The Application of Artificial Intelligence in Military Operations A Warning Bell for International Humanitarian Law
n 9 Esfand 1404 (March 1 2026) the United States and Israel launched joint aggressive operations against Iran. This military operation was rapidly labeled as the first AI-based war. Reports indicate that the U.S. Army used Anthropic’s Claude tool to strike approximately 1000 targets in the first 24 hours. This AI system significantly assisted in war planning by optimizing target selection analyzing intelligence data and determining precise coordinates through satellite imagery. The use of the Claude system is part of the Pentagon’s Maven Smart System project.
Introduction
On 9 Esfand 1404 (March 1, 2026), the United States and Israel launched joint aggressive operations against Iran. This military operation was rapidly labeled as the "first AI-based war." Reports indicate that the U.S. Army used Anthropic’s "Claude" tool to strike approximately 1,000 targets in the first 24 hours. This AI system significantly assisted in war planning by optimizing target selection, analyzing intelligence data, and determining precise coordinates through satellite imagery. The use of the Claude system is part of the Pentagon’s "Maven Smart System" project.
The Maven system, developed by Palantir Technologies and rooted in the Maven project of the 2010s, performs three primary functions: 1) Target identification, 2) Matching munitions with identified targets, and 3) Assessing battle damage. By integrating 179 data sources, including satellite imagery, surveillance systems, and classified intelligence, the Maven system identifies threats and targets. Utilizing classified data from satellites, surveillance systems, and other intelligence sources, the system creates conditions for real-time targeting options in warfare.
The increased use of artificial intelligence in weapon systems shortens the "kill chain" (i.e., reducing the time between target identification and engagement). This leads to the "compression of the decision-making process," in which human actors, rather than exercising independent judgment and decision-making, become heavily reliant on algorithmic recommendations. Given the absence of any binding regulations regarding the responsible use of military AI, the risks associated with the application of AI in weapon systems are increasingly escalating.
The Maven system has reduced the duration of the "kill chain" to under 72 seconds (and in some cases, 20 seconds), minimizing human oversight (requiring only a single click for final approval). Furthermore, the density of civilian centers has driven the system to approve strikes even with low confidence levels, significantly increasing civilian casualties.
Consequences of AI Application in Military Dimensions
Although artificial intelligence is deployed as a tool to enhance operational accuracy and efficiency, its military use has raised serious concerns regarding accountability and the protection of civilians. Target identification systems are only as reliable as the data they process. These concerns regarding humanitarian law intensified when one of these attacks on the first day of the war struck the "Shajarat al-Tayyiba" elementary school in Minab (Hormozgan Province). This attack, carried out using Tomahawk missiles, resulted in the martyrdom of 156 civilians, more than 120 of whom were children aged between 7 and 12.
In a statement to the UN Security Council (September 2025), the International Committee of the Red Cross (ICRC) warned against three significant applications of AI in the military domain: 1) Autonomous Weapon Systems (AWS), 2) Decision-support systems, and 3) AI-based cyber systems.
Regarding decision-support systems, the ICRC acknowledges that "high speed and scale, combined with automation bias, may lead to blind endorsement by the human operator, replacing human judgment." This concern is clearly evident in the operation of Israel’s "Lavender" system (approving targets in 20 seconds without genuine human review) and the US "Maven" system (unverifiable information speed). Furthermore, the ICRC recalls its historical experience that "claims of new tools being more precise and humane" have always been made by developers of new weapons, yet "the outcome on the battlefield has never significantly improved the situation of civilians; rather, it has led to accelerated and widespread destruction with horrific consequences."
International Humanitarian Law (IHL) requires military operations to adhere to the principles of distinction, proportionality, and precaution. Commanders must distinguish between civilians and combatants and ensure that military advantage outweighs potential harm to civilians. When AI systems are deeply embedded in the targeting cycle, determining whether these humanitarian obligations are met becomes extremely difficult. If an attack based on algorithmic recommendations results in civilian casualties, attributing responsibility becomes complex, leaving the chain of accountability between commanders, programmers, private technology firms, and machine learning systems ambiguous.
Another major concern in this domain lies in the opacity of the AI decision-making process. Many AI models operate as "black boxes"—even their creators cannot fully explain how the system arrived at a specific output. When such systems are used to generate targeting recommendations, military personnel may struggle to reject or confirm algorithmic results. This increases the risk that human oversight becomes merely ceremonial. In conditions of highly complex decision-making, military commanders may simply approve the list of targets generated by the algorithm, rather than conducting independent and thorough human verification.
However, governments have yet to take concrete steps to mitigate the risks associated with the military use of AI. The rapid integration of AI into actual military operations demonstrates that regulatory efforts struggle to keep pace with technological adoption. The Pentagon’s Maven project exemplifies how military strategists view AI as essential for maintaining operational superiority.
In the absence of binding regulations, major powers continue to rapidly integrate AI into military operations. The war in Iran provides an early glimpse of future conflicts, where AI is no longer limited to logistical or intelligence support but also shapes targeting decisions and operational planning. As military competition among major powers intensifies, the inclination to adopt AI-based capabilities will grow. Governments fear that accepting restrictions on AI use would place them in a weaker strategic position relative to their rivals.
Nevertheless, technological superiority should not come at the cost of ethical responsibility and legal accountability. Binding legal mechanisms must be developed to confirm that governments and individuals—not machines and algorithms—bear the legal and ethical responsibility for attacks carried out. This could help prevent accountability gaps in the military use of AI. Furthermore, governments must strive to establish clear standards regarding "transparency," "accountability," and "human control" for the use of AI in warfare. This includes ensuring genuine human participation in decisions regarding the use of lethal force, establishing control mechanisms for military AI systems, and strengthening legal frameworks governing emerging military technologies. Unfortunately, leading governments in the development and use of military AI (including the United States and Israel) have refused to negotiate the necessary binding rules. The United States has explicitly opposed any "international legal regulation of military AI" and has even sidelined its own legal advisors who advocated for the enforcement of humanitarian rules. This refusal to regulate, coupled with the breathtaking speed of technological development, heralds a future where algorithmic "black boxes" hold the power to decide the lives and deaths of millions, while judges at the International Criminal Court remain helpless in even defining "criminal intent" within such processes.
Conclusion
The attack on the Minab school on the first day of the war against Iran marks a tragic turning point in the history of military technologies. This incident clearly demonstrated that artificial intelligence, at least under current conditions where it lacks effective human control and oversight, facilitates serious violations of fundamental rules of humanitarian law (particularly the principle of distinction and the principle of precaution). Therefore, governments must immediately commence negotiations for a binding international instrument that: a) restricts the use of AI systems in targeting, b) bans autonomous lethal weapon systems (LAWS) without effective human control, and c) mandates transparency and continuous accountability mechanisms.
Any use of AI-based targeting systems must require "sufficient time and opportunity" for human review and approval. Approval processes as rapid as the 20-second validations used by systems like Maven and Lavender, or the blind reliance on AI outputs, lack the necessary human competence and must be prohibited. Governments utilizing military AI must document records of all stages—including data collection, algorithm training, recommended outputs, and reviewed actions—and make them accessible to international monitoring mechanisms. Without transparency, any attempt to "audit" military AI is impossible.
Furthermore, companies such as Google, Amazon, Microsoft, and Palantir, which cooperate with militaries without restriction, must establish clear internal standards to prevent use in illegal targeting and publicly release their independent findings.
The Minab incident has proven that the risks experts in human rights and humanitarian law have warned about for years are not "conspiracy theories," but undeniable realities. As the ICRC explicitly stated in its address to the Security Council: "We cannot (and should not) allow these systems to continue developing and being used without supervision and regulation. No new technology has ever protected civilians better than existing rules. The same holds true today."
Ehsan Mohammadi, Expert at the Center for Political and International Studies
(The opinions expressed are those of the authors and do not purport to reflect the opinions or views of the IPIS)