Weapons that detect targets and apply force to them based on sensor inputs have been around for years. Such systems are distinct from those where a human operator sets the specific location and point in time at which force should occur. For the large part, these have been ‘defensive’ weapon systems, often fixed in their location and co-located with their human operator, and with fixed ‘target profiles’ limiting the types of targets they can engage.

However, in recent years technological developments have led to an increase in the capabilities of weapon systems with autonomous functions, including a larger geographical area and duration of operation, and the potential for more complex or changeable target profiles. This expanded scope of independent operation has raised concerns regarding how human ‘control’ is applied to such systems, or how they are understood to be sufficiently ‘predictable’. The changing nature of target profiles has also raised related concerns, as well as perceived risks of a ‘dehumanization’ in the use of force. These themes are central to international legal discussions regarding constraints on autonomy in weapons systems.

Introduction