The upcoming second Summit on Responsible AI in the Military Domain (REAIM) will be a platform for multistakeholder dialogue on the implications of integrating artificial intelligence (AI) into military systems. Hosted in Seoul by the Republic of Korea, Singapore, the Netherlands, Kenya, and the United Kingdom on September 9-10, the Summit intends to generate ideas and political momentum towards ensuring the responsible development, deployment, and use of AI in the military domain and the establishment of relevant norms. Over a thousand participants from government, international and non-governmental organizations, members of industry, and academia are expected to attend, promising timely and relevant discussions.
The REAIM Summit is one among multiple diplomatic initiatives and normative processes, but it stands out as the one multistakeholder forum that unites a large spectrum of States and other significant actors. Uniquely, it provides a platform for a broad and inclusive discussion on AI in the defense and military sectors. As such, the REAIM Summit may serve as a possible uniting forum in the midst of conflicting State positions on how to move forward with global regulation of military AI. Thus, while the second REAIM Summit may have its own normative impact, it will also serve as a catalyst for other diplomatic and normative processes.
The Previous Summit
The second REAIM Summit follows its inaugural edition, which was organized in the Hague in February 2023 by the Netherlands and the Republic of Korea. The first REAIM Summit was attended by 57 States and around 2000 representatives from across government, industry, academia, and civil society. Over its two days, there was a series of interactive panels open to all attendees and a closed Ministerial level meeting for government discussions at the political level. The Summit was thus the first global multistakeholder conference to address the topic of military AI in a broader thematic context than what until then was rather limited to lethal autonomous weapon systems (LAWS). Specifically, this broader focus enabled discussion of AI-enabled decision support applications, AI applications related to nuclear weapons, and States’ and NATO’s organizational adaptations for integrating AI, for instance.
The first REAIM Summit produced a normative output in the form of a Call to Action that recognized the importance of “responsible AI” (RAI). It called for continued efforts to engage in global dialogue on RAI in the military domain and encouraged knowledge-building and sharing of good practices yet remained relatively vague in terms of substance. The Call to Action was supported by most States attending, including China, France, the United Kingdom, and the United States. However, it was not endorsed by Brazil, India, and South Africa, which were cautious regarding building norms outside the UN, although they were supportive of the Summit. (Notably, Russia had not been invited to the Summit at all, due to its full-scale invasion of Ukraine in February 2022).
The first Summit and its Call to Action led to an international process. The Netherlands and the Republic of Korea organized regional workshops, aimed at getting States to focus on legal and governance challenges in military AI, as well as considering how the new technology would affect each region. In-person workshops took place in Singapore, Santiago, Istanbul, and Nairobi, and there was an online workshop for Europe and North America. These workshops followed a two-track approach of sessions with experts from government, academia, and civil society, followed by closed meetings for governments to discuss regional approaches. The process went global by going regional.
In addition, the Netherlands established the Global Commission on Responsible Artificial Intelligence in the Military Domain (GC REAIM) to help promote mutual awareness and understanding on the issue. The Global Commission will have a two-year mandate, during which it will produce a strategic guidance report intended to help governments better understand RAI. The goal of the Global Commission is to foster expert knowledge and input for the broader multistakeholder process.
The Summit’s Prospects
The second REAIM Summit has three main themes: (1) understanding the implications of AI on international peace and security; (2) implementing responsible application of AI in the military domain; and (3) envisioning the future governance of AI in the military domain. Like the first Summit, there will be mix of public and closed meetings. The first day will feature plenary sessions and side events focusing on the key themes. On the second day there will again be a Ministerial Roundtable. Throughout the Summit there will be exhibitions and live demonstrations by participants from industry, talk shows and interviews, and side events for youth.
Based on discussions with diplomats and organizers of the Summit, there should be around 40 Ministers and Vice Ministers attending the Summit. Interestingly, more Defense Ministers are likely to attend than Foreign Ministers, indicating that Ministries of Defense are becoming alert to the need to engage in these discussions. There should be a good geographical spread of countries represented at Ministerial level, with several African and Latin American Minister-level attendees. The Summit will again be attended by four of the five permanent members of the UN Security Council. Like last year, Russia has not been invited, making the Summit not entirely inclusive and universal.
Following last year’s Call for Action, the second Summit will again have an outcome document, this time called a Blueprint for Action according to a document circulated among diplomats, which is in turn intended to pave the way for a Plan for Action in 2025. The Blueprint has been negotiated and agreed ahead of the Summit by a group of 35 countries, including China. It contains affirmations and commitments across the three themes of the Summit that are useful for getting first commitments by States to start taking action domestically and internationally. It does not specially call for negotiations on regulation, nor recommend a particular way forward on RAI governance. Rather, it stresses the need for continued dialogue and better understanding of the issues. The organizers hope that some of those that did not endorse the Call for Action, such as Brazil, India, and South Africa, will come on board this time.
After the Summit, the REAIM co-hosts are planning to bring a new resolution on the military use of AI to the UN General Assembly (UNGA) this October, putting the issue on the UNGA agenda for the first time. They have also established a core group of countries as co-sponsors of the resolution. It will likely call for UN Member States to submit their views on how to ensure the responsible use of AI in the military domain and request the UN Secretary-General (UNSG) to compile them in a report. A new UNGA process would not be intended to replace future REAIM Summits, at least to start, but would begin UN consideration of the topic with a consultative approach.
The Summit’s Nexus to the GGE, UNGA, and Other Processes on Military AI
The second REAIM Summit is part of and interacts with other diplomatic initiatives and normative processes related to military AI, which may be indirectly influenced by the Summit.
First and perhaps most closely related, at the first REAIM Summit, the United States presented a draft Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy, which was finalized in November 2023. The Declaration aims to build international consensus around responsible behavior and thereby guide States’ development, deployment, and use of military AI. The United States organized an inaugural plenary meeting of States endorsing the Declaration in March 2024. To date, 54 States have signed.
The relationship between the U.S. Political Declaration and the REAIM process is complementary. The U.S. Declaration is more concrete than the REAIM’s Call to Action and has a more operational focus. In addition, the U.S. led process allows information-sharing among governmental representatives in a closed forum where they can be more open than at the United Nations or in public settings. In this context, the REAIM Summit can serve as a platform for increasing awareness of the U.S. Declaration and fostering international agreement on RAI at the political level.
Second, since the inaugural REAIM Summit, the only formal, intergovernmental process related to the military use of AI has been the Group of Governmental Experts on Lethal Autonomous Weapon Systems (GGE on LAWS). Though convening yearly since 2016, it has made incremental progress. The Group is now mandated “to further consider and formulate, by consensus, a set of elements of an instrument.” There has been some movement in positions—for example, France and Germany have come out in favor of the so-called two-tier approach, which would seek a prohibition on autonomous weapon systems that cannot be used in compliance with international humanitarian law and regulation for those that can be. However, other major military States such as Japan, India, the Republic of Korea, Russia, the United Kingdom, and the United States remain opposed to any new legally binding instrument.
While the second REAIM Summit will not resolve these controversies, the Summit can allow for further discussions on and related to them. Expert panels may also present new ideas, notably regarding potential regulations of LAWS under the two-tier approach. In addition, the REAIM Summit places the discussions on LAWS within the broader scope of the military use of AI, including on how to regulate civilian technologies that may have crucial military applications. As such, the REAIM Summit provides a platform for further debate on the future ethical and legal permissibility of LAWS.
A third related process is also underway at the United Nations. Austria brought a new Resolution on LAWS to the UNGA in October 2023, following its Conference entitled Humanity at the Crossroads: Autonomous Weapons Systems and the Challenge of Regulation held in April 2023 in Vienna, which produced an outcome document endorsed by 36 States. It called for clear prohibitions and regulations to preserve the human element in the use of force, arguing that the international community was at an “Oppenheimer moment,” in which “geopolitical tensions threaten to lead a major scientific breakthrough down a very dangerous path for the future of humanity.”
The Austrian resolution created an agenda item on LAWS for the first time and invited all stakeholders interested in the issue to submit their views to the UN Secretary-General. Based on these submissions, the UNSG issued a report on the way forward in July 2024. He reiterated his joint call made with the President of the International Committee of the Red Cross in October 2023 for States to negotiate a legally binding instrument by 2026. The ground is therefore prepared for the UNGA to take over, should the GGE on LAWS fail to agree on starting negotiations.
The REAIM Summit shares similar features to the UNGA process, which is an inclusive and near universal yet purely State-driven forum. Unlike the REAIM Summit, however, the UNGA has the authority to formally adopt legally binding instruments and political commitments. Thus, REAIM may set the grounds for further deliberations and potential negotiations at the UNGA, not just on LAWS, but on other aspects of RAI. Insights, ideas, and joint understandings from the REAIM Summit are likely to feed into the UNGA’s work. Should a process on RAI be introduced at the UNGA in addition to the Austrian-led process on LAWS, there may be overlap or synergies between the two processes.
Outlook
The second REAIM Summit will increase global awareness and understanding of RAI in the military domain, ensuring it remains a priority for policymakers. But there are certain limitations to this process, notably that it is still perceived as a mainly western initiative that does not include all the major military powers, such as Russia. Nevertheless, its transparent and multi-stakeholder approach should enable it to shape norms and the implementation of norms for the development and use of military-related AI technologies.
In the next few years, the UNGA may play a greater role in this space, with potential new processes on LAWS and RAI. As AI technology develops, there are also likely to be increasingly more issues under the umbrella of “AI in the military domain,” meaning that more diplomatic and normative initiatives may be brought to the UNGA. To manage related processes, the UNGA First Committee could introduce a thematic debate on the topic or create a new Cluster on the military use of AI.
As parallel yet potentially competing initiatives continue in this area, no comprehensive process will be able to address all the complex issues of the development, deployment, and use of AI. The REAIM process will therefore continue to be an important forum for all stakeholders – notably including industry and academia – to come together to exchange information and best practices on the challenges of ensuring responsible AI in the military domain as well as to lay the ground for common vision and understanding.