Justice and Warfare AI: The Case of Ukraine and The Threat to Democracy

By Vittoria Patti

Source: TIME

Uses of AI in Warfare

In the past decade, we have witnessed a radical technological revolution driven by rapid advances in AI. In fact, AI has adopted an increasingly prominent role in everyday life, ranging from areas such as education and healthcare to agriculture and transportation. Warfare is no exception. With the growing development and deployment of AI-powered weapons that operate autonomously, without any human intervention, AI has reached a point of no return: the ability to make decisions on life and death. These AI-powered systems include both lethal and non-lethal autonomous weapons that can identify and engage targets without human input, as well as surveillance platforms and data processing algorithms that guide battlefield strategies. While some of these instruments are being promoted as enhancing precision and minimising error, removing human judgement from the process comes with its own complications.

The latest attested instance of AI use on the battlefield can be identified within the ongoing conflict in Ukraine, one that has been defined as “the first AI war”. In fact, on 12 March 2022, Ukrainian soldiers in Kyiv retrieved wreckage of the Zala KUB-BLA, a kamikaze Unmanned Aerial Vehicle (UAV) produced by the Kalashnikov Group of Soviet origin. The drone is capable of real time recognition and classification of detected objects using AI through a particular autonomous mode. The testimony of its use signaled to the world that AI armaments are no longer merely a futuristic prediction, but that they have already entered the violent realities of the present day.

This event is not isolated. In recent years, other global actors such as the United States and China have invested heavily in AI-enabled weaponry, accelerating what can be called the “AI arms race”. As defence budgets inflate and private businesses compete to supply increasingly advanced systems, parallel concerns on the matter have emerged as the international community attempts to cope with the impending consequences of this evolution.

The Accountability Gap

Such innovations carry not only ethical dilemmas, but also prompt broader legal questions on accountability: who is legally liable when violations of the laws of armed conflict occur? Who can respond to these determinations in a court of law? Certainly, a machine cannot be put on the stand or be sent to jail. So, to whom should the fault be attributed? To programmers, who designed the algorithms at the root of these systems; to industrial and defense companies, who manufactured the tool; to commanders ordering activation; to soldiers executing; or to governmental leaders allowing their employment? With more people involved in the process in comparison to traditional weapons, the line of attribution is blurred and it becomes increasingly harder to trace back responsibilities along the accountability chain.

This challenge leads to what scholars refer to as the accountability gap: 

“the impossibility of transferring accountability to machines combined with the absence of any moral agent to whom responsibility could be ascribed [...], a situation when no entity could be legally held accountable for the wrong done”. 

In essence, when a machine performs a task autonomously, accountability is dispersed so widely that it may disappear altogether.

However, in warfare, accountability is not optional. It is an essential means to uphold the rule of law and therefore guarantee respect for and compliance with human rights. Moreover, by ensuring the punishment of wrongful acts, accountability also serves to deter future offenders and prevent harm on an international scale. On the other hand, when actions are delegated to non-human actors, the absence of accountability measures threatens the protection of victims from abuse and endangers the very safeguards of justice. If states can claim that harm was caused by a non-human, and therefore a non-prosecutable actor, the door opens to impunity. Worse, this loophole may even encourage riskier uses of force as officials hide behind the shield of technology.

Why Does It Matter for Democracy?

In democracies, accountability of state governments and their officials is more crucial than ever. Democratic accountability represents a fundamental obligation to the citizens, the assurance that those in power are answerable for their actions, especially when the latter involve violence. Yet, if militaries can outsource such critical and sensitive decisions and actions to AI, no one is left to answer to them. The accountability gap that results from AI use in war thus has concerning implications for democracy itself.

In fact, a lack of accountability measures weakens the system of checks and balances between the branches and agencies of a government, a structure carefully designed to prevent abuses of power. In turn, this leads to decreased trust in public institutions, undermining the principle that democratic governments are answerable to voters and putting the whole democratic order in jeopardy. Over time, an absence of accountability mechanisms can fuel political polarisation and give rise to the dangerous perception that governments are no longer responsive or transparent.

The consequences of this could be catastrophic and extend beyond the domestic sphere. In fact, as democracies betray their own standards of morality, dignity and human rights, there will be no way of maintaining the credibility of an international system of rules and norms that was built on those same standards, paving the way for chaos and disorder. Therefore, democratic erosion constitutes a risk not only for democracies themselves, but for the international community as a whole.

The Road to a Solution

The most widely supported solution identified by experts is the establishment of a mandate for meaningful human control – clear and enforceable boundaries that would require the active involvement of a person at every stage of AI applications for military purposes. The implementation of this concept into reality would ensure that decisions never rest solely on a computer and that actions are never carried out solely by a robot, leaving room for accountability to be pursued. It would reaffirm the principle that life-and-death choices remain firmly in human hands, preserving a space for justice.

Yet, the path to such regulation is far from simple. In fact, existing international law remains ill-equipped to deal with the unique challenges posed by these new technologies. Codes such as the Geneva Conventions were drafted long before the emergence and spread of AI; consequently, they lack the needed specificity to regulate the unique issues that come with AI. For example, Article 36 of the 1977 Additional Protocol I (API) to the Geneva Convention establishes a mandate for the legal review of new weapons. However, the statute’s vagueness may lead to challenges in its application to this case, as AI technologies have multiple purposes and may not fall within the weapons category. 

Another norm that applies to emerging AI technologies in warfare is the Martens Clause, introduced in the preamble of the 1899 Hague Convention II, which establishes that all scenarios not mentioned in the text should still fall under "the laws of humanity and the dictates of public conscience”. The section was meant as a reserve mechanism to govern situations not explicitly spelled out in International Humanitarian Law, such as AI. However, its wording risks being too vague, as “the laws of humanity” and “the dictates of public conscience” are subject to different interpretations.

Meanwhile, diplomatic efforts have also struggled to make progress. Countries engaged in negotiations, as within the Group of Governmental Experts forum under the UN Convention on Certain Conventional Weapons, have encountered resistance due to complex power dynamics and geopolitical fragmentation, failing to find an effective compromise. Where the dual and rapidly advancing nature of the technology makes an agreement more difficult, entities have come to disagree not only on the scope of restrictions but also on the backbone definitions of concepts such as autonomy and human involvement. This unresolved debate finds itself in a wider crisis of global governance, further hindering consensus.

Nevertheless, the urgency of creating appropriate compliance mechanisms continues to grow sharper. Without a decisive initiative (ideally in the form of a new binding treaty that could tackle the shortcomings of current international legal frameworks) society risks facing a subtle and gradual dissolution of the core democratic value of accountability. Regulating AI in warfare is not only a matter of safety; it is about defending the deeper political values of democracy and justice.

---

Vittoria Patti is a geopolitical analyst and current Head of Communications and Media at Orizzonti Politici, a think tank based in Italy. She holds a Master’s degree in Conflict Studies & Human Rights from Utrecht University in the Netherlands and a Bachelor’s degree in Political Science & International Relations from the University of South Florida in the United States of America, where she was also a legislative intern at the Florida House of Representatives. With a strong interest in law and justice, her research explores the intersection of legal studies with global security and foreign affairs. At the moment, she is working on a project as part of a research fellowship on European defence industry and policy.