Japan Govt Unveils Guidelines for Managing AI-Incorporated Defense Systems; Aims to Cancel Research of Systems Deemed High Risk

The Yomiuri Shimbun

12:59 JST, June 7, 2025

The Defense Ministry has unveiled guidelines for managing the risks associated with defense equipment incorporating artificial intelligence, with the aim of ensuring the use of AI remains within the scope of human control.

The guidelines clearly state that the government will not permit the research and development of defense equipment if it is found to be Lethal Autonomous Weapons Systems (LAWS) , in which a human is not involved in selecting targets or deciding which targets to attack.

According to the guidelines, risk management for research and development must be conducted in three stages: classification of AI equipment, a legal review and technical review.

Equipment will be examined under these guidelines based on how the judgement of the AI system impacts destructive capabilities, dividing research and development targets into high-risk and low-risk categories.

If deemed high-risk, the government will assess compliance with international and domestic laws prior to the commencement of research and development. This includes missile launches that are assisted by AI to identify targets.

If deemed LAWS, the system’s development and research will be canceled.

After the legal review is complete, the process moves on to a technical review. This stage verifies that the design allows for human control and ensures safety through mechanisms that reduce AI malfunctions.

To ensure an effective review, the ministry will need the cooperation of defense contractors that design equipment incorporated with AI, requiring them to disclose AI algorithms and other relevant information.

The ministry plans to finalize the specific methods for ensuring cooperation through future discussions with the companies.