Government's first AI Strategy Council meeting: Prime Minister directs "In-depth discussion for identifying benefits and risks."

While the rapid evolution of generative AI has harnessed potential applications in a variety of fields, there are growing concerns about risks such as information leakage and disinformation. The government has reorganized the AI Strategy Implementation Council and newly launched the AI Strategy Council (Chairperson: Professor Yutaka Matsuo of the Graduate School of Engineering at the University of Tokyo). The AI Strategy Team is already in place for coordinating at the working level in the ministries concerned.

Prime Minister Fumio Kishida attended the first meeting and stated, "In the AI Strategy Team set up under the AI Strategy Council, ministries should work together closely under the leadership of the Special Advisor to the Prime Minister, Hideki Murai, based on today's discussion, and move forward quickly with its wide range of missions to maximize AI potential benefits and respond to risks posed by it." The discussions and results of the meeting will be reflected in consideration of the Integrated Innovation Strategy, the Basic Policies for Economic and Fiscal Management and Structural Reform, and the development of international rules.

The government's AI strategy was created in 2019 and has since been revised twice for the 2021 and 2022 annual editions. RIKEN, the National Institute of Advanced Industrial Science and Technology (AIST), and the National Institute of Information and Communications Technology (NICT) have played a principal role in the research and development of AI, and the practical application of the technology has been promoted by various ministries and private companies.

Japan is, however, lagging behind other countries in the development of ChatGPT and other generative AIs, as well as in its applications and rulemaking. Following the secretariat's explanation of how the use of AI should be promoted, how concerns and risks should be addressed, and how development should proceed, various opinions were expressed by each council member.

Regarding AI, not only the technologies themselves but also the social implications must be discussed and International coordination is strongly needed in future discussions. The methodology for controlling AI would range from law enforcement (stringent approach) to guidelines and self-regulation (moderate approach), but an argument in favor of choosing one or the other is not productive. Even if regulations are introduced, technology is advancing so quickly that these regulations may be unable to keep up. When discussing the challenges we face concerning generative AI, it is necessary to clarify what general problems derive from AI.

Japan's medium- to long-term technological capabilities for the development of large-scale language models should be sustained. This development cannot be managed by a conventional national project system. With rapid changes anticipated, the government's budget scheme, in which an estimate for a research and development project is requested in the summer and the project begins the following year, cannot keep up with technological progress.

On the basis of these various views, the AI Strategy Team will identify issues that need to be addressed, and will flesh them out and then finalize them into policy. Clarifying the legal nature of AI uses under the framework of copyright laws: some cases are subject to criminal penalties. The AI Strategy Team met a few days later at the council meeting to discusss the relationship between AI and copyright as one of the responses to specific issues.

In the development and learning stages of AI, the current copyright law allows, in principle, the collection and reproduction of copyrighted works as training data, the creation of training data sets, and the use of such data sets for training to develop AI without the permission of the copyright holder.

In the generation and use stage of images, it would, however, constitute copyright infringement if similarity or dependence are found between images generated by using an AI and existing works, and such images are uploaded for publication or copies or sold. In other words, uploading AI-generated material to social networking sites without exercising caution could unknowingly infringe on someone's copyright and, in some cases, be subject to criminal penalties.

The Agency for Cultural Affairs will inform the public through seminars and other means, as well as organize discussions and raise awareness and understanding of the issues concerning generative AI with the assistance of criminology scholars, lawyers, and other experts.