Seoul summit charts framework on responsible AI military use

A comprehensive framework establishing guidelines and principles for the ethical and regulated use of artificial intelligence in military operations spanning nuclear weapons systems, endorsed by 61 countries, was adopted at a high-stakes summit in Seoul.

The two-day 2024 summit on Responsible AI in the Military Domain or REAIM concluded Tuesday with the adoption of a "Blueprint for Action," drawing 1,952 participants from 96 countries, including 38 ministerial-level officials, according to South Korea's Foreign Ministry.

The blueprint warned that "AI applications in the military domain could be linked to a range of challenges and risks from humanitarian, legal, security, technological, societal or ethical perspectives that need to be identified, assessed and addressed."

The blueprint notably stressed the "need to prevent AI technologies from being used to contribute to the proliferation of weapons of mass destruction (WMDs) by state and non-state actors including terrorist groups."

The document also emphasized that "AI technologies support and do not hinder disarmament, arms control and non-proliferation efforts; and it is especially crucial to maintain human control and involvement for all actions critical to informing and executing sovereign decisions concerning nuclear weapons employment without prejudice, to the ultimate goal of a world free of nuclear weapons."

The blueprint outlined 20 clauses divided into three key sections: the impact of AI on international peace and security, the implementation of responsible AI in the military domain, and the envisioned future governance of AI in military applications.

The blueprint highlighted the importance of applying AI in the military domain "in a responsible manner throughout their entire life cycle and in compliance with applicable international law, in particular, international humanitarian law."

The adherence is crucial to "harness the benefits and opportunities of AI while adequately addressing the risks and challenges involved," according to the blueprint.

The 61 countries endorsing the blueprint included South Korea, the United States, Japan, France, the United Kingdom, Switzerland, Sweden, Pakistan and Ukraine, according to the Foreign Ministry in Seoul. China notably sent a government delegation to the meeting and attended a ministerial-level dialogue on Tuesday, but the country ultimately chose not to support the blueprint.

During the dialogue, Dutch Defense Minister Ruben Brekelmans said, "The application of AI in the military domain confronts states with fundamental ethical, legal, and societal questions, for example, in relation to a lack of human judgment and control, other types of malfunctions, and the potential lowering of the threshold for violence."

"For armed forces, not deploying an AI system can also come at high costs," Brekelmans said. "It is therefore important to balance the development and deployment of AI applications with thorough risk assessments, conducted by teams of well-trained AI experts and military personnel."

Kenya's Defense Minister Soipan Tuya also disclosed that Kenya will be part of a core group of countries, alongside the Netherlands and South Korea, "spearheading the first UN resolution on responsible AI in the military domain."

"This landmark resolution represents a global commitment to addressing the ethical and security implications of AI in warfare, and Kenya is proud to contribute to this important initiative," Tuya said.

The Foreign Ministry said last week Seoul intended to advance follow-up discussions on the REAIM Blueprint for Action in the upcoming regular session of the UN General Assembly, set to convene later this September.

The REAIM meeting serves as a 1.5-track international multilateral platform that brings together official government representatives as well as private stakeholders from industry, academia and international organizations to address the responsible use of AI in the military domain.

The summit was convened to foster a balanced understanding of the advantages and risks posed by AI in military applications for global peace and security, while also playing a pivotal role in shaping international norms for its responsible and ethical use.

The inaugural summit, co-hosted by South Korea and the Netherlands in The Hague last year, set the stage for this year's second iteration, co-hosted by South Korea, the Netherlands, Singapore, Kenya and the United Kingdom.

President Yoon Suk Yeol has reaffirmed South Korea's determination to lead global AI governance, advocating for a new digital order through pivotal initiatives such as the New York Initiative in Sept. 2022, the Paris Initiative in June 2023 and his address at the UN General Assembly in Sept. 2023.

The 2024 REAIM Summit further solidifies Yoon's digital vision, reinforcing South Korea’s leadership in military AI governance and elevating its status as both an AI powerhouse and a "global pivotal state" -- a key foreign policy initiative of the Yoon administration, according to the Foreign Ministry.