New york: A global commission focused on the military use of artificial intelligence (AI) has urged the formation of legally binding agreements to ensure that decisions regarding the use of nuclear weapons remain under human control, rather than being delegated to AI systems. This recommendation is part of a broader effort to responsibly integrate AI into military operations.
According to Yonhap News Agency, the Responsible AI in the Military Domain (REAIM) Global Commission, a collaborative initiative spearheaded by the Netherlands and South Korea, put forth this recommendation in its inaugural guidance report on AI governance within the military sphere. The announcement was made by Dutch Prime Minister Dick Schoof during a high-level United Nations Security Council meeting in New York, dedicated to AI and international peace and security.
The report emphasized the necessity of keeping critical decisions related to nuclear weapons under human authority, citing the moral, legal, and strategic considerations involved. It advocated for the pursuit of these commitments through national policy declarations or an international agreement. The report emerges amidst growing concerns about a lack of effective global governance regarding military AI use and underscores the need to establish norms and guidelines, particularly for decisions involving weapons of mass destruction.
The commission, established in 2023 through the joint efforts of South Korea and the Netherlands, aims to provide guidance on AI use in the military by involving experts in the field. The report also recommended aligning AI development and use in the military with ethical principles and international law to prevent unlawful applications and ensure adherence to existing obligations, such as the right to use force and the manner in which it is employed during conflicts.
Additionally, the report called for the implementation of national policies that ensure human responsibility throughout the AI system life cycle, emphasizing the importance of rigorous testing, evaluation, verification, and validation. To further facilitate responsible AI integration, the guidance advocated for the establishment of a permanent and inclusive multilateral dialogue, as well as the development of an expert network to share knowledge and raise awareness of AI governance frameworks.