1. Introduction
It is certainly desirable to incorporate autonomy into weapons systems and battlefield decision-making processes, as this can enhance precision and reliability without sacrificing the robustness and flexibility of human intelligence. However, the use of autonomous weapons in armed conflict raises specific legal challenges, particularly with regard to attributing criminal liability for unlawful targeting. It is therefore important not only to establish a basis for attributing liability in connection with autonomous weapon systems, but also to determine appropriate mechanisms for compensating damage caused by their use. Since their emergence, military AI-enabled technologies have challenged conventional conceptions of criminal liability and raised complex questions concerning legal frameworks for providing compensation for harm.
There are numerous challenges in determining liability for serious crimes and violations of international humanitarian law committed through the use of autonomous weapons. Although the management of military operations remains primarily a human task, autonomous systems operate on the basis of target signatures and parameters previously established by programmers, which enable them to search for, identify, track, and prioritize targets. If this mode of operation results in the killing of civilians or the destruction of civilian objects, complex questions of criminal law arise. These include questions concerning the foreseeability of the autonomous system’s operation under human supervision—the so-called the problem of many hands—and whether there is a sufficient basis for attributing criminal liability at an appropriate level.
To determine the issues of criminal liability arising in cases where weapons systems unlawfully use lethal force, it is necessary to consider the framework established by the four Geneva Conventions. The Geneva Conventions require states to adopt criminal measures to punish persons who commit, or order the commission of, grave breaches of the Conventions. Grave breaches of the Geneva Conventions constitute war crimes, with violations of the principle of distinction providing a particularly clear example. An autonomous weapon system may target a civilian or a combatant outside the battlefield for a variety of reasons. Such systems may misinterpret the circumstances in which they operate or behave in ways that their programmers did not anticipate.
The present study aimed to examine how international criminal law responds to the challenges posed by the use of autonomous weapon systems, with particular emphasis on the criminal liability of those involved in the commission of international crimes involving such weapons. The development of any modern weapon is a complex process involving multiple teams, individuals and organizations at every stage, from the initial conceptualization and design through development, production, and ultimately deployment. A sophisticated autonomous weapon system is therefore rarely developed by a single individual. Rather, it is the product of numerous actors and groups working on different subsystems that may have complex interdependencies. In such circumstances, attributing individual responsibility for a violation—whether intentional or accidental—and identifying the person or persons responsible for the subsequent conduct of a weapons system that results in the commission of an international crime may be particularly difficult in attempting to initiate criminal proceedings. Against this backdrop, this study sough to examine whether the existing framework of international criminal law is capable of adequately addressing the challenges arising from crimes committed through the use of autonomous weapons. Given the involvement of numerous actors in the design, development, programming, training, production, and deployment of these systems, the central question is: Who bears criminal liability for international crimes committed through the use of military AI?
2. Literature Review
It seems there is a lack of Persian-language scholarship examining approaches to international criminal liability arising from the use of AI-enabled weapons systems in the commission of international crimes. Leave aside general discussions on autonomous weapon systems in Iran, it seems there is no Persian-language research that directly and specifically addresses the subject.
At the international level, given the growing importance of military AI and AI-enabled weapon systems, prominent scholars have produced significant scholarship on AI-enabled autonomous weapon systems. Although some of these publications touch upon questions of international criminal liability arising from the deployment of such weapons in armed conflict, it seems there is no study that comprehensively examines the specific subject matter addressed in the current inquiry.
Moreover, the existing scholarship tends to address international criminal liability for autonomous weapons in the military domain in a fragmented and inconsistent manner. No research seems to have systematically engaged with the central hypothesis of this research, either by articulating and substantiating it or by challenging and refuting it with a comparable degree of coherence and methodological rigor.
3. Materials and Methods
This study employed a descriptive–analytical research methodology. It used a library research protocol to collect the data from relevant scholarly and legal sources.
4. Results and Discussion
Although the benefits of AI for humanity are undeniable, the concerns surrounding criminal liability for crimes committed through the use of military AI cannot be readily disregarded. Autonomous weapon systems require states and legal scholars to reconsider existing legal doctrines and approaches to attribution. States should therefore recognize AI-enabled systems as having a degree of legal (electronic) personality. At the same time, holding AI-enabled systems themselves criminally liable or imposing criminal penalties on them is incompatible with fundamental principles of criminal law and does not appear legally or conceptually coherent. Rather, human moral agents must remain accountable for the conduct of these systems. However, because multiple individuals and groups may be involved in the design, development, production, programming, deployment, operation, or command of military AI, identifying and holding a specific individual criminally responsible may be difficult, if not impossible, under existing legal frameworks. Accordingly, manufacturers, operators, commanders, programmers, corporate leaders, and senior defence officials should be subject to appropriate forms of accountability where they exercise relevant authority or effective control over autonomous weapon systems and where their conduct contributes to the commission of international crimes through the use of such systems.
According to the findings, none of the existing or proposed concepts provides an adequate and comprehensive solution. For this reason, the development of autonomous weapons should be stopped. However, this solution is unlikely to be achieved, as no state wants to lose the arms race or fall behind in the development of a powerful military, which could result from abandoning autonomous weapons. Such a development is even more unrealistic given the contemporary geopolitical situation in certain parts of the world, including Taiwan, Kosovo, Syria, Karabakh, and Palestine. Therefore, greater efforts should be made to develop an appropriate framework for determining the responsibilities of those involved in the development and use of autonomous weapons. One possible alternative is to ban or restrict the production and use of AI-enabled autonomous weapons within the framework of international humanitarian law. Strict liability is also one possible legal framework for establishing accountability for AI-related harms. Although there has so far been considerable resistance to this form of liability in international jurisprudence—particularly because of the requirement to establish a specific state of mind or degree of fault—the acceptance of strict liability would not only eliminate legal uncertainty but also resolve many of the problems associated with identifying the human actors behind autonomous agents and holding them accountable.
Moreover, liability extends far beyond individual criminal liability. State liability for the deployment of autonomous weapon systems that result in serious violations of the laws of armed conflict is also significant. States acquiring new weapon systems or modifying existing ones should consider imposing product liability requirements on developers, manufacturers, and sellers of such systems to help cover the costs of compensation in the event of a breach of international legal obligations. Finally, multilateral cooperation among states is necessary to ensure that AI-enabled weapons are equipped with black boxes that can assist the International Criminal Court or other international tribunals in determining their role in the commission of international crimes. These black boxes should contain a record of the individuals involved in the production, design, programming, planning, and operation of the systems, as well as relevant information concerning their respective roles.
5. Conclusion
Although attributing direct criminal liability to a specific individual in this context is complex, an interpretive application of Articles 25 and 28 of the Rome Statute, together with the doctrine of hierarchical responsibility, can provide a basis for attributing liability to individuals or groups of individuals who play a decisive role in the design, development, deployment, or command of AI-enabled autonomous weapon systems. Accordingly, human actors who occupy a position within the chain of supervision or command and exercise effective control over others may be held liable.