Skip to content
Commentary | 2 February 2024

What does global military AI governance need?

Image of Mahmoud Javadi

Mahmoud Javadi |AI Governance Researcher at Erasmus University Rotterdam (EUR)

Image of Michal Onderco

Michal Onderco |Professor of International Relations at Erasmus University Rotterdam

Arms Control China Conventional Arms Control Cyber Emerging technologies Hybrid warfare Risk Reduction United States Global Security

We currently face an absence of globally acknowledged governance frameworks for overseeing the comprehensive life cycle of artificial intelligence (AI) in military domains, encompassing its development, deployment, and utilisation.

In 2023, two initiatives intriguingly came into existence simultaneously. The first Summit on Responsible Artificial Intelligence in the Military Domain (REAIM) took place in February 2023, was hosted by the Netherlands, and attracted 2,000 attendees. At its end, 57 states endorsed its final declaration, Call to Action. On the last day of the Summit, the United States (US), probably catching the REAIM conveners off guard, unveiled the “Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy” (the Political Declaration), its own pioneering initiative to “build international consensus around responsible behaviour and guide states’ development, deployment, and use of military AI.” This declaration extended an invitation to nations globally to participate. As of January 2024, fifty-one countries have joined this initiative, including all member states of the European Union (EU).

Both REAIM and the Political Declaration fall short in their ambition and capability to establish global rules for military AI. They primarily aim to foster coalition-building based on a shared vision of military AI. Their distinctiveness lies in determining the composition of the coalition and the specific vision underpinning them.

The US appears to be strategically framing the Political Declaration as a means to tackle (perceived) concerns and threats from near-peer competitor China. The declared aim of the Political Declaration is to “define in far more granular detail what … irresponsible state behavior looks like in the development and use of this technology.” Given this context, we assert that REAIM and its forthcoming Global Commission on AI – where the Hague Centre for Strategic Studies (HCSS) serves as the Secretariat – are better configured to chart the global military AI governance course.

Military AI and global leadership in the Biden era

Influential thinkers and strategists in the US often draw comparisons between nuclear weapons and AI, describing the latter as “unprecedented and, in some ways, even more terrifying technology.” The analogy has reignited the push for the US to exhibit global leadership in AI arms control, mirroring its past “significant achievements” in averting nuclear war, impeding nuclear proliferation, and shaping an international order that fostered decades of (cold)peace among great powers.

It should come as no surprise that the Political Declaration was set in motion precisely during the REAIM summit. This move seems to convey the notion that any steps towards the governance of military AI necessitate American leadership. Contrary to expectations, both the organizers and key participants of REAIM were caught off guard by the launch of the Political Declaration. As a result, REAIM’s Call to Action apparently avoided to mention the Political Declaration, as a subtle way to express distance from the American decision. Importantly, this omission apparently aimed to retain endorsement for REAIM from certain states, notably China.

The Biden Administration explicitly acknowledges that China possesses both the “intention and, increasingly, the capacity to reshape the international order.” This assertion underpins the primary motivation behind the Political Declaration, which seeks influence on China’s behavior in military AI capabilities. This objective gains heightened significance following the failure of American and Chinese leaders to commit to a ban on AI in autonomous weapons. The 2022 US National Security Strategy also reiterates Washington’s commitment to responsibly managing the competition between the US and China. As part of this responsible approach to competition management, US initiatives in the past two years have specifically aimed to restrict China’s access to high-end semiconductors used in AI, given concerns that such technology is enhancing Beijing’s military capabilities. While these measures faced criticism from the Chinese Premier in his remarks at the 2024 World Economic Forum in Davos, it is nonetheless reasonable to infer that there is a perception of China as an irresponsible state, prompting the need for regulation of its behaviour in the realm of military AI.

The US has affirmed its commitment to outcompete its rivals, particularly China. Despite being hailed as the most exhaustive state-level multilateral confidence-building measures for military AI, the Political Declaration aims primarily to furnish Washington with a strategic tool for building a coalition of states that share its vision for the responsible use of military AI outside the confines of the UN setup.

To this end, it incorporates a precise definition of military AI and outlines ten measures that firmly encapsulate the American vision of this technology. This viewpoint has remained unchanged since February 2023, even in the face of increasing support from various states since that time. These measures are exclusively directed towards states, rendering the initiative less inclusive but more focused on coalition-building to promote a narrative favourable to the United States and, apparently, its China strategy. To these ends, the Biden administration diligently cultivated support for the Political Declaration in a series of informal background briefings with senior officials throughout the autumn of 2023. The ultimate endorsement of the Political Declaration by fifty-one nations is a fruit of that labour.

REAIM: A democratic, decentralised, and depoliticised approach to global military AI governance

By contrast, REAIM represents a different set of values. The most important difference is its inclusivity and multistakeholder engagement. The first Summit drew participants from a hundred countries, including eighty government representatives, along with delegates from knowledge institutions, think tanks, industry, and civil society organisations. The Summit featured four high-level sessions, approximately thirty-five breakout sessions, twenty AI demonstrations, an Academic Forum, Innovation hubs, and a student hub.

REAIM’s second Summit is scheduled to take place in South Korea in September 2024, with expectations that the Global Commission on AI will be eventually established before the convening of the Summit. The Commission aims to “raise all-round awareness, clarify the definition of AI in the military domain, and determine how this technology can be developed, manufactured, and deployed responsibly.”

On the summit’s final day, REAIM’s Call to Action received endorsements from 57 participating states, including China and all EU member states except Austria and Ireland. Unlike the Political Declaration, endorsements for REAIM are not continuous. Nevertheless, the second edition of the Summit, coupled with the forthcoming Commission, may open avenues for additional support and endorsements from state and non-state stakeholders.

REAIM’s final declaration calls for nine specific actions. The bedrock of these actions revolves around multi-stakeholder depoliticised dialogue for knowledge-building and exchange of best practices for developing, deploying, and using AI in the military domain. These actions give prominent roles to academia, knowledge institutes, think tanks, civil societies, private sectors, and industries for contribution to the establishment of a “normative framework on military AI” aligning with the Political Declaration; however, its sole reliance on states and an overarching focus on Sino-American rivalries set it apart.

REAIM avoids a top-down approach. It aspires to be a democratic forum to gather ideas and facilitate dialogue among diverse state and non-state stakeholders. This approach, along with the broad range of participants, indicates that REAIM intends to focus on technical dialogue rather than becoming entangled in politicised discourses on military AI in the context of geopolitical rivalries. It is not without its imperfections. Some states, such as Iran, have voiced discontentment over their exclusion from the first Summit. The forthcoming Commission and the Summit in South Korea provide an opportunity to address such criticism.

Since the trajectory of AI development requires states alongside private entrepreneurs, technologists, thinkers, and companies, the need for an inclusive, multi-stakeholder, democratic, depoliticised, and decentralised forum becomes inevitable for any prospective legally binding convention in the future. Global military AI governance requires an extensive coalition founded on a consensual vision. REAIM has taken the first decisive step toward this objective.

Global military AI governance requires an extensive coalition founded on a consensual vision. REAIM has taken the first decisive step toward this objective. Mahmoud Javadi and Michal Onderco

Charting a path for global military AI governance

To sustain the current momentum spurred by REAIM, it is imperative for the EU and its member states to actively engage and allocate resources to ensure its success. It may not be characterised as overly ambitious since it is focused on advancing knowledge and understanding military AI, according to the EU statement in the 2023 Conference on Disarmament. Nevertheless, it serves as a crucial prerequisite for any efforts, including the Political Declaration.

The evolving landscape of global military AI governance is both expansive and intricate, particularly when viewed within the context of strategic rivalries. The establishment of a robust framework for military AI governance is a responsibility that should fall under the purview of the UN, renowned for its legitimacy as a truly global forum based on international law. Until the UN assumes a central role in this regard, addressing the challenges associated with regulating military AI becomes more attainable by aligning with the democratic, depoliticised, and decentralised principles articulated by REAIM.

The ultimate objective of the Political Declaration is not something Europeans shy away from, as evidenced by the widespread endorsement from nearly all European states and all the EU member states in particular. However, the crucial consideration lies in the strategies and approaches to attain this goal. This becomes particularly significant given the potential impact of the U.S. presidential elections this year, which could once again influence all multilateral state-level initiatives.

If the US attempts to lead from the centre in the governance of military AI, it becomes imperative to also lead from behind. Therefore, the wisest policy choice for Washington is to acknowledge that achieving the visions and missions of REAIM would contribute significantly to fulfilling the Political Declaration.

The opinions articulated above represent the views of the author(s) and do not necessarily reflect the position of the European Leadership Network or any of its members. The ELN’s aim is to encourage debates that will help develop Europe’s capacity to address the pressing foreign, defence, and security policy challenges of our time.

This work has been funded by the REMIT project, funded from the European Union’s Horizon Europe research and innovation programme under grant agreement No 101094228.

Image credit: Wikimedia Commons / Pixabay