On the right-hand side of the frame, three women wearing hijab with their backs to the camera look out on the ruins of a missile-struck building with its roof caved in and surrounded by piles of rubble.

The United States’ Missile Strike on Minab’s Shajareh Tayyebeh Elementary School: Part I

International Humanitarian Law, the Protection of Schools and Children, AI, and the Shifting U.S. Posture on Law’s Restraint in War

Editor’s Note

This article is the first installment in a two-part series.

Nine months have passed since a missile strike on the first day of the U.S.-Israel-Iran war hit the Shajareh Tayyebeh Elementary School in Minab, Iran, destroying the structure and killing at least 150 persons – schoolgirls and schoolboys, plus a handful of school personnel. All public evidence indicates that the U.S. military launched that Feb. 28, 2026, strike. Yet, to this day, the United States has neither acknowledged its own responsibility nor credibly demonstrated that some other actor was responsible. We aim, in this and a subsequent article: first, to analyze the available evidence against international humanitarian law (especially IHL’s rules on the protection of civilians, including children); second, to situate that evidence within the context of shifting U.S. positions on law’s role in restraining warfare; and third, to consider what paths to accountability (criminal or otherwise) might exist were the strike fully attributed to, or acknowledged by, the United States.

This Part I article concentrates on the first two of those points. We begin by summarizing the IHL analysis of the Minab school strike that we have just published as part of a special Contemporary Practice of the United States section in the October issue of the American Journal of International Law. We then address information in investigative reporting published after our analysis was in final proof. Ultimately, we assess that the strike violated fundamental IHL rules. 

In our Part II article, we consider questions of accountability: are there “reasonable grounds to believe” the strike was a “war crime” – as the United Nations’ Independent International Fact-Finding Mission on the Islamic Republic of Iran determined after our essay went to press? And besides adjudication, what forms might accountability take? 

IHL, U.S. Interpretations, and the Minab School Strike

In assessing the strike pursuant to IHL, our AJIL essay emphasized U.S. interpretations of that body of law. The United States has assumed the obligation to comply with IHL through its ratification of the four Geneva Conventions of 1949 and its acceptance of many customary IHL norms. But compared with many other states and commentators, the United States’ interpretations of IHL often have been more permissive, asserting greater leeway in the waging of war. Notwithstanding our concerns regarding such permissiveness, we found that, even when measured against the United States’ own positions:

  • There is a strong case that the strike violated IHL’s principle of distinction, as operationalized through the rule on doubt; 
  • Available evidence indicates clearly that the United States violated its IHL obligations to take feasible precautions before launching a strike; and
  • These violations occurred on the back of disturbing U.S. institutional and rhetorical shifts away from avoiding and mitigating civilian harm, including a specific departure from longstanding support of U.N. initiatives to minimize attacks on schools and other wartime harms against children.

We conclude that if the attack on the Shajareh Tayyebeh school does not catalyze a political and institutional reversal of current trends, it is destined to be a devastating marker of the intensifying U.S. rejection of legal restraint in war.

Doubt and the Principle of Distinction: One of IHL’s “cardinal principles” (ICJ ¶78) is that of distinction: at all times, belligerents must distinguish between civilian persons, the civilian population, and civilian objects, on the one hand, and combatants, persons directly participating in hostilities, and military objects, on the other. Military operations, including attacks, must be directed only at the latter, never at the former. Under the view that prevails globally, if there is uncertainty – doubt – regarding the status of a person or object, that person or object is to be presumed civilian (see Arts. 50(1), 52(3), Additional Protocol I).

The United States’ Department of Defense Law of War Manual (2023) accepts this broad outline, yet advances a permissive interpretation of the rule on doubt. After stating that “commanders and other decisionmakers must presume that persons or objects are protected from being made the object of attack unless the information available at the time indicates that the persons or objects are military objectives,” the Manual rejects an evidentiary threshold and specifies that decisionmakers may authorize or engage in a strike if they have an “honest and genuine belief” that the target is a military objective (§ 5.4.3.2).

Even on this commander-centric and epistemically permissive approach, we assessed that a strong case can be made that the Shajareh Tayyebeh strike violated the rule of doubt, based on available reporting. Crucial in this respect are reports that, within the Department of Defense, it was “widely known” that intelligence pursuant to which the structure was targeted was severely out of date, and that, when authorizing the strike, commanders bypassed explicit warnings to that effect within the relevant database.

Precautionary Obligations and Duty of Feasible Action to Vet Targets: Whether or not the strike violated the principle of distinction, precautionary requirements remain applicable. The United States in its Law of War Manual accepts the IHL obligations to take “feasible precautions” for the protection of civilians and civilian objects” and to take “feasible” action to verify targets (§§ 5.2.3; 5.5.3). However, its approach to these rules is permissive, given that the prevailing view would require a belligerent to do “everything” or “all” that is feasible (§§ 5.2.3.1; 5.2.3.2; 5.5.3). Relatedly, the United States’ recognition of the widely affirmed duty to take “constant care … to spare the civilian population, civilians, and civilian objects” appears to be conditioned on that duty not exceeding the Manual’s interpretation of the obligation to take feasible precautions for the protection of civilians and civilian objects (§ 5.2.3.5).

Even on the U.S. approach to precautions, the available facts are damning. The attack was a scheduled, deliberate strike, in the first hours of a war of choice. A month of naval mobilization had preceded the war. During that period, prior to the pressure of ongoing hostilities, the feasibility of verifying that a potential target was not civilian was at its apex (compare Law of War Manual, §§ 5.2.3.2; 5.5.3). 

Had such a verification process taken place, it would almost certainly have revealed the target to be a school. For almost a decade, an outdoor play area and distinctive pastel-colored painting on the structure were visible on Google Earth. Images of the courtyard included numerous persons whose shadow-length, given the time of day, revealed them to be children. The school appeared in Iranian business listings, and its years-long online presence included photographs matching the colored walls visible from satellite imagery, as well as videos of a thriving school. Instead, we assessed, the United States was “laser-focused on maximizing the tempo of operations” (AJIL n.80).

Considering this evidence as a whole, it is clear that the United States violated its general obligation to take constant care to spare the civilian population, as well as the specific requirement that it take feasible precautions to verify targets. 

Departures from Restraints on Harms to Civilians and Civilian Objects, Including Children and Schools: From 1999 onwards, the United States joined more than a dozen United Nations Security Council resolutions which identified certain “grave violations” against children amid armed conflict – among them, attacks on schools – and urged parties to end such practices. The United States endorsed the Safe Schools Declaration, asserting: “We apply safeguards, additional to those required by international humanitarian law, when conducting military operations that may affect schools.” It cited in particular the “Department of Defense … action plan to improve its ability to mitigate and respond to civilian harm.” That endorsement occurred on Jan. 13, 2025, a week before President Donald J. Trump’s inauguration. By the time of the missile strike on the Shajareh Tayyebeh school thirteen months later, the United States had departed significantly from its support of such initiatives (AJIL n.101).

As detailed in our essay, a rhetoric of “lethality” emerged early in the new administration, exemplified by the mantra of Secretary of Defense Pete Hegseth, “[m]aximum lethality, not tepid legality.” Institutional changes followed suit. Hegseth fired top military lawyers, the president renamed the agency the “Department of War,” and the United States dismantled and marginalized the internal systems of civilian harm mitigation that its Safe Schools Declaration endorsement had highlighted. It did so without replacement or alternative, even though there was no budgetary or other imperative for that change (AJIL n.89). 

Reporting Subsequent to Our Initial Analysis

Our AJIL essay relied on government officials’ statements and on investigative reporting by multiple news outlets, including Airwars & Pax, BBC News (here and here), CBC, CNN (here and here), the Financial Times (here and here), The New York Times (here, here, here, and here), NPR, ProPublica (here and here), Sky News, and The Washington Post (here and here). In the weeks since that analysis went to final proof, investigative reporting on the missile strike continued. Of particular note is “‘Inside the Kill Chain’ That Destroyed an Iranian School” by Bloomberg reporters Ben Bartenstein and Krishna Karra. This reportage not only corroborated factual evidence upon which we had relied, but also surfaced new information warranting additional reflection.

Feasibility of Vetting and Dismantlement of Civilian Harm Mitigation Action Plan: Our conclusion that the United States violated its precautionary obligations, including its duty to take feasible action to verify that a potential target is in fact a military objective, is bolstered by the Bartenstein and Karra report. Regarding verification, they wrote: “Work that had unfolded in previous conflicts over weeks or months – matching sites from the library, such as buildings, airfields or bases – was completed in a matter of days.” The challenges of hasty verification were compounded by the exclusion of civilian harm mitigation personnel from the process of analyzing and selecting targets – “a consequence of the cuts in staffing” during the prior year, they wrote, “but also a decision not to involve the group in planning and operations.”

Artificial Intelligence: A footnote in our initial analysis observed that the Minab school “strike may also invite questions about the possible use of artificial intelligence (AI) in target-identification,” but continued that “limited details on this aspect at this stage preclude detailed commentary here.” (n.73) The Bartenstein and Karra report elaborated on the role of AI, allowing further comment.

Their report discussed the commander’s role in setting “guidance for selecting and prioritizing a menu of potential targets.” These targets then were developed using the Defense Department’s Maven Smart System (“made by Palantir Technologies Inc.”) in order to: link “disparate information to places, objects, and dates”; display “imagery, analysts’ previous detections, reports, history, warnings and other material”; and apply “mission-specific objectives to enable prioritization of targets.” 

According to Bartenstein and Karra, the U.S. Central Command process entailed an assessment of the lawfulness of the target, of compliance with precautions, and of the sufficiency of intelligence; when asked whether striking the Shajareh Tayyebeh target would be legal, AI’s “answer given to the senior commander was yes.” Bartenstein and Karra reported: “Some Centcom personnel expected Maven to flag stale information or inconsistencies in the underlying intelligence assembled for potential targets,” emphasizing that it was “not clear why they had such expectations.”

This account of AI’s role in the strike, which is more detailed than any information previously available, reinforces our analysis. It is wholly consistent, for example, with our finding that the United States violated its precautionary obligations under IHL. (For more on the specific relevance of AI integration in that respect, see Jessica Dorsey’s excellent post at Opinio Juris.)

Implications for Analysis under the Rule on Doubt: The Bartenstein and Karra account is potentially more significant for analysis of the rule on doubt undergirding IHL’s foundational principle that belligerents must distinguish between civilian objects and military objectives, and may only direct attacks at the latter. One way of construing their account is that the AI interface could have led to a scenario in which no individual commander or decisionmaker appreciated the level of doubt that was plainly manifest in the system as a whole. To be clear, this is not the only way to read Bartenstein and Karra. Nor is it clear that this is in fact what happened, particularly given an earlier report in which CNN’s Zachary Cohen indicated that commanders had bypassed warnings that were visible to them. Nevertheless, the possibility that no individual commander appreciated what the system did warrants consideration.

If in fact the doubt which was manifest in the U.S. system as a whole was not concentrated in the mind of any individual decisionmaker, then it may be that the attack would not have violated the rule on doubt, as advanced by the United States, given that U.S. interpretations emphasize a commander-centric analysis and a standard of good faith. But that does not end the matter. We pointed out in our AJIL essay (AJIL n.72) that such a scenario “would raise an interpretive question regarding the necessary locus of doubt in applying the rule.”

In this article, we underscore what one of us (Dannenbaum) argued in a recent Just Security podcast discussion (from 38:50): The United States’ commander-centric approach to the rule on doubt is inapt, for the reason that IHL rules are meant to engage the responsibility of the state, not just individual humans.

IHL does more than guide the decision-making of individual commanders. It also guides state action ex ante (in the interests of preventing and mitigating unwarranted harm), and further provides the basis for state responsibility ex post (in the interests of accountability and reparation when such harm occurs). Thus, in guiding state action, IHL should inform not only the conduct of the final authorizing commander, but that of everyone in the state’s system, including those persons involved in structuring the internal flow of information; in this instance, everyone engaged in designing, incorporating, and relying upon systems of artificial intelligence. Such a whole-of-state approach is necessary to ensure that doubts are communicated to targeting analysts and to authorizing commanders, and then integrated effectively by them – necessary, that is, to preserve IHL’s presumption of civilian status. In the situation at hand, given the significant doubt manifest in the U.S. military system as a whole, a formidable case can be made (even assuming the authorizing commander was unaware that intelligence was stale) that the United States violated IHL’s principle of distinction, as operationalized through the rule on doubt.

In sum, based on the totality of publicly available information, a powerful case can be made that the United States violated multiple rules of IHL, including the principles of distinction and feasible precautions. In Part II, we will examine what this entails for accountability at the levels of both state and individual responsibility.

Filed Under

, , , , , , , , , ,
Send A Letter To The Editor

DON'T MISS A THING. Stay up to date with Just Security curated newsletters: