese government announced its first AI technology strategy in March 2017. After the strategy’s execution went somewhat astray due to inconsistent government frameworks, a new advisory body was established, publishing a second AI strategy that focused on human resources and R&D in June 2019. Upon further review, the government issued a revised AI strategy in April 2022 that shifted the focus to the deployment of AI systems in society (Table 1).
Structural problems within the Japanese government explain the multiple revisions of the country’s AI strategy—specifically, only the Ministry of Internal Affairs and Communications (MIC) and METI have any expertise on AI technologies, and the expertise they do have is limited in the government. In particular, the Cabinet Office (CAO), which plays a central role in developing strategies for the entire government, relies on short-term transfers of human resources, mostly from the MIC and METI.
History and Priorities of Japan’s AI Governance Policy
To date, Japan’s AI strategies have not emphasized ethics or governance as much as other countries. In fact, references to AI governance in Japan’s 2022 strategy are overwhelmingly low when compared to equivalent strategies of other countries announced between 2020 and 2022 (Table 2).
However, this does not mean that the Japanese government was unaware of the ongoing global debates over AI ethics and governance. Rather, the Japanese public was just not as interested in AI ethics and governance. Japan had been studying AI ethics for years. The MIC published guidelines for R&D and the use of AI in 2017 and in 2019, respectively; the CAO released a report on AI and human society in March 2017 and published “Social Principles of Human-Centric AI” in 2019 to address the pressing issues surrounding the Organisation for Economic Co-operation and Development’s (OECD) AI Principles.
Trends in global AI governance shifted “from principle to practice” after the release of the OECD AI Principles, and METI took the lead on AI governance in Japan. A 2021 report hinted at nonbinding guidelines from the perspective of “governance innovation.” The following January, METI released “Governance Guidelines for the implementation of the AI Principles ver. 1.1.” The 2022 governance guidelines marked the confirmation of Japan’s so-called “soft-law approach” to AI governance: promoting voluntary governance efforts by businesses through the release of nonbinding guidelines by the government.
AI Governance Policy After Generative AI
The release of ChatGPT at the end of 2022 generated interest in Japan regarding AI risks. In May 2023, the CAO established a new AI Strategy Council, which quickly published a “Tentative Summary of AI Issues.” The summary primarily focused on addressing potential risks posed by generative AI and, with that focus, differs from Japan’s previous AI strategies.
However, the 2023 summary still emphasized the importance of promoting voluntary efforts by businesses (the soft-law approach) as an immediate objective. In June 2023, the council instructed the MIC and METI to merge existing AI guidelines and to update them with generative AI in mind. The official result of these efforts was published in April 2024 as the “AI Guidelines for Business.” (These guidelines are in line with the international code of conduct formulated by the Hiroshima AI Process after the 2023 G7 summit hosted by Japan.) A revised version of the guidelines was published in March 2025.
In early 2024, the ruling party began to consider legislation involving AI regulations, basing their proposals on trends in AI governance policies in the European Union and the United States. Details are described below.
Approaches of Japan’s AI Governance Policy
Cultural Background of Japan’s AI Governance System
Two Types of Risk: Real Social Risks and Artificial General Intelligence (AGI) Risks
In discussing AI governance, it is important to distinguish what societies generally perceive as two different types of risk posed by AI technology.
The first type consists of real social risks caused by the use of technology that conflicts with shared social norms, such as ensuring safety and protecting human rights. These risks are not unique to AI technology. Innovative technologies in general bring real social risks, accentuating the need for technology governance more broadly. These risks become apparent and recognized as social problems whenever new technology diffuses into society.
The second type of risk centers on anxiety about future AGI technology—specifically, the fear that current AI technology will develop into AGI in the future and, as a result, threaten the survival of humanity (so-called existential risk). This risk perception is specific to AI technology and emerges as technological advancement proceeds, typically at the so-called “hype” stage immediately before technology diffuses into society. Because of perceived existential risk, the need for governance can be imperative for some societies, and may result in ex ante regulations for AI development based on precautionary principles.
Japan’s Perceptions of Future AGI
There is a significant difference between Japanese and Western public perceptions of anxiety concerning future AGI. AI governance researchers in Japan have identified several factors to explain this, including differences in religious background and varying popular culture views on human and AI (machine) relationships (Table 3).
In short, the Japanese people are relatively more optimistic about the future social impact of AI technology than Western populations are. Thus, there is almost no discussion in Japan about introducing strong ex ante regulations across sectors from the perspective of the precautionary principle. This difference in social perception of AI risk is thought to be the reason why Japan’s AI strategies emphasize a soft-law approach as policies for AI ethics and governance. Japan’s perception of AI risk might parallel that found in China, India, and other Asian countries.
Approach 1: Soft-Law Approaches
Promotion of Voluntary Efforts Through Soft-Law Approaches
One of the characteristics of Japan’s AI governance system is that it consistently maintains the so-called soft-law approach, a policy of promoting voluntary initiatives by businesses based on nonbinding guidelines from the government, rather than introducing binding legal regulations (hard law) as an initiative applicable to all sectors related to the development and use of AI.
The main reason for this difference is that Japan does not share the West’s anxiety about future AGI. In addition, ex ante regulation based on the precautionary principle is quite difficult to enact in the Japanese legal system, and the rigidity of laws and regulations makes agile governance difficult and strongly inhibits innovation, which, as mentioned previously, is a core focus of Japanese AI policy.
One of the characteristics of Japan’s AI governance system is that it consistently maintains the so-called soft-law approach, a policy of promoting voluntary initiatives by businesses based on nonbinding guidelines from the government.
A soft-law approach makes agile governance possible for business, but that only applies to the reduction of unintentional risks, especially by large companies, since guidelines often do not reach small- and medium-sized enterprises (SMEs), and malicious actors will not comply with the guidelines. A soft-law approach may work in Japan because most Japanese businesses accept and comply with such norms as part of the country’s social culture of government-industry relations.
Balance with Regulations in Specific Fields
In specific sectors where real social risks are already apparent due to the diffusion of AI technologies, the Japanese government does discuss introducing regulations. In the application sectors of AI technologies such as finance, autonomous driving, and medical care, where significant impacts on life and physical safety can be predicted when a system malfunction occurs, the need to review and modify existing regulations has been actively discussed and even implemented. For example, in the field of autonomous driving, legal reforms concerning safety rules made it possible for unmanned vehicles to be operated on public roads in 2020. The Digital Agency, a Japanese administrative agency established in 2021, is continuing to work with relevant ministries and agencies to update its strategies and safety rules in light of technological advances.
Policy and legal discussions are also underway in other areas where real social risks from generative AI are apparent, especially challenges associated with copyright issues. As for the risks related to dis/misinformation by generative AI, Japan experiences this threat primarily in the wake of natural disasters and internet fraud; countermeasures to these challenges are being discussed, not from the context of AI regulations, but rather from the context of digital platforms regulations.
Approach 2: Interoperability as an International Contribution
Japan’s Emphasis on Interoperability for Global AI Governance
Another characteristic of Japanese AI governance strategy is an emphasis on international interoperability. In fact, Japan has actively worked to ensure interoperability of AI governance systems around the world, especially by using the frameworks of the G7 and OECD.
Japan has been actively involved in the formulation of OECD AI Principles since 2016, and it has played a leading role in the Hiroshima AI process since the emergence of generative AI. In addition, after the establishment of AI Safety Institutes (AISIs) in the United Kingdom and the United States at the AI Safety Summit in 2023, Japan responded swiftly, announcing its own AISI the following month. These actions suggest high interest in establishing international standards related to AI governance.
Notably, the first description in the “AI risk” chapter of various AI strategy documents is about Japan’s “international contribution.” In other words, Japan’s AI governance strategy is more concerned with ensuring interoperability than with reducing domestic risks caused by AI technologies.
Japan’s Neutral Position in Global AI Governance
Japan’s interest in interoperability reflects a recognition that it does not have the ability to set de facto standards like an AI superpower such as the United States could, nor can it promote global standardization processes like the European Union. Japan has therefore focused its efforts on interoperability, seeking not to intervene in the formation of regulations or guidelines introduced by the European Union or the United States but rather to ensure that core principles for interoperability are agreed upon internationally.
In this regard, Japan is in a relatively advantageous position for advancing interoperability within the global AI governance system. While Japan is a part of the G7 as a major developed country, it has a relatively neutral position vis-à-vis other actors. Moreover, since Japan is the only Asian country in the G7, it could play a leading role in alleviating concerns about Western bias in developing global standards for interoperability.
Recent Trends in Japan’s AI Governance Policies
Japan’s New AI Act
Background
Though Japan has long taken a soft-law approach to AI regulation, as mentioned above, discussions involving the (hard law) legal system emerged in early 2024 as regulatory considerations for advanced AI models evolved in the European Union and the United States.
In February 2024, the AI Project Team of the Japanese Liberal Democratic Party (LDP) presented an outline of a bill that would impose notification and reporting requirements for businesses specifically related to advanced AI systems. However, the motivation for the proposal was not necessarily to reduce the risks caused by AI systems in Japan, but rather to not be criticized as a country with inadequate legislation when social problems arise in the world due to advanced AI systems. In other words, the LDP’s proposal can be interpreted as part of a pattern of promoting international interoperability of the legal system in the fields of advanced AI systems.
The LDP recommended that the government develop the minimum necessary legal framework for AI models that pose extremely high risks while also emphasizing the importance of more general guidelines. At a meeting of the AI Strategic Council in December 2024, Prime Minister Ishiba stated that Japan will build an AI legal system that will be “a model for the world.” In February 2025, a bill known as the AI Act was approved by the Ishiba Cabinet and submitted to the Diet.
The official name of the AI Act—the Bill on Promotion of R&D and Utilization of AI-related Technologies—indicates that the focus of the bill is promoting AI innovation. Moreover, the bill is a so-called “basic law” and does not include any new legal regulations or penalties, instead outlining the establishment of a new AI Strategic Headquarters within the government, the direction of the government’s basic measures, and the responsibilities of private companies and other stakeholders involved (Table 4).
As for basic measures related to AI governance, this bill includes the development of guidelines in line with international norms, analysis of cases of infringement of individuals rights, consideration of countermeasures, and guidance for businesses operating AI systems. In this sense, the AI Act follows Japan’s AI governance policy to date: promoting voluntary efforts through guidelines consistent with international norms; responding to risks and reviewing ex post facto regulations within existing legal frameworks in principle; and promoting international cooperation for interoperability.
Implementing Japan’s AI Act
The AI Act was passed by the Diet on May 28, 2025, and came into effect on the same day. However, the government could face major issues after its promulgation, namely a lack of requisite expertise to implement the actions stipulated in the bill.
In order to implement the AI Act—including measures related to AI governance—it is essential to quickly strengthen the AI-related capabilities of government organizations by recruiting new talent and training existing officials.
Since civil servants in the Japanese government are, in principle, employed for life, there are few officials with knowledge of AI technology, with the exception of some in the MIC and METI. In order to implement the AI Act—including measures related to AI governance—it is essential to quickly strengthen the AI-related capabilities of government organizations by recruiting new talent and training existing officials.
Another concern is that the establishment of the new AI strategic headquarters may create new vertical divisions and fragmentation of authority within the cabinet. It will be necessary to develop an institutional framework to advance cross-sectoral strategic planning in the cabinet to promote a comprehensive digital innovation strategy.
The Recent Transformation of Global AI Governance
A Shift from AI Safety and Risks to AI Opportunity and National Interest
The landscape of global AI governance has transformed drastically since the beginning of 2025. Specifically, the priorities of the second Trump administration and the themes addressed at the Paris AI Action Summit in January–February 2025 suggest emphasis on AI innovation instead of AI governance, and countries around the world are now shifting their policies in favor of AI innovation.
These shifts include promoting investment in AI (as seen in the United States, France, and the European Union) and deregulation (as seen in the United States, Canada, the United Kingdom, and the European Union). However, as each country emphasizes the promotion of its own national interests, conflicts among major developed countries are coming to the surface—in particular, tension between the United States and the European Union under the Trump administration, which were highlighted when Vice President JD Vance strongly criticized EU digital and AI regulatory policies at the Paris summit. There also seems to be a divergence between the United Kingdom and France, since the United Kingdom refused to sign the summit’s declarations. Under these circumstances, there is concern about the future capacity of the G7 to promote international cooperation on AI governance.
Another trend in the debate on global AI governance is the increased participation of countries in the Global South. India served as a co-chair of the 2025 Paris AI Action Summit and will chair the next AI summit; its presence is expected to continue to increase in the future. China, an AI superpower, has been advancing a standardization strategy domestically, promoting a resolution on AI capacity building at the United Nations, and actively engaging developing countries in the Global South.
Japan’s Coordinating Role in a Period of Transformation
Although global AI policies are shifting from governance toward innovation, the former will remain important due to the potential social risks of AI. As a promoter of a soft-law approach domestically and interoperability globally, Japan’s coordinating role on the diplomatic stage will be increasingly important.
If tension among G7 members continues, Japan, which is in a relatively neutral position and has been actively working on the Hiroshima AI process in recent years, can coordinate bilaterally with the United States and the European Union and encourage a consensus position on AI governance.
Japan can also play a role in strengthening relations with countries of the Global South, including India, with which it has developed strategic ties in recent years. Japan established the Friends Group of the Hiroshima AI Process in May 2024, and will continue to prioritize outreach to the Global South as China tries to enhance its own influence in the developing world. But the capacity challenges in the Japanese government noted above will have to be addressed in order to sustain Japan’s global leadership role in AI governance.
U.S.-Japan Cooperation on AI Governance
U.S.-Japan Collaboration on AI During the Biden Administration
U.S.-Japan collaboration on AI strategies progressed steadily during the Biden administration. The April 2024 Kishida-Biden joint statement specifically described Microsoft’s investment in Japan and cooperation in the field of AI by research institutes in both countries. AI governance strategies were discussed in the “Japan-U.S. Dialogue on the Digital Economy.” However, most of the discussions on AI governance focused on cooperation through multilateral institutions such as the G7 and the OECD.
In the context of economic security, the Japan-U.S. Economic Policy Consultative Committee (the so-called economic “2+2”) has been underway since July 2022, and the second joint statement (November 2023) includes AI as an emerging and critical technology, emphasizes G7 collaboration on AI governance, and encourages collaboration between the National Institute of Standards and Technology (NIST) and Japan’s Information Technology Promotion Agency (IPA) as partner AISIs.
U.S.-Japan Collaboration on AI During the Trump Administration
U.S. relations with major developed countries, such as the members of the European Union, have become more complex since the inauguration of the second Trump administration in January 2025. However, as indicated by the February 2025 summit meeting between Prime Minister Ishiba and President Trump, the U.S.-Japan relationship seems relatively stable as of this writing.
In a joint statement, the two leaders indicated that they will seek to collaborate to lead the world in the development of AI and other critical technologies in order to elevate their economic partnership to the next level. To that end, they instructed the relevant ministers to strengthen U.S.-Japan economic cooperation.
In terms of AI investment, it is noteworthy that SoftBank, Open AI, and others announced the Stargate Project in the presence of President Trump on January 21, 2025; these companies also announced a partnership on “Cristal Intelligence” on February 3, just before the bilateral summit meeting.
The direction of AI policy under the second Trump administration is gradually becoming clear after the repeal of the Biden administration’s executive order on AI policies and the issuance of a Trump executive order for a new action plan. Despite these steps, as the debate surrounding global AI governance undergoes transformational changes, the Trump administration’s AI governance strategy remains uncertain. Given that the United States and Japan are partners that can have a significant impact on global AI governance, the first challenge will be to develop a common understanding of the current situation and then to craft a joint strategy for the future direction of AI governance.
Recommendations
AI Capacity Building for the Japanese Government
The Japanese government proposed the AI Act recognizing the increased importance of AI technologies in society. Strengthening AI expertise across the government is imperative to fully execute the act’s stipulations.
Therefore, the Japanese government should develop a plan to increase the capacity for managing AI—not just at the cabinet level, but across the entire government—by building an ecosystem that includes academia and the private sector. Japan’s coming AI Strategic Headquarters should pursue the following:
- Formulate comprehensive AI, digital, and innovation strategies and relay instructions for execution to government ministries based on extensive and up-to-date knowledge of AI technological and industrial trends.
- Coordinate and publish AI governance guidelines in an agile way through a multi-stakeholder process in line with international norms.
- Enable the entire government to respond quickly to various real social risks and take appropriate countermeasures.
In addition, the government should immediately promote the use of AI in all ministries and develop an initiative to recruit and develop AI talent.
Strengthening Japan’s Coordination Role on Global AI Governance
Japan’s leadership role on AI governance will need to expand beyond the G7 to include following initiatives:
- Bilateral coordination with key players such as the United States and the European Union and enhanced outreach to developing countries of the Global South, such as India.
- Building on efforts not only in the G7 and OECD but also other institutions such as the Quad (U.S.-Japan-Australia-India cooperation mechanism) and the United Nations.
Sharing a Common Understanding of Future Global AI Governance
Because Japan and the United States, as partners, could have significant influence on global AI governance, they must develop a common understanding on AI governance and a shared future vision for global AI governance systems. Priorities should include the following:
- The future development of AI technology and governance in a global context.
- The role of G7 initiatives in global AI governance strategy as well as the involvement of Global South countries such as India.
- Responding to China’s engagement of the Global South with economic security in mind.
In addition, ongoing collaboration between research institutes such as NIST and IPA and between businesses in both countries will form the basis for future cooperation on AI.
Because Japan and the United States, as partners, could have significant influence on global AI governance, they must develop a common understanding on AI governance and a shared future vision for global AI governance systems.
Conclusion
The global AI governance system is currently undergoing a major transformation as the emphasis in the policy debate shifts from AI safety to innovation. The Trump administration played a major role in this transformation, but its global strategy for AI governance remains unclear.
Japan, which has also emphasized innovation and interoperability and has taken a soft-law approach domestically via guidelines and general principles, is preparing to enact its AI Act “as a model for the world” and should be expected to play a coordinating role in the transformation of global AI governance policy.
Under these circumstances, Japan must strengthen its government’s AI expertise, including diplomatic capabilities, to develop a common understanding with the United States about the use of AI technology to support innovation and economic growth, as well as to shape the policy debate on future visions for global AI governance.
Tagui Ichikawa is a specially appointed professor at Hitotsubashi University’s Institute of Innovation Research.
This report is made possible with support from the government of Japan.
