Japan’s AI regulations for business

aI's development is accelerating exponentially. Although its regulation is in its infancy, one dramatic innovation seems to be leading to the next at a breakneck pace. Generative AI technologies for creating art, literature, journalism reports, translations, and even academic papers are just one amazing example.

Historically, the rise of ingenious technologies has raised social, political and legal issues, and due diligence is required to ensure that new technologies are a net positive for society. AI is perhaps the most influential invention in human history, and while businesses are already reaping the benefits of intelligent automation, AI will also bring about a myriad of civilization-defining issues, including human rights, ethics, privacy, bias, and worker rights. It is not an exaggeration to say that it causes.

There is no global regulatory framework for AI, and so far liberal democracies have adopted different approaches aimed at balancing risks and benefits. A common concern is that overregulation has a negative impact on innovation.

Still, in December 2023, the EU opted for comprehensive and binding regulation through the AI ​​Act. This is a system targeted at specific digital environments, overseen by the European AI Secretariat. The UK has chosen a lighter, pro-business approach that divides governance between existing regulators. The United States is evolving along the lines of the United Kingdom, requiring a federal AI bill of rights and a recent executive order directing federal agencies to develop AI regulatory plans.

Japan is developing its own approach to AI regulation. Japan has traditionally adopted a civil law system, but in some areas of law it follows the common law approach of the UK and US, with ministries and agencies issuing guidelines and current laws governing the emerging legal issues of AI. are doing.

industry guidelines

In January 2024, the Ministry of Internal Affairs and Communications and the Ministry of Economy, Trade and Industry published the “Draft AI Guidelines for Business,” which updated and integrated the previous principles and guidelines, and the G7 subsequently compiled the “Hiroshima AI Process Comprehensive Policy Framework.” Hiroshima Summit in May 2023.

These guidelines are intended for developers, providers, and users of AI in business and government. Users for purposes other than business activities are not eligible. The same goes for data providers, as developers and providers should be responsible for handling data.

These guidelines outline three fundamental values ​​for an AI society. Diversity and inclusion. and a sustainable society. These are centered around ten common guiding principles: human-centeredness, safety, fairness, privacy protection, security, transparency, accountability, education/literacy, fair competition, and innovation. These guidelines call on AI developers, providers and users of AI for business purposes to follow common guidelines and care for people and the environment by:

• Combating bias in the development and use of AI through data collection and management procedures and algorithmic means.
• Security measures (such as ensuring verifiability during AI development).
• Protection of privacy such as inappropriate input and personal information.and
• Agile governance, where environmental/risk analysis, goal setting, system design, operations, and evaluation are continuously and rapidly rotated by multiple stakeholders, rather than fixed rules and procedures. The draft guidelines include governance action items and an appendix with practical examples.

Business actors involved in advanced AI are encouraged to follow the Hiroshima Process International Guiding Principles for All AI Stakeholders and the Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems. Developers of advanced AI systems must also comply with the Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems.

Sector-specific guidelines have also been published, such as the use of generative AI by education and healthcare providers. Although the regulations are not binding, AI stakeholders in Japan are expected to understand and follow these and other relevant guidelines.

Current law on AI issues

In addition to the guidelines, Japan has governed AI through existing laws rather than industry-specific laws. The development, deployment, and use of AI, like any other business activity, may be subject to regulations such as copyright laws, privacy laws, unfair competition laws, antitrust laws, and the Economic Security Promotion Act. , Attorney Law. Here are some examples:

Copyright law. AI raises questions regarding infringement in learning/development, infringement in production and use, and copyrightability of products created by AI. For example, AI knowledge construction utilizes information that may be subject to copyright.

As a general rule, permission from the copyright holder is required to use copyrighted materials. A narrow exception has been established: permission is required for such use unless the purpose is to “enjoy oneself or allow another person to enjoy the ideas and emotions expressed in the work.” there is no. However, if the user has an entertainment purpose (e.g., overfitting), the user may be subject to claims of copyright infringement, including monetary damages and injunctive relief (e.g., destruction of the trained model). there is.

Infringement in the creation and use of AI will be judged using traditional infringement criteria (similarity and dependence on existing copyrighted works), as if it were created by a human without using AI. . Breaches during the generation and use stages can expose not only the AI ​​user but also the provider of the generated AI to liability. The copyrightability of AI-generated works follows the traditional interpretation of copyrighted works. If the instructions to the generated AI are just ideas that do not lead to expression, they will not be considered creative and will not be considered copyrighted works.

Personal Information Protection Law. The current Personal Information Protection Act applies to the use of personal information by AI. When providers of generative AI (such as OpenAI, which provides ChatGPT) acquire personal information for the purpose of developing generative AI, they must clearly indicate the purpose of use and ensure that the actual usage does not deviate from the personal information. I need to get the information. Make the purpose of use clear and do not violate rules for obtaining confidential information such as race or creed. In June 2023, we notified OpenAI in Japanese of the purpose of use of personal information and warned them to appropriately acquire sensitive information.

Users of generation AI also need to be careful with their personal information. For example, a company that manages personal information. Personal information about our employees may be entered when using generated AI to improve work efficiency. In this case, you must be careful not to violate the law.

When personal data is used for training of generative AI, the person who entered the data provided the personal information to a third party (in the case of using foreign generative AI, the personal information was transferred to a foreign country) may be considered. illegal.

Competition law. The government is strictly monitoring AI for antitrust violations. In March 2021, a report titled “Algorithms, AI, and Competition Policy” sponsored by the Fair Trade Commission was released. Pricing and price research algorithms, depending on how they are used, can encourage price competition, while competitors can also use AI to coordinate price adjustments, disrupt trade between competing users, and defraud consumers. It states that there is a possibility of erroneous pricing, intentional ranking manipulation, etc.

Economic Security Promotion Act. In May 2022, AI was designated as a “Specific Important Technology” (including use by foreign countries for the purpose of harming national security). As a result, government support is now available to accelerate AI research and development. At the same time, measures to prevent technology leaks overseas, protection of trade secrets, and ensuring sound and fair research are also required.

Japan's regulatory scheme for AI may seem piecemeal or insufficient to some, but it has thoughtful government and industry guidelines, and this light touch of applying existing laws to AI scenarios. The approach reflects an intention to protect society and individual rights without suffocating AI. It can spur innovation and undermine global competitiveness.

Although this policy is not legally binding, it actively addresses guidelines that AI developers, providers, and users would be wise to follow. When AI-related legal disputes arise, whether the parties follow the guidelines can affect the outcome. Although this “wait and see” strategy may seem reactive as AI is introduced into Japanese society, it is flexible and can accommodate the realities of AI in Japan.

So Sato Law Office
3-4-1 Tokyo New Kokusai Building No. 937
Marunouchi, Chiyoda-ku, Tokyo 100-0005
Phone number: +81 3 6275 6080
Email: plc@innovationlaw.jp
innovationlaw.jp/ja/