AI Governance in Japan Ver. 1.1

AI Governance in Japan Ver. 1.1 Report from the Expert Group on How AI Principles Should be Implemented July 9, 2021 Table of Contents 1. Introduction 2. AI Governance Trends in Japan and around the World A. From AI principles to governance B. Risk-based approach C. Assessment and classification of risks D. Architecture of AI governance (1) Goals: AI principles (2) Horizontal intermediate rules (a) Legally non-binding guidelines (b) Legally binding horizontal regulations (c) International Standards (3) Rules focused on specific targets (a) Regulations on specific use (b) Regulations on specific sector (c) Regulations on Use of AI by Government (4) Monitoring and enforcement (a) Monitoring (b) Enforcement E. International harmonization and alignment between layers 3. Ideal approaches to AI Governance in Japan A. Suggestion of Governance Innovation B. Opinions of stakeholders (1) Opinions of industries (2) Opinions heard from the Expert Group and on other occasions (3) Consumer Perspective C. Ideal approaches to AI governance in Japan (1) Legally non-binding corporate governance guidelines (2) International standards (3) Legally binding horizontal regulation (4) Regulations focused on specific targets D. Future Issues (1) Ensuring incentives to use the non-binding intermediate guideline (2) Guidance on the use of AI by the government (3) Harmonization with other countries governance (4) Coordination between policies and standards (5) Monitoring and enforcement 4. Concluding Remarks 1. Introduction There has been discussion around the world on AI governance, or how systems comprising AI as an element, AI services that make the systems available, other related services, and developers, users and service providers should be governed. In Japan, the AI Strategy 2019, revised based on follow-ups this year, and the Integrated Innovation Strategy 2020 are requesting relevant ministries to discuss ideal approaches to AI governance in Japan, including regulation, standardization, guidelines, and audits, conducive to the competitiveness of Japanese industry and increased social acceptance, for the purpose of operationalizing the AI Principles, taking domestic and international AI trends into account. Similar discussion can also be seen in Europe and the United States, where basic policies about regulations on AI systems and more specific regulations have been discussed and published. Global Partnership on AI, launched in June 2020, is an international effort for operationalizing OECD AI Principles. While the discussion on AI governance is developing in Japan and around the world, it is not easy to design actual AI governance. Some may think that horizontal regulation can address issues unique to AI such as lack of explainability. On the other hand, solution to the issues can be sector-specific or use-case-specific because AI can apply to various fields and uses and can raise different issues in each application. Discussion on the design of appropriate monitoring and enforcement mechanisms is also required in order to make the governance effective. We need to structure this complex and multi-layered governance while avoiding hindering innovation as well as addressing concerns regarding AI systems and services. AI governance is an urgent issue that we cannot solve without the knowledge and experience of experts from various fields. The report defines AI governance as design and operation of technological, organizational, and social systems by stakeholders for the purpose of managing risks posed by the use of AI at levels acceptable to stakeholders and maximizing their positive impact. The report discusses trends in AI governance in Japan and around the world, summarizes the development of AI governance discussion, elaborates risk discussion, and defines a universal structure comprising goals, intermediate rules, and rules focusing on specific areas such as sectors. It maps AI principles, horizontal regulations, guidelines, standards, and regulations on specific applications, specific sectors and use by the government in the universal structure with a monitoring and enforcement mechanism. It discusses ideal approaches to AI governance in Japan by taking domestic and international AI trends into account. It proposes the AI governance architecture that is ideal in Japan at the moment and identifies issues that have not been fully discussed. This manual import source is derived from official METI PDF text exposed through the official web viewer for: https://www.meti.go.jp/shingikai/mono_info_service/ai_shakai_jisso/pdf/20210709_8.pdf