EJIKU Justine
The world must never reach a point where humans lose control of AI and machines in combat. That is a red line that humanity must never cross.
War, it would seem, has finally become efficient. Decisions that once demanded the burden of human reflection, who lives, who dies, what is gained, and what must be spared are now increasingly assisted, if not shaped, by algorithms that neither see suffering nor feel its weight. This quiet transformation stands in stark contrast to the origins of International Humanitarian Law (IHL), born from the anguish of Henry Dunant after the Battle of Solferino, a moment that demanded not greater efficiency in killing, but greater humanity in war.
From that foundation emerged a legal order anchored on a careful balance between military necessity and the enduring imperative to preserve human dignity. Yet, as artificial intelligence assumes an ever more prominent role in contemporary hostilities, particularly in the Gaza conflict, one is compelled to ask whether this balance is quietly being rewritten. Can a framework designed to humanize war withstand a battlespace in which the tempo of violence is accelerated, the scale expanded, and critical judgments increasingly mediated by machines?
This article argues that while artificial intelligence promises precision and operational advantage, it simultaneously unsettles the human-centred logic of IHL, inviting a necessary and urgent re-examination of how law can meaningfully regulate warfare in an age where humanity risks becoming an afterthought. The Conflict in Gaza among other things reveals to us that the world must never reach a point where humans lose control of AI and machines in combat. That is a red line that humanity must never cross.
As early as 2017, commentators such as Tom Simonite warned that the rise of military AI could rival the transformative and deeply unsettling impact of the nuclear bomb, fundamentally altering not only how wars are fought but how power itself is projected and perceived. In the years that followed, this once-speculative comparison has gained renewed force, with contemporary reflections capturing the stark reality of technological evolution in even more direct terms: that cloud and AI are the bombs and bullets of the 21st century.
The Gaza conflict provides a particularly illustrative case study of how AI is operationalized in practice. The Israeli military, long associated with technological sophistication, has increasingly incorporated data-driven systems into its targeting processes. At the heart of its operations lies AI-based systems which are reportedly capable of sifting through intercepted communications, movement patterns, mapping the daily routines and locations of key individuals and other intelligence inputs, allowing for the identification of high-value targets and the use of AI powered weapons to carry out attacks.
The scale and speed enabled by such systems mark a significant departure from traditional targeting practices. Senior military officials, including Aviv Kochavi, have acknowledged that AI-powered processes have dramatically increased the volume of targets generated, with the system reportedly capable of producing hundreds of potential targets within a single day, figures that far exceed previous operational capacities.
At the same time, the integration of artificial intelligence into targeting has introduced new layers of complexity in the assessment of civilian harm. Reports indicate the use of algorithmic tools to estimate the presence of civilians in targeted structures, employing simplified evaluative models such as colour-coded risk indicators to inform strike decisions. While such systems are presented as enhancing precision and compliance with legal obligations, their reliability remains contested. The opacity of AI decision-making processes, combined with the limitations of available data, raises serious questions about the accuracy of such assessments.
Accounts from individuals familiar with these systems describe a process where targets are generated, reviewed, and acted upon at high speed, often under pressure to maximize output. In such an environment, human oversight while formally retained may risk becoming procedural rather than substantive. If the application of force is increasingly mediated by systems prioritising efficiency, scale, and speed, the extent to which the foundational principles of humanity, military necessity, and careful judgment can be meaningfully upheld becomes an issue not only of legal interpretation, but of the very future of restraint in warfare.
Although it has not been conclusively established that the following incidents were directly aided by artificial intelligence, the pattern of outcomes suggest so, and reflects the very risks and operational challenges associated with its use in modern warfare. On 27 May 2024, a strike in Rafah later described by Benjamin Netanyahu as a “tragic mistake” killed at least 45 civilians and injured hundreds in a densely populated refugee camp. Earlier, on 1 April 2024, the drone strike on a World Central Kitchen convoy killed seven aid workers after a chain of misidentifications and overlooked intelligence.
On 25 August 2025, the bombing of Nasser Hospital resulted in the deaths of journalists and civilians, with the Israeli military admitting the strike had been approved but its consequences misjudged. On 13 July 2025, a strike near a water distribution point in Nuseirat refugee camp attributed to a “technical malfunction” killed several civilians, including children. On 23 March 2025, the killing of 15 rescue workers, including Red Crescent paramedics and a UN staff member, exposed “professional failures” and breaches of operational protocol.
These incidents, viewed alongside casualty figures of 72,045 fatalities and 171,686 injuries as of February 2026 reported by the Gaza MoH, underscore how contemporary targeting processes can magnify the risks of misidentification, malfunction, and disproportionate harm, thereby raising serious concerns for compliance with IHL.
At the forefront of IHL lies the principle of distinction, codified in treaty and customary law, which obliges parties to an armed conflict to at all times distinguish between civilians and combatants, and between civilian objects and military objectives. The reliance on algorithmically generated target lists, particularly where individuals are identified and struck within their homes as we have seen in the conflict in Gaza, raises difficult questions as to whether such systems can reliably meet this standard.
These concerns are compounded by the contested legal status of members of non-state armed groups such as Hamas and Palestinian Islamic Jihad. In conflicts not universally recognised as international in character, combatant status does not formally attach, meaning that individuals may only be lawfully targeted while directly participating in hostilities. We now have seen from this conflict that the presumption that individuals located in civilian settings such as their homes constitute lawful targets risks stretching this threshold beyond its intended limits.
Closely linked is the principle of proportionality, which prohibits attacks expected to cause incidental civilian harm excessive in relation to the concrete and direct military advantage anticipated. The apparent incorporation of pre-determined or system-generated collateral damage thresholds within AI-assisted targeting processes raises the question of whether such deeply contextual and value-laden assessments can be meaningfully reduced to algorithmic outputs.
The other principle requiring parties to a conflict to take all feasible precautions in attack including the duty to verify targets and minimise civilian harm appears difficult to reconcile with reports of accelerated targeting cycles in a conflict like that in Gaza where AI plays a central role. When decisions are made at speed, and in high volume, the risk emerges that precaution becomes procedural rather than substantive, undermining its protective purpose as intended under IHL.
A fact-finding committee convened by the International Criminal Tribunal for the Former Yugoslavia to examine NATO’s 1999 bombing campaign against the Federal Republic of Yugoslavia emphasized that the obligation to take precautions is one of feasibility rather than absoluteness. Commanders retain a degree of discretion in determining which resources to employ and how to employ them, taking into account the practical realities on the battlefield. Yet, this very flexibility underscores a core definitional problem: the precautionary duty is inherently subjective, dependent on what a reasonable commander could achieve under the circumstances.
Gaza’s experience seems to point to the fact that the use of AI further complicates compliance because the “feasibility” of supervising or constraining a system whose behaviour is not certain is difficult, if not impossible, to define with precision. Indeed, we have now seen that when human judgment in targeting and the use of force is ceded to computer-controlled processes, the very essence of life-and-death decision-making becomes mechanized, risking superfluous suffering for combatants and civilians alike but also provokes deep unease in the public conscience.
The attacks allegedly carried out of mistake in Gaza validate the fear of many scholars about the inability of AI and autonomous weapons to anticipate or estimate the consequences of an action and lack of guarantee for the systems to consistently perform as intended, without failures or unintended effects. These incidences have compounded the fact that weapons can act in ways neither foreseen nor intended by the operators, making it inherently difficult to ensure compliance with IHL.
The historical foundations of International Humanitarian Law, as articulated in the 1868 St. Petersburg Declaration, remind us that the central purpose of warfare is to diminish the military capacity of the enemy, and nothing more. Any use of weapons or methods that unnecessarily intensifies suffering or renders death inevitable exceeds this legitimate aim and stands in direct opposition to the laws of humanity. In the contemporary context of AI-enabled and autonomous weapon systems, this principle assumes renewed urgency. While technology offers unprecedented efficiency and operational advantage, it must not eclipse the human-centered imperatives of restraint, accountability, and moral judgment.
The challenge before policymakers, military planners, and legal scholars is therefore clear: to ensure that the transformative potential of AI in warfare is harnessed without undermining the enduring humanitarian principles that have guided armed conflict for over a century. Failure to do so risks not only legal violations but a profound erosion of the ethical and moral fabric that underpins the law of war.
The Author is an alumnus of the All Africa IHL Moot Competition 2024, Team Uganda.
This post was created with our nice and easy submission form. Create your post!
Written by
Did this story move you? Every gift goes directly to EJIKU Justine — writers on Muwado earn from reader appreciation, not algorithms. Even $1 makes a difference.
Muwado weekly chart
Get Africa’s top 10 stories every Thursday
No account needed — just your email.
Want to follow EJIKU Justine and get notified every time they publish?
Create a free Muwado account →