The Hiroshima AI Process: Japan’s Role in Shaping Global AI Governance

Author: Inge Odendaal

In the past year, the debate over artificial intelligence (AI) has shifted from whether to regulate^[1] it to how to regulate it effectively. Governments worldwide are grappling with the challenge of establishing regulations that can keep pace with the rapid developments in AI, respect state sovereignty, and establish international norms and practices. Japan, a global powerhouse known for innovation and resilience, has played a key role in this debate through its position as the G7 chair, the facilitator of the Hiroshima AI Process, and the host of the 18^th United Nations (UN) Internet Governance Forum (IGF) in Kyoto.^[2]  This article provides an overview of the Hiroshima AI Process, outlines the main debates among members and discusses Japan’s role and position in the broader discussion on AI regulation.

Generative AI is a category of AI capable of producing various digital outputs, such as text, images, audio, video, software, and code.^[3] Notable examples include large language models (LLMs) such as ChatGPT^[4], which can generate human-like text-based responses, and Dall-E 2, which can create realistic images based on descriptions.^[5] Generative AI models are available for free use, both privately and commercially, making them accessible to people around the world. This has led to an increase in the number of users and developers of generative AI and has also extended their reach across international borders.

The Hiroshima AI Process: Establishment, Opportunities, and Challenges

The Hiroshima AI Process was initiated in response to the growing recognition of the need for international cooperation in regulating generative AI. At the G7 Digital and Technology Ministers’ Meeting in Takasaki in April 2023, ministers discussed the need for international dialogue on the compatibility of various AI governance frameworks and the opportunities and challenges posed by this technology.^[6] These discussions were elevated to the G7 Summit in Hiroshima in May, where leaders decided to establish the Hiroshima AI Process as a platform for coordinating the formulation of international rules and approaches to AI regulation. Currently, individual G7 member states maintain their domestic regulations and policies, while the Organisation for Economic Cooperation and Development (OECD) has its principles and guidelines.^[7] Since the talks in May, G7 members have been tasked with exploring mitigation measures domestically, drawing inspiration from other countries, to coordinate measures at an international level. ^[8]

To this point, G7 members agree and acknowledge the numerous advantages of generative AI. Beyond its practical implications, this technology has the potential to foster innovation and entrepreneurship, enhance productivity, support healthcare systems, and contribute to addressing global issues, such as climate change.^[9] Generative AI could even play a role in safeguarding and preserving minority languages. However, members have also noted that policy recommendations resulting from the Hiroshima AI Process should focus on the “responsible” use of this technology. ^[10] These models pose significant risks, including the potential for manipulation, copyright infringement, human rights violations and privacy and security breaches. Moreover, they are susceptible to AI hallucinations, which generate misinformation.^[11] Compounding these concerns is the fact that many of these platforms are freely accessible to anyone.

Debates in Regulatory Approaches

The path to regulating generative AI remains the subject of debate amongst G7 member states. Two primary regulatory approaches have emerged, each supported by different G7 members. The first approach is conceptualised as a “hard-law-based” approach, emphasising obligations related to governance, transparency, and security, akin to the European Union’s AI Act.^[12] In contrast, there is the “soft-law-based” approach that prioritises non-binding guidance over comprehensive regulations.^[13] Japan advocates for the latter. The country has always favoured the integration and use of AI, which aligns with its belief that regulation should not stifle creativity and innovation.^[14] Therefore, guidelines surrounding AI in Japan have not been about restricting AI but utilising AI for a society that promotes human dignity, diversity, and inclusion.^[15]

While some may argue that the OECD principles and the outcomes of the Hiroshima AI Process are non-binding, such an assessment overlooks their significance.^[16] These non-binding principles serve as international reference points and establish standards for AI governance. Regulations governing technology that transcends borders hold implications for all stakeholders, ranging from developers, individual users, manufacturers and corporations. Moreover, collaboration among states with a shared interest in safety and security aids in preventing double standards and reduces risks. Japan is aware of the importance of these factors.

Japan’s Engagement

Since the Hiroshima AI Process’ inception, Japan has translated its commitment into tangible actions. To date, the country has taken on the role of facilitator in the process, coordinating various engagements, such as ministerial meetings, and assisting in the publication of OECD’s reports.^[17] Japan also hosted the UN’s Internet Governance Forum in October in Kyoto.^[18] Furthermore, it is not only its international engagements that demonstrate the country’s commitment to becoming a hub for innovation in this field but can also be seen in its exciting domestic AI-related developments. These developments include the integration of generative AI into government processes, ^[19] the recent announcement of an economic package for AI development,^[20] and public-private partnerships working toward creating their own LLM.^[21] 

Looking Ahead

During the IGF, Prime Minister Fumio Kishida announced that the G7 members intend to publish a draft of international rules for generative AI by the end of the year.^[22]  In the rapidly evolving landscape of AI development and regulation, one certainty is Japan’s determination and commitment to play a key role in global AI regulations. This commitment involves promoting innovation and cooperation while simultaneously addressing the multifaceted challenges and risks associated with generative AI.