Preventing Autonomous Weapon Systems from Being Used to Perpetrate Intentional Violations of the Laws of War: TLI Think! Paper 7/2018

TY - GEN

T1 - Preventing Autonomous Weapon Systems from Being Used to Perpetrate Intentional Violations of the Laws of War

T2 - TLI Think! Paper 7/2018

AU - Homayounnejad, Maziar

AU - Overill, Richard Edward

PY - 2018/4/17

Y1 - 2018/4/17

N2 - Autonomous Weapon Systems (AWS) are essentially weapon systems that, once activated, can select and engage targets without further human intervention. While these are neither currently fielded nor officially part of any nation’s defense strategy, there is ample evidence that States and defense contractors are currently developing AWS for future deployment. In the near-term, such weapons will mainly follow the wide-area loitering concept, which largely attacks targets from a distance. In the longer term, it is likely that militaries will develop and field bipedal / humanoid robots with a degree of manual dexterity, and the capability to undertake both rescue missions, and ‘close quarters combat’ with human subjects. This raises questions on how such machines may be used in times of armed conflict, occupation or internal strife. Given their potential capability for human-like physical interactions, there is a concern – following empirical research – that such machines may be deliberately programmed to commit inhumane acts against civilians, such as rape, broader sexual violence, and torture. Furthermore, as AWS units will be programmed to logically follow ex ante instructions, there is a real risk that such terrorizing acts will occur more easily, more frequently and, once initiated, indefinitely; potentially unconstrained by the moral or ethical brakes, or the physical limitations of human soldiers.However, there are also a number of legal and technical safeguards that may mitigate this risk, thereby realigning the technology of AWS with its original and intended purpose. These include: 1) capability limitations in the near-term; 2) positive design-led safeguards; 3) a ‘kill switch’ function; 4) pre-emptive legal safeguards to enhance non-proliferation efforts; and 5) digital evidence to aid the prosecution of war crimes. Importantly, none of these are individually watertight, but collectively they offer a package of safeguards that should sufficiently mitigate the risk of sexual violence, torture and other forms of repression using AWS.

AB - Autonomous Weapon Systems (AWS) are essentially weapon systems that, once activated, can select and engage targets without further human intervention. While these are neither currently fielded nor officially part of any nation’s defense strategy, there is ample evidence that States and defense contractors are currently developing AWS for future deployment. In the near-term, such weapons will mainly follow the wide-area loitering concept, which largely attacks targets from a distance. In the longer term, it is likely that militaries will develop and field bipedal / humanoid robots with a degree of manual dexterity, and the capability to undertake both rescue missions, and ‘close quarters combat’ with human subjects. This raises questions on how such machines may be used in times of armed conflict, occupation or internal strife. Given their potential capability for human-like physical interactions, there is a concern – following empirical research – that such machines may be deliberately programmed to commit inhumane acts against civilians, such as rape, broader sexual violence, and torture. Furthermore, as AWS units will be programmed to logically follow ex ante instructions, there is a real risk that such terrorizing acts will occur more easily, more frequently and, once initiated, indefinitely; potentially unconstrained by the moral or ethical brakes, or the physical limitations of human soldiers.However, there are also a number of legal and technical safeguards that may mitigate this risk, thereby realigning the technology of AWS with its original and intended purpose. These include: 1) capability limitations in the near-term; 2) positive design-led safeguards; 3) a ‘kill switch’ function; 4) pre-emptive legal safeguards to enhance non-proliferation efforts; and 5) digital evidence to aid the prosecution of war crimes. Importantly, none of these are individually watertight, but collectively they offer a package of safeguards that should sufficiently mitigate the risk of sexual violence, torture and other forms of repression using AWS.

U2 - 10.2139/ssrn.3123254

DO - 10.2139/ssrn.3123254

M3 - Other contribution

VL - TLI Think! Paper 7/2018

PB - Transnational Law Institute, KCL

ER -