In the past years, a growing number of voices are calling for urgent discussion on weapon systems with increasing autonomy. The discourse on these emerging technologies takes place at the political level under the auspices of the Convention on Certain Conventional Weapons (CCW) but the issue is also widely reflected upon in academic articles, conference papers, (governmental) reports and other papers. As the issue of autonomous weapons is multifaceted and multidisciplinary, the community involved in the discourse is too. What all of these actors have in common, however, is that they are all part of a discourse within which semantic disputes are prominent. Although different terms are suggested to describe autonomous weapons, a number of terms stand out. Particularly prominent terms are ‘autonomy’, ‘target selection and attack’ and ‘human intervention’. These terms are the basis of a widely used and broadly accepted definition describing autonomous weapons as weapons that are capable of selecting and attacking targets without human intervention. At first glance, the definition and language seem quite clear; nevertheless, upon further examination this definition reveals a number of complications. The aim of this paper however is not to propose a definition that would solve linguistic disputes (if such a definition would even be viable); rather it takes a more external perspective with the purpose to illustrate how a common vocabulary can complicate the discourse on autonomous weapons when the terms involved are not commonly understood or lack consistent interpretations. Hence, this article functions as a map for understanding the debate on autonomous weapons–imperative for anyone who would decide to participate in it.