Stopping Killer Robot Proliferation to Tyrants and Terrorists: Why A Global Ban Is Neither Necessary Nor Sufficient

Abstract Autonomous weapon systems (AWS) – military robots able to select and engage targets with lethal force on their own – are a subject of voluminous ethical debates, with many calling for a universal ban on AWS. One of the most prominent arguments for such a ban is based on the threat of anti-personnel AWS – killer robots – to terrorist groups and totalitarian regimes. As these weapons would be ideal for terrorist or oppressive ends, and as having such ends realized is very bad indeed, this would be a disastrous outcome. Although agreeing with this threat assessment, I argue that a universal ban on AW would be neither necessary nor sufficient to stop these outcomes, while the efforts to create a ban divert resources from other more feasible countermeasures. To show this, I critically analyze two most detailed extant accounts of the proliferation threat put forward by ban proponents. Based on that analysis, I propose a number of more specific policy measures that could prevent most dire forms of killer robots’ proliferation while remaining politically feasible and compatible with subsequent pursuit of more ambitious non-proliferation agenda.