Growing Consensus on Policy at UN Discussions on AWS but Skepticism Towards Non-Binding Principles & Practices

Although agreement on the substance of a framework continues to develop, states expressed skepticism over proposals for non-binding practices and principles and continue to call for legally binding rules on the development and use of Autonomous Weapon Systems (AWS). 

These discussions were the 2nd in a series of 3 intersessional meetings between the GGE meeting from 7-11th March 2022 and the forthcoming GGE taking place from 25-29th July 2022, and the focus was on human control, human-machine interaction, ethical considerations; and responsibility and accountability in the development and use of AWS. 

Following the 6th Review Conference in December 2021, in which states failed to agree to begin negotiations, several written proposals have been submitted for the normative and operational framework. Among the proposals, there is broad agreement on key issues such as characterisation of AWS as systems that rely on autonomous functions to select and engage targets, the need for human control, judgment, or supervision over such systems, the recognition that systems that cannot comply with international humanitarian law should be prohibited and the need for regulations to ensure AWS compliance with legal requirements and ethical principles. However, divergences remain around whether non-binding principles and practices would suffice, or whether legally binding rules are needed. 

During these discussions, the Group of 6 (G6) states that have proposed non-binding principles and practices (Australia, Canada, Korea, Japan, UK, and USA) took the floor to argue in favor of their proposal, suggesting that it is the only proposal able to achieve consensus within the CCW. However, several states, including Argentina, Austria, Chile, Ireland, Mexico, New Zealand, Palestine, Switzerland, and Uruguay expressed doubts around whether the proposal would be sufficient to safeguard against the legal, ethical, and humanitarian risks posed by AWS and to ensure meaningful human control over the use of force. 

On the topic of human control, judgment, and human-machine interaction, there was a broad consensus on the centrality of the human element in ensuring that the development and use of AWS comply with existing legal requirements and ethical principles. However, concerns were raised that the G6 proposal fails to state that AWS that are incapable of being used with sufficient human control should be prohibited – with the implication that insufficient human control would be acceptable. Additional concerns were raised around the G6 proposal’s use of the word ‘design’ under systems that should be prohibited, with states commenting that weapon systems are rarely specifically designed to breach international law, but rather that their use in practice could lead to non-compliance. 

On ethics, states commented that it is insufficient to merely state that ethics are applicable without specifying how ethical principles apply to an agreed framework. Several states, including Ireland, Chile, Germany, Mexico, and Switzerland suggested that ethics should inform the boundaries between acceptable and non-acceptable weapon systems, with systems used without a sufficient degree of human control being prohibited and positive obligations to ensure limitations around the broad scope of systems that incorporate autonomy to select and engage targets. Other states also commented that systems that rely on prejudicial data sets raise ethical issues, and should be prohibited. The Stop Killer Robots campaign and the ICRC argued that AWS that are designed to target humans specifically would be unethical and should be prohibited. 

On responsibility and accountability, all states agreed that it is humans and not machines that are morally and legally responsible for the use and development of AWS. China stated that responsibility and accountability must also apply to people who research, develop, manufacture, deploy, and use AWS. Italy stated responsibility and accountability are necessary at all levels of command, including the strategic level. The Stop Killer Robots campaign added that  AWS would further undermine accountability for perpetrators of unlawful violence, and would make providing retributive justice to victims even more difficult.

Although differences over issues around human control, ethics, responsibility and accountability continue to narrow, a key divergence remains over the way forward. While states have recognised the potential value of the G6 proposal for practices and principles, many states continue to argue that legal limits are needed to establish the boundaries of acceptability over the development and use of AWS. The introduction of non-binding principles and practices would risk creating an ecosystem in which AWS are developed and used without first establishing clear limits, which would pose clear legal, ethical and humanitarian risks and endanger international security. 

The 3rd in the series of intersessional meetings at the UN will take place from 27-28th June 2022, followed by the CCW GGE at the Palais des Nations, Geneva from 25-29th July 2022. The Stop Killer Robots campaign will be present for both and continue to advocate for negotiations to begin on new international law, to reject automated killing and ensure meaningful human control over the use of force.