New Technologies in International Law / Tymofeyeva, Crhák et al.

are imagined to be, comprised of a group of individual but identical elements 36 (perhaps 10’s or hundreds, but theoretically thousands). 37 The elements are comparatively small, 38 but because each member is identical there is no single ‘leader’. This way the swarm continues to function where individual members become inoperative. One key benefit of this type of swarm is its adaptability. Indeed, a swarm may be used for virtually any mission. They might, for example, act as cloak, protecting a piloted aircraft by disturbing a ground-based radar detection system. 39 A swarm might also be deployed into an urban battlefield environment with each element programmed, as previously discussed, to target individuals or objects selected according to certain predefined criteria. The technology undoubtably has benefits, which is, no doubt, a driver behind swarm development, 40 and deployment. 41 However, here, it is also important to distinguish fully autonomous swarms. There is no widespread agreement, but three terms regularly appear in the wider debate regarding AWS. These are, (i) human-in-the-loop; (ii) human-on-the-loop; and (iii) human-out of-the-loop systems. For present purposes, only an overview is of these elements is necessary. First, a human-in-the-loop system is one in which a human is involved in the decision-making process. For example, a human might identify the target (marking it with a cursor), before authorising a swarm deployment. There may be no more human involvement, however, in this case at least one human forms part of the wider weapons system. 42 A human-on-the-loop system is one which can operate independently, but where a human supervises the decision-making process. Here, the human can intervene at any time, if the system malfunctions or operates in an unexpected manner. A useful example of this is the PHALANX weapon System which has existed for decades. 43 36 See e.g., slaughterbots (n 24). 37 See generally e.g., David Hambling, ‘ The US Navy wants swarms of thousands of small drones’ (24 October 2022, MIT Technology Review). Available at accessed 1 November 2023. Also see, slaughterbots (n 24). 38 Hambling, ibid. Indeed, one of the reasons swarms are both desirable and controversial, is that they have a relatively low cost and rudimentary design. 39 See in general e.g., Claudia Conte C, Verini Supplizi S, de Alteriis G et al, ‘Using Drone Swarms as a Countermeasure of Radar Detection’ (2023) 20 Journal of Aerospace Information Systems 2. Coco A, Dias T, and van Benthem T, ‘Illegal: The SolarWinds Hack under International Law’ (2022) 33(4) European Journal of International Law 1275. 40 See e.g., United Kingdom Government, ‘Press release: £2.5-million injection for drone swarms’ ( Gov. uk , 28 March 2019) accessed 1 November 2023. 41 The Israel Defence force is believed to have deployed an armed swarm for the first time in 2021. See e.g., Hambling D, ‘Israel used world’s first AI-guided combat drone swarm in Gaza attacks’ ( New Scientist , 30 June 2021) accessed 7 November 2023. 42 See e.g., ‘Javelin’ ( Lockheed Martin ) accessed 1 November 2023. 43 Raytheon, ‘Phalanx Weapon System’ ( Raytheon ) accessed 1 November 2023.

20