Trustworthy AI Is the Goal in the Artificial Intelligence Basic Planx

◆In December 2025, the “Artificial Intelligence Basic Plan” (hereinafter referred to as the Basic Plan) was approved by the Cabinet. It strongly emphasized the policy of aiming to become “the world's most AI-friendly country for development and utilization,” while placing “trustworthy AI” as its core concept. Examining “trustworthy AI” through the lens of the OECD's AI Principles reveals a strong emphasis on human-centered values, fairness, robustness, and safety. Looking at the approaches of the United States and the EU, the items they prioritize and their methodologies differ from Japan’s. For Japan to continue leading the creation of international cooperation frameworks, there are many challenges, including the harmonization of evaluation criteria.

◆The Basic Plan cites “high-quality data” as a key strength for Japan in realizing “trustworthy AI,” citing healthcare, research, and industry as examples. However, progress in data infrastructure and utilization varies significantly across sectors. Healthcare and research, being highly public-oriented, benefit from government and research community leadership in establishing foundational frameworks, making data and usage environments relatively easier to develop. In contrast, while cross-sector and cross-industry collaboration is advancing in the industrial sector, inter-company collaboration across the sector remains slow. In addition, digital transformation (DX) has lagged among small and medium-sized enterprises (SMEs), making it difficult to say that “high-quality data” currently exists to a significant degree across the entire industry.

◆The Basic Plan also expresses concern about Japan’s lag in AI-related development and investment. The international upper echelons of general-purpose foundational models are led by the United States and China, with Japan's presence currently limited. Furthermore, considering the uneven international distribution of computational resources (GPU clusters) and the gap in private investment scale, it suggests Japan may not yet have sufficiently established the prerequisites for continuously developing and operating competitive AI models.

◆In light of the principle of “trustworthy AI,” the Basic Plan’s direction—developing AI foundation models suited to Japanese culture and customs, ensuring a certain level of autonomy, and building social implementation using high-quality data in areas requiring reliability—is desirable. However, resolving challenges such as data preparation outlined in the Basic Plan will take some time, and concerns remain regarding the concrete measures, execution capability, and speed required for its realization. The government has indicated its policy to compile a public-private investment roadmap by spring 2026, necessitating continued close monitoring of developments.