The duty to verify the target when using autonomous weapons

Abstract

The article examines the conditions for compliance with the duty to verify targets under Article 57(2)(a)(i) of Additional Protocol I to the Geneva Conventions when lethal autonomous weapon systems (LAWS) are employed. The central problem arises where the human decision to employ a weapon system precedes the determination of the specific object of attack, while its detection, identification, and individuation are subsequently performed autonomously by the system. The article aims to establish the criteria under which reliance on autonomous target identification may satisfy the obligation to do everything feasible to verify that an object is a lawful target. Drawing on a doctrinal analysis of Article 57(2)(a)(i), the ICRC Commentary, State military manuals, and contemporary scholar­ship, the article identifies the normative components of the duty of target verification and applies them to autonomous target selection. It argues that autonomy does not alter the standard imposed by Article 57(2)(a)(i), but changes the manner in which that standard is implemented. Although developers of LAWS do not thereby become direct addressees of the duty to verify, information generated through testing and valida­tion concerning system accuracy, errors, and opera­tional limitations becomes legally relevant to those responsible for an attack. The article proposes a legal test under which compliance depends on whether reliance on autonomous target identifi­cation is justified in light of the systemʼs validated reliability, conditions of use, quality of its infor­mational basis, known limitations, and residual uncertainty.