2023 Volume 7 Issue 1 Pages 69-100
This paper (i) proposes the "rules (regulations)" necessary for the research and development, use, sale and supply of AI systems and services, (ii) respects voluntary efforts to comply with them, (iii) establishes "rules (harmonised standards, technical standards and requirements)" to act as "de facto mandatory standards" in the sale and supply of services, (iv) establishes management system standards to plan, implement, check and act on them, (v) establishes a legal basis for disciplining these mechanisms, and (vi) establishes the 'AI Supervisory Commission (tentative name)', (vii) which will unveil the concept of AI regulation with the 'Japanese AI System Conformity Assessment System'.
On the other hand, instead of continuing the trial-and-error approach of considering soft law, which is expected to be exclusively self-disciplining and may not need to be observed, the aim is to overcome the resistance to hard law associated with the introduction of regulations (such as legal prohibitions on substantive issues), against which there are persistent objections, and to address the issues of principles and guidelines that have been discussed so far. To achieve this goal, this paper attempts to move away from the theory of law and regulation avoidance, which leaves compliance to voluntary initiatives and does not regulate through hard law, to 'optimise' AI regulation in the light of international trends, and to propose new measures that will contribute to future AI regulatory policy in the research, development and use of AI.
