By Branka Marijan
Published in The Ploughshares Monitor Winter 2023
After close to a decade of effort, there is still no treaty regulating autonomous weapons systems. But even if naysayers are convinced that no treaty is possible, it is still the case that crucial steps are being taken. And, if progress seems to come at a snail’s pace, we must remember that arms control and disarmament efforts rarely happen quickly.
Recent progress
On November 1, the First Committee of the United Nations (UN) General Assembly, which deals with disarmament and global threats to peace, approved the first-ever resolution on autonomous weapons. The draft, which is expected to be adopted at the General Assembly plenary session in December, highlights the “urgent need for the international community to address the challenges and concerns raised by autonomous weapons systems.” It was approved by 164 states, while five voted against and eight abstained.
This result reflects the growing recognition of the need to regulate the use of artificial intelligence (AI) and autonomy in weapons systems. What follows this resolution is not clear. Perhaps concrete politically binding measures, ideally a legally binding treaty. One thing is certain: more dialogue and pressure on states to crystallize their positions on this issue will help to move the regulatory process forward.
The steps to a successful outcome
This UN resolution did not happen in a vacuum. Discussions on autonomous weapons at the UN Convention on Certain Conventional Weapons (CCW) have been happening since 2014, kept alive by an active civil society and a few key states. Many experts to the CCW have provided member states with important insights into the technological developments that have, over time, become more significant. However, the CCW as a forum has run its course because of its inability to make consensus work and opposition by key states, such as Russia.
In 2022, 70 states reacted to the CCW’s stagnation by presenting a joint statement at the UN General Assembly, calling for more efforts on autonomous weapons. This group, which included the United States, the United Kingdom, and the Republic of Korea, provided cross-regional representation.
Several states with technically advanced militaries also started to look for additional ways to address certain aspects of wider military use of AI that had been somewhat neglected in UN discussions, such as battlefield decision-making and war-gaming. This past February, the government of the Netherlands, with the government of the Republic of Korea as co-host, organized the Responsible Military AI (REAIM) Summit in the Hague. The Summit attracted many states, with 57, including China, agreeing to the call for action.
If progress seems to come at a snail’s pace, we must remember that arms control and disarmament efforts rarely happen quickly.
At the close of the Summit, the United States introduced a Political Declaration on responsible military AI and autonomy. The declaration urges that “states should ensure that relevant personnel exercise appropriate care in the development, deployment, and use of military AI capabilities, including weapon systems incorporating such capabilities.” What is meant by “appropriate care” will likely be discussed by the United States and those of its allies that have endorsed the declaration so far. It is unlikely that either China or Russia will join this U.S.-dominated effort.
The 2024 REAIM Summit is set to be hosted by the Republic of Korea and we could soon learn more about the relevance of this gathering to efforts on regulating autonomous weapons. Russia was absent from the 2023 Summit, while China, even though it joined the call to action, has not been an enthusiastic supporter of greater regulation. So, it is possible that these efforts will become little more than preaching to allies.
Other efforts, such as the United Kingdom’s Safety Summit, while focused on existential risk, did not address aspects of defence but catered to the views of key industry leaders.
Some states have expressed concern about technological competition and AI arms racing, over which they have little control. Latin American and Caribbean countries have called for the prohibition of weapons that do not employ meaningful human control in the selection and engagement of targets. This regional group, which met in Costa Rica in February, released the Belen Communiqué, which outlined the position of these states: that a legally binding instrument that prohibits systems that function outside of human control is necessary.
All these efforts have been important in the leadup to the First Committee resolution and will remain important following its adoption. The various summits and meetings show that the CCW is not the only forum where states are discussing concerns related to autonomous weapons.
Future steps – and hurdles
The First Committee resolution does not bridge the divide between those who favour concrete regulation and those who favour voluntary measures. Although an overwhelming majority of states voted for the resolution, the group that voted against or abstained is significant and will play a large role in future resolutions and the regulation of autonomous weapons.
Both Russia and India voted against the resolution; both have opposed efforts at CCW, not seeing a need for new law on autonomous weapons. Interestingly, China, Israel, and Türkiye abstained. China’s unwillingness to agree to regulations that are not on its terms means that a comprehensive binding instrument is unlikely. Israel is a key player in military AI, having developed innovative AI weaponry, including sophisticated loitering munitions, drone swarms, and new tanks; it, too, has spoken against new law. Türkiye’s abstention reflects its positioning of itself as an exporter of medium-sized drones and loitering munitions.
As well, rapid advancements of AI capabilities, including large language models like ChatGPT, show that autonomous weapons are only one type of AI-enhanced capabilities used by modern militaries; all deserve attention.
The steps that lead to the regulation of autonomous weapons will include further resolutions. No steps can be taken without political will, particularly from middle powers such as Canada. These middling states will need to show leadership and political savvy in an arena dominated by the great AI powers.
When Austria proposed the resolution at the First Committee some weeks before it was approved, it did so after years of study and many consultations with other national governments and civil society. Other middle powers, particularly Canada, should follow this model.
