Japanese Govt Plans to Double Staff Tasked with Checking AI Safety; Move is Part of Basic Plan for Increasing Development, Use of AI

The Yomiuri Shimbun

15:21 JST, December 11, 2025

The government plans to double the number of staff at a government-affiliated body that works to confirm the safety of artificial intelligence technology, according to a draft of its basic plan on AI.

The plan, which is set to be approved by the Cabinet within this month, will be the first of its kind that the government has compiled.

In order to support the development of AI, the government also seeks to loosen regulations on the protection of personal information. According to the draft, the government aims to balance technological innovation with risk management to make Japan the country with the best environment in the world for developing and using AI.

The basic plan sets the direction of the government’s future AI policy based on the law to promote the research, development and utilization of AI-related technologies, which was enacted in May. The draft was submitted to meetings such as the one held Wednesday by the ruling Liberal Democratic Party’s Headquarters for the Promotion of a Digital Society.

The draft expresses concern that Japan is lagging behind other countries in developing and investing in AI, which has a direct impact on the nation’s industrial competitiveness and national security. “The gap is becoming more conspicuous year by year,” the draft said.

The draft stipulates that the government will be “aggressive in closing the gap,” capitalizing on Japan’s strengths such as having high-quality data. It has set four principles as key to this goal: promoting the utilization of AI; improving development capabilities; better governance; and transforming society.

To increase AI adoption at government ministries and agencies, employees who assume managerial positions will be urged to proactively use the technology to set an example. The government will also provide subsidies to small and medium-sized enterprises to adopt AI. In the draft, the government sets out a plan to submit a bill to revise the Law on the Protection of Personal Information to the Diet at an early stage to support the training of AI models, which is essential for the development of the technology. The government intends to allow the acquisition of information without individual consent when the purpose is to generate statistical data.

However, the draft also refers to the risks of AI, such as criminal misuse and the spread of misinformation and disinformation. “It is crucial to dispel the public’s concerns,” the draft said. It states that the number of staff at the AI Safety Institute, a government-affiliated institution established in 2024, should be “immediately” expanded to about double its current level, which is roughly 30. The institution is expected to engage in activities such as developing a system to evaluate the safety of AIs.

During a meeting of an expert panel on the draft, some people asked the government to set numerical targets for metrics such as rate of AI usage by citizens and the amount of public-private investment in the technology. However, the government rejected the idea, concluding that it is too early to do this.