Hitoshi Nasu on “The Alpagu Autonomous Attack Drone: What the Future Holds for the Humanity in the Battlefield?”
Today’s post is something of a follow-up to our recent discussion about the U.S. airstrikes on the drone capabilities of Iranian-backed militias (see here). Professor Hitoshi Nasu of the University of Exeter examines the legal issues associated with a sophisticated new Turkish drone that incorporates artificial intelligence (AI) technology to deadly effect.
Among other things, Professor Nasu assesses some of the challenges of reviewing the legality of this system, including its advanced target recognition process, that is, “the machine’s ability to detect and recognize targets using real-time image processing capabilities.” He also unpacks the weapon’s effects, particularly as the whether they would be unlawfully indiscriminate.
This is a great opportunity to get up to speed with the latest developments in weaponized drones and to consider what integrating AI software might mean on the battlefield. (In 2014 I touched upon the implications of drones being linked to facial recognition technology in “The Hyper-Personalization of War: Cyber, Big Data, and the Changing Face of Conflict”)
Here’s Professor Nasu’s very interesting (and timely!) essay:
The Alpagu Autonomous Attack Drone:
What the Future Holds for the Humanity in the Battlefield?
by Professor Hitoshi Nasu
On June 18, 2021, a successful test of Turkey’s fixed-wing autonomous tactical attack drone, Alpagu, was reported. The Alpagu, which was developed by STM Defense Technologies Engineering and Trade Inc., hit its target with pinpoint accuracy. This drone is the latest addition to the inventory of autonomous weapons equipped with artificial intelligence technology.
The focus on the topic of autonomous weapons systems has recently shifted from academic debates in international forums to combat operations on the battlefield. With the reported use of the Kargu-2 drone during the 2019 conflict in Libya, an effective deployment of drones such as Bayraktar TB-2 during the 2020 Nagorno-Karabakh conflict, and most recently, Hamas’ use of several autonomous GPS-guided submarines during the May 2021 fighting in the Gaza Strip, this shift in focus is to be expected.
Like Israel’s Harop “Suicide Drone”, the Alpagu is a loitering munition capable of creating pinpoint damage to neutralizes targets that meet pre-set criteria. Essentially, it is a hybrid between a drone and a guided missile. Weighing less than two kilograms, it is highly maneuverable and utilizes image tracking software to identify and strike targets.
As with all new weapons and weapons systems, the Alpagu must be subject to weapons review to ensure its ability to comply with the law of armed conflict. This obligation falls not only on Turkey but also on any other States that are studying, developing, acquiring or adopting it.
All the States that are party to the Additional Protocol I to the Geneva Conventions have an obligation to conduct weapons review. Article 36 of the Additional Protocol I reads:
In the study, development, acquisition or adoption of a new weapon, means or method of warfare, a High Contracting Party is under an obligation to determine whether its employment would, in some or all circumstances, be prohibited by this Protocol or by any other rule of international law applicable to the High Contracting Parties.
There is no standardized methodology for conducting weapons review. The range of legal obligations, against which this review is conducted, varies depending on each country’s treaty commitment.
Even so, two general principles are universally accepted under customary international law: (1) a prohibition on the use of weapons that indiscriminately affect both lawful targets and civilians (i.e., indiscriminate weapons); and (2) a prohibition on the use of weapons that are calculated or of a nature to cause superfluous injury or unnecessary suffering.
The first principle is designed to protect civilians from the effects of the weapon, whereas the second principle aims to ensure that the injury or suffering inflicted upon lawful targets does not outweigh the military necessity underlying the choice of particular weapon.
In light of these two principles, particular attention should be paid to the following factors when reviewing the legality of the Alpagu drone: the weapon’s ability to recognize and discriminate the intended target; and the weapons effect upon the target.
The Alpagu drone is equipped with an image processing software for detection, tracking and classification of targets. Like the Kargu-2 attack drone, it is designed to be an anti-personnel weapon capable of autonomous precision strikes. It is designed to produce minimum collateral damage by detonating an explosive charge close to its target.
As this author discussed elsewhere, the machine’s ability to detect and recognize targets using real-time image processing capabilities has its limits. This is especially the case when it cannot rely on visual indicia that are commonly observed among legitimate military targets. The target recognition algorithm must be robust enough to identify protected civilians and those who are wounded, sick, or otherwise express an intention to surrender so as to suspend engagement with them as required by law.
These problems can be circumvented by limiting the weapon’s operating parameters to a particular battlefield environment or by defining the range of targeting options more narrowly than the range permitted by law. For example, the Alpagu’s object classification software can be programmed to only detect and engage high priority targets, such as those who are firing weapons. Whether the Alpagu drone is indiscriminate in nature or not may well be situation-dependent.
A legally significant technological consideration is the risk of error. Systematic errors that arise from either the design of a weapon system or the weapon system’s interaction with the environment in which it is used are relevant considerations. Such errors increase the uncertainty and unpredictability of the weapons system which can lead to biased applications during the deployment. These systematic errors should be addressed during the development and manufacturing stages but can also be compensated for by adjusting the operational parameters.
Random errors, on the other hand, are more difficult to control. As Backstrom and Henderson observed, the use of artificial intelligence for target recognition inevitably requires the legal standard for lawful target identification to shift from a question of subjective belief to an objective and measurable quantity expressed as a statistical probability (p. 495). Exactly what constitutes a legally acceptable margin of error remains debatable. At any rate, field testing alone does not provide a sufficient fidelity of data to be reliable in the field.
As a hybrid between a drone and a guided missile, the Alpagu is capable of neutralizing its target with pinpoint accuracy. Although it is not clear which type of live ammunition was used during the latest test, the Alpagu appears to cause damage through an explosive charge that is detonated close to the target, rather than with a kinetic force created by physical impact.
Its effect is unlikely to be indiscriminate because of the limited range of its impact, with minimum collateral damage expected. Nonetheless the Alpagu’s particular method of directing lethal effects must be examined to ensure that it does not cause superfluous injury or unnecessary suffering, which is prohibited universally under customary international law.
There are no clear-cut criteria for what types of weapons cause superfluous injury or unnecessary suffering. There is, however, a limited range of specific weapons that are already banned for that reason – certain types of explosive projectiles, expanding bullets, and non-detectable fragments. These weapons are specific manifestations of what constitutes a weapon that is calculated or of a nature to cause superfluous injury or unnecessary suffering and, as such, provide a baseline for legal assessment of new weapons.
In particular, the lethal effects produced with pinpoint accuracy by the Alpagu drone may be seen as comparable to those created by explosive projectiles. It is prohibited to use projectiles of a weight below 400 grammes that are explosive or charged with fulminating or inflammable substances. But the prohibition is very specific to the use of light-weight explosive projectiles designed to explode upon impact with the human body.
Due to its light-weight flight capacity, the Alpagu drone is arguably moving closer to prohibited explosive projectiles than other guided projectiles that already exist. It may nonetheless be distinguishable due to its weight and fuzing mechanism (unless it is designed to explode upon impact with the human body). Even though the lethal effect of the Alpagu drone is similar or comparable to those of prohibited explosive projectiles, it may still be assessed as lawful for reasons of legal technicality.
Weapons Law “Paradox”
The legality of any new weapon can be a subject of debate. The Alpagu drone could be characterized as inhumane since its targeted lethal effect and pinpoint accuracy leaves no chance of survival. Despite this potential characterization, the Alpagu is not unlawful because of a legal technicality.
If, on the other hand, the Alpagu drone were to be designed to incapacitate human targets by releasing novel chemical agents tailored to enhance temporary physical disabling effects, the legal analysis would be significantly different. In this case, the Alpagu would be prohibited as a means and method of warfare because of a comprehensive ban on chemical weapons and associated munitions and devices. This is so even if it may be seen to be more humane to leave human targets physically disabled for a certain period of time than killing them.
The development and improvement of advanced technologies are expected to enable further miniaturization of explosive-laden drones with the capacity to swarm multiple targets. Due to the “paradox” of weapons law, such advanced attack drones will continue to be deemed lawful despite the fact that they can be designed to pursue comparable military outcomes in a more humane way.
At the same time, a variety of counter-drone technologies, such as jamming, electromagnetic weapons, and laser, are becoming readily available. A widespread availability of these capabilities will negate or significantly reduce tactical advantage of these autonomous attack drones.
The future regulation of the Alpagu drone, and its more advanced models, hinges on the intricate balance between technological competition, strategic choice and humanitarian considerations.
About the author:
Hitoshi Nasu is Professor of International Law at the University of Exeter. Prior to his current appointment, he held academic posts at the Australian National University, where he was also Co-Director of the Centre for Military and Security Law and the Australian Network for Japanese Law. Currently, he is also a Senior Fellow at the Stockton Center for International Law, United States Naval War College.
He publishes widely in the field of public international law, with particular focus on international security law, the law of armed conflict and the law of weaponry. His expertise extends to a wide range of international security law issues, such as collective security, peacekeeping, the protection of civilians in armed conflict, and in different domains including maritime, cyber and space. In particular, he has produced numerous publications that address a variety of legal issues arising from military applications of new technologies, such as nanotechnology and artificial intelligence, as well as contemporary security challenges in the Indo-Pacific theatre.
As always, remember what we like to say on Lawfire®: gather the facts, examine the law, evaluate the arguments – and then decide for yourself!