2025 Group of Governmental Experts on Lethal Autonomous Weapon Systems
Second Session (Geneva, 1–5 September 2025)
Asia-Pacific Institute for Law and Security
Utrecht University
Thank you, Chair!
I am delivering this intervention on behalf of the delegation of Utrecht University, and my own delegation, that of the Asia-Pacific Institute for Law and Security (APILS).
Many thanks to you, Chair, and your team, for your tireless efforts to find a workable solution to the problem of characterisation. Thank you also to all the delegations that have shown flexibility over the past few days and put forward constructive proposals for new language.
Part of the difficulty that the group has encountered with respect to the word “identify” arises from the different uses of the word.
For some, the identification of targets seems to mean the process of developing target profiles. By target profiles we mean sets of criteria that intended targets must meet. For example, to neutralise radar systems, a target profile may include all objects emitting electromagnetic radiation of a frequency that is known to be used by the radar system.
For others, the identification of targets seems to mean the process of recognising persons or objects in the operational environment that meet a target profile. Put differently, identification in this sense involves the matching of actual persons or objects encountered in the environment against a pre-set target profile. In our example, this would involve using sensors to detect whether a particular object in fact emits electromagnetic radiation at a specific frequency.
Using the word “identify” in the first sense—developing its own target profiles—would radically narrow the application of the elements that the Group has been drafting. It would then not cover the much wider range of systems that would appropriately fall under regulations and compliance measures on the second tier of the two-tier approach.
In our view, even where a system matches, without human intervention, persons or objects against target profiles, in order to then select and engage them, creates challenges for compliance with IHL. It is precisely these kinds of challenges that the compliance-enhancing measures in boxes III through V would helpfully address.
We believe that the proposed new paragraph 1bis comprehensively addresses this problem, both by explaining what the inclusion of the identification function means and what it does not mean. We are therefore very supportive of paragraph 1bis. We also express our appreciation to ICRC for proposing this new language.
I thank you, Chair!
