The Way Forward on Autonomous Weapons after the Vienna Conference

International efforts to regulate autonomous weapons systems may soon change drastically. The slow pace of progress in current discussions has led to dissatisfaction among participants from civil society and states hoping to curb the use of autonomous weapons, especially small and middle-sized Global South and European countries like Brazil and Ireland. New initiatives have now been started to breathe life into ongoing efforts. The Vienna Conference on Autonomous Weapons Systems, hosted by Austria from 29–30 April, is one such initiative. During the Conference, Austria gauged the interest of other states in breaking drastically with existing discussions and moving ahead with drafting new legally binding measures. Another disappointing year in the current track of negotiations may push a large group of states towards Austria’s position.

Autonomous weapons systems (AWS) lack an internationally agreed-upon definition, but for the purposes of this article are understood to be systems that use AI and the processing of sensor data to select and apply force to targets without further human intervention. The main venue for discussions on AWS is the Group of Governmental Experts on Emerging Technologies in the Area of Lethal Autonomous Weapons Systems (GGE on LAWS), a subsidiary body of the Convention on Certain Conventional Weapons (CCW). Discussions on AWS have been held yearly since 2013 within the CCW framework, but progress has been slow even on basic issues: the latest round of meetings features characterising and defining what lethal autonomous weapons actually are as a core topic on its agenda. In the meantime, Israel is using AI tools to locate targets in the Gaza Strip and Ukraine is using autonomous drones to carry out strikes on Russian targets. Dissatisfaction with the slow pace of discussions in the face of increasing use of AWS has led to a proliferation of new initiatives in an effort to ensure progress.

The Vienna Conference and Its Broader Context

The Vienna Conference, titled ‘Humanity at the Crossroads: Autonomous Weapons Systems and the Challenge of Regulation’, is the latest in a number of global and regional efforts to make headway. With this Conference, Austria championed the cause of a group of states seeking strict, legally binding measures to prohibit some AWS and regulate others. Here, these states will be referred to as ‘progressive’, as opposed to ‘conservative’ states advocating for less far-reaching measures or a slower pace in moving towards new measures. This excludes those pushing back against almost all new measures, notably Russia. 

With over a thousand participants including representatives of 144 countries, the Vienna Conference was attended both by prominent progressives like Brazil, Palestine and Mexico and conservatives like India, China and the US. A major role was accorded to civil society organisations and academia, who tend to fall into the progressive camp and had representatives speak during every panel discussion. Notably, all major powers and states with a realisable interest in the development of AWS attended.

The Conference showed its progressive leanings in a variety of ways. Perhaps the most important point of contention between the progressive and conservative factions is the question of whether to implement prohibitions and regulations of AWS through legally binding or voluntary measures. The Vienna Conference’s hosts as well as the vast majority of panellists made it clear that a legally binding instrument, such as a CCW Protocol or a wholly new treaty, is the best way to address the issue. In addition, Austria made it clear that it was prepared to start negotiating such an instrument even without those states that have shown reservation so far. The Chair’s Summary of the Vienna Conference ends with the line: ‘This Chair’s Summary affirms our strong commitment to work with urgency and with all interested stakeholders for an international legal instrument to regulate autonomous weapons systems’.

Furthermore, while both progressives and conservatives agree that some measure of human control is needed to make sure that AWS comply with international law, progressives tend to argue for more far-reaching requirements under the moniker ‘Meaningful Human Control’. At the Vienna Conference, one panel discussion focused solely on human control and legal accountability. Here, panellists clearly preferred the phrase ‘Meaningful Human Control’, and at several points used terms taken directly from a Palestinian working paper that spearheaded the progressive effort at the 2023 GGE on LAWS session.

In a final major area of discussion, progressive states tend to push for regulations on AWS to be based not only on law, but also on ethics. Austria has been a major proponent of the inclusion of ethical considerations in AWS regulation, and on the development of discussions in GGE on LAWS in recent years stated: ‘We particularly regret the mere lip service to the ethical considerations which have been part of our work since its very beginning’. Ethical concerns were thus the topic of a full panel discussion at the Vienna Conference.

Where to Now?

The GGE on LAWS will meet again in August, but the slow pace of progress over the past 11 years makes it likely that this round of meetings will not satiate the hopes of progressive states. The increasing attention given to autonomous weapons may mean that another disappointing outcome – either lacklustre discussions at this year’s sessions or another meagre report after the tentative conclusion of its current mandate in 2025 – will push many progressive states to accept Austria’s assessment that the GGE on LAWS cannot fulfil their goals and that an alternative path must be taken with only ‘interested stakeholders’. However, moving away from current discussions too quickly risks causing a rift between states that would be too difficult to bridge in the future: states categorically opposed to autonomous weapons may negotiate legally binding rules that are too strict for others to join. This would leave the world with a treaty that regulates these systems only in countries with no interest in developing them, and which will never be adopted by states where it might have a practical effect.

To prevent such a situation, states that seek to prohibit and regulate autonomous weapons to the greatest extent should focus on two key points. First, they should keep contributing actively to ongoing discussions with the aim of warming other states up to potential legally binding measures. Slow progress has not meant a lack of progress, and the GGE on LAWS has notably agreed on Guiding Principles on AWS and a two-tiered approach of prohibitions and regulations. Abandoning these discussions would leave them without some of the voices most committed to pushing them forward, thereby only further hampering progress and bringing different tracks of discussions into conflict with each other rather than making them mutually reinforcing.

Second, if progressive states do decide to start negotiations on legally binding measures, they must moderate some of their positions and take into account the needs and interests of states that wish to develop autonomous weapons. It has become clear that some of the demands made by progressive states will never be acceptable to conservative states: for example, the US roundly rejects overly strict language on control over AWS after their activation because it would forbid existing systems that have been in use for years by many states without legal controversy, such as missile defence systems. Including such measures in a legally binding instrument would therefore prevent a significant group of states from ever joining it. 

A potentially more productive approach would be the development of parallel measures: a lenient legally binding instrument combined with stricter voluntary measures. This would increase the likelihood of conservative states joining the instrument while maximising practical regulation through voluntary means. A more lenient instrument would leave out some more contentious goals, like the inclusion of ethics in regulation of AWS, to focus on the main goal of the progressive bloc: the establishment of legal barriers to the development and use of AWS that cannot be adequately controlled by humans. The extent of control necessary would remain a difficult issue. However, conservatives in the GGE on LAWS already acknowledge the need for various measures to ensure human control, as demonstrated in a 2023 working paper by a group of conservative states. Moving closer to the conservative position on control seems the only feasible way for the progressive bloc to get the conservatives to agree to a legally binding instrument. Stricter voluntary measures will prove easier to find agreement on, as both conservative and progressive states have already expressed support for these: one example is the 2023 Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy, which was supported by 54 states.

Austria has made it clear that it hopes the Vienna Conference will drive a group of progressive states towards a strict legally binding instrument. A similar hope inspired the adoption of the Treaty on the Prohibition of Nuclear Weapons (TPNW). This treaty seemed the perfect fulfilment of progressive aspirations, including strict prohibitions and a reference to ethics in its preamble. Those advocating for the same approach to autonomous weapons should consider whether the TPNW was successful in its attempt to do away with nuclear weapons: less than half of UN member states have signed the treaty, and we have rarely been closer to the use of nuclear weapons. Moving towards a legally binding instrument on autonomous weapons while ignoring the need to keep engaging with those expressing hesitation may irreparably damage ongoing discussions on autonomous weapons while producing an instrument that will only be in force where autonomous weapons do not actually exist. The suggested approach of seeking a lenient legally binding instrument combined with stricter voluntary measures will prove highly challenging, and there is a real chance that even its optimal fulfilment will not lead to the achievement of the progressive states’ end goals. Using diplomacy to prevent the harm that autonomous weapons may inflict, however, means engaging productively with states that are willing to inflict this harm.

The Vienna Conference put ‘humanity at a crossroads’ in its title, and put discussions on AWS at a crossroads with its content. Now, the challenge is to find the path that leads to effective and widely accepted prohibitions and regulations on AWS.

The views expressed in this Commentary are the author’s, and do not represent those of RUSI or any other institution.

Have an idea for a Commentary you’d like to write for us? Send a short pitch to commentaries@rusi.org and we’ll get back to you if it fits into our research interests. Full guidelines for contributors can be found here.