Traditionally it is argued that AWS are a challenge for IHL, but Chapter 4 pleads the opposite. IHL principles and rules are long and well-established and, therefore, it is up to roboticists to prove that AWS will reliably comply with IHL. For this reason, the chapter begins with an analysis of the inherent values of IHL and the two orders of requirements established in IHL in order to guarantee the legitimacy of AWS. First, a priori requirements that concern the legitimacy of AWS as weapons per se (Article 35 and 36 API). The second set of requirements are a posteriori; they regard the legitimate use of AWS. In this respect, a thorough analysis of the rules of Precaution (Article 57 API), Distinction (Articles 49; 51; 52 API) and Proportionality (Articles 51(5)(b) and 57(5)(a)(iii) API) is provided. While the rules of Precaution enshrine the obligations impending upon military commanders, other rules, such as the Principle of Distinction and Proportionality, pose specific challenges in how AWS will be able to comply with such rules. For that purpose, the chapter looks at new technological developments which may facilitate the compliance with the IHL rules (facial recognition; collateral damage estimative methodology), but by no means insure respect for IHL in all circumstances.

Chapter 3 has a more philosophical character as far as it seeks to understand the richness of the concept of ‘autonomy’ when applied to technology of warfare, in order to build a coherent and well-structured legal discourse. This concept is a multidisciplinary one, and it has different semantic values depending on the discipline. The chapter argues that computer language is immersed in a phenomenon named ‘misappropriation of language’. Thus, when arguing about autonomous systems, it is important to clarify that ‘autonomy’ should not be understood in the legal or philosophical way but rather as the capacity of an artificial system to develop tasks without human supervision. Once this concept has been clarified, the chapter examines the ontological argument that has been argued by some scholars and roboticists. That argument sustains the possibility of considering autonomous systems as legal agents, under the Latin maxim ‘qui facit per alium, facit per se’. The chapter explains why the legal theory of delegation cannot be applied to AWS and why AWS should not be seen as independent legal agents but rather remain within the category of weapons systems.