What's new

Welcome to uhufu | Welcome My Forum

Join us now to get access to all our features. Once registered and logged in, you will be able to create topics, post replies to existing threads, give reputation to your fellow members, get your own private messenger, and so, so much more. It's also quick and totally free, so what are you waiting for?

Targeting mistakes and other unintended engagements in armed conflict: The explosion at Al-Ahli Hospital in Gaza


Staff member
Mar 20, 2024
Reaction score
Armed conflicts are marred by misidentifications and miscalculations. Civilians carry the brunt of identification mistakes and faulty military equipment, often being harmed not intentionally, but due to the carelessness of parties to conflict. In the past months, reports of unintended engagements have been particularly frequent. On 5 May, it was announced that Russia has ‘accidentally bombed its own territory’. In April, Israel characterised a strike that killed World Central Kitchen aid workers as a ‘grave mistake’ that ‘followed a misidentification’. In December, the IDF mistakenly killed three Israeli hostages, and an investigation revealed ‘a string of errors and flaws’ that led to their deaths. Unintended engagements are also a main concern raised in relation to autonomous weapons systems.

What unites these military engagements is that the way in which they unfold and/or their consequences are not intended by the party to conflict, that is, they are neither desired nor foreseen as a virtual certainty. There is a tendency to frame unintended engagements as ‘accidents’, as the inevitability of ‘things going wrong’ (ch. 25) in armed conflict, which introduces a bias in the way we think and talk about them. However, as I have written elsewhere, that a particular outcome was not intended by a party to conflict does not mean that the party’s conduct leading to that outcome did not constitute a violation of international humanitarian law (‘IHL’). Equally, the fact that civilians suffer (even extensive) harm does not mean that a party to conflict violated IHL. The extent to which mistakes, malfunctions and other unintended engagements violate IHL depends on the interpretation of particular obligations from this regime, and their application to the conduct of parties to conflict.

This post interprets four obligations of relevance to the topic of unintended engagements. These are the obligation to take precautions in attack, the obligation to take precautions against the effects of attacks, the prohibition of indiscriminate attacks and the prohibition of making civilians the object of attack. The obligations will be discussed in their application to the explosion that occurred at the premises of the Al-Ahli Hospital in Gaza.

On October 17, 2023, an explosion in the parking lot of the hospital killed and injured many civilians. The facts around the explosion remain disputed. According to Israel, the explosion was caused by a misfiring rocket launched by Palestinian Islamic Jihad (‘PIJ’). According to the Ministry of Health in Gaza, the explosion resulted from an attack by the Israel Defense Forces (‘IDF’). A Human Rights Watch investigation found that the most likely origin of the explosion was a misfired rocket by PIJ, with many Palestinian organisations contesting these findings. Facts matter, and the need for a full and independent investigation is clear. What seems to transpire from the reports is that what unfolded at Al-Ahli Hospital was not intended by either party. The post will review separately alternative scenarios in which the explosion is attributed to PIJ and to the IDF, as these scenarios engage different obligations.

The post will not engage with the question of conflict classification since the relevant obligations, as reflected in Additional Protocol I (‘AP I’), also exist under customary law in non-international armed conflicts and are thus binding on both Israel and PIJ whatever the classification. The post is confined to an examination of IHL and does not engage with other legal regimes, such as international human rights law and international criminal law.


One version of events is that the explosion was caused by a failed PIJ rocket launch. Two obligations will be examined under that version: first, the obligation to take precautions against the effects of attacks, and second, the prohibition of indiscriminate attacks.

Obligations to take precautions against the effects of attacks

Parties to conflict must take precautions against the effects of attacks (art. 58 AP I; ICRC Customary IHL, Rule 22). This is a positive obligation to take feasible measures to protect civilians and civilian objects under a party’s control. Though the name by which this obligation is known refers to ‘attacks’, its application is broader. Under treaty and customary law, parties to conflict, to the maximum extent feasible, must take necessary precautions to protect the civilian population, individual civilians and civilian objects under their control against the dangers resulting from military operations. Nothing suggests that the obligation is confined to dangers arising from attacks by the opposing side. The phrase ‘dangers resulting from military operations’ can equally capture the dangers from operations by the party under whose control civilians are located. Thus, the obligation is aimed not only at protecting civilians from actions of the adversary, but also from the military operations of the party under whose control they find themselves.

In considering the content of the obligation to take precautions against the effects of attacks, the Danish Military Manual affirms that ‘the obligation is closely linked to the principles of military necessity and distinction, and it covers any danger resulting from any military operation’ (emphasis in original text, p. 194). A broad understanding of the obligation would operationalise the protection of civilians against dangers arising from military operations required by IHL (art. 51(1) AP I). Specific examples of ways to discharge this precautionary obligation can be found in the military doctrine of Croatia, Italy and Madagascar. Possible measures to restrict civilian harms include the adaptation of weapons to the environment of the defence position and the provision of information and warning of defence measures potentially endangering civilians so as to allow them to adjust their behaviour to upcoming military operations.

Importantly, this obligation is conditioned by a standard of feasibility. In other words, PIJ will only be in breach of this obligation if the group failed to take ‘feasible’ precautions. The interpretive weight falls on the notion of ‘feasibility’ and the factors relevant for its analysis. According to the United Kingdom, for example, what is feasible in particular circumstances (both for the relevant precautions in attack and against the effects of attacks) must take into account ‘humanitarian and military considerations’ (s. 5.36 and ftn 224). Although the standard of feasibility may seem fairly deferential to parties to conflict, it does not entail unlimited discretion. In its Central Front: Ethiopia’s Claim 2 Partial Award, the Eritrea-Ethiopia Claims Commission found Eritrea in breach of the obligation to take feasible precautions (in attack) for two main reasons: first, for lack of essential care in engaging targets, and second, for ‘failure to take appropriate actions afterwards to prevent future recurrence’ (paras. 110-113).

Applied to the explosion at Al-Ahli Hospital, a case can be made that PIJ failed to take feasible precautions to protect the hospital and those being treated or sheltering on its premises. According to the IDF, PIJ ‘fired a barrage of ten rockets at Israel from a cemetery near the Al-Ahli Baptist Hospital in Gaza’, and the explosion was caused by ‘errant rocket fire’. The location from which PIJ fired may be problematic – firing rockets in such proximity to the hospital creates a distinct risk of harm. Further, the weapons used may raise issues, given the high rate of launch failures. And finally, if it was indeed a failed launch, it follows a series of prior failed rocket launches that caused harm to civilians in Gaza. Failures to take appropriate protective action to mitigate the risk of future harm following such incidents could also indicate that the group fell short of its precautionary obligations.

Prohibition of indiscriminate attacks

A slightly more contentious claim would be that PIJ’s failed launch violates the prohibition of indiscriminate attacks. For this argument to be made, two cumulative conditions need to be satisfied. First, the failed launch must qualify as an attack. Second, the attack must be indiscriminate.

First condition – definition of attack

For the prohibition of indiscriminate attacks to be engaged, a particular conduct must qualify as an attack. According to art. 49 AP I, an attack is an act of violence against the adversary, whether in offence or in defence. The adversary is understood broadly to capture not only the military forces of the opposing side, but also persons and objects under its control (para. 2). Does it matter that PIJ’s rocket never reached the adversary? The question turns on whether the definition of attack contains a result element (whether actual harm in areas under the control of the adversary, or at least some concrete endangerment) or is concerned with conduct only (the initiation of an act of violence against the adversary, whatever its actual consequences).

On one view, the act would not qualify as an attack if it merely fell on territory controlled by the launching party, without at least endangering the adversary. On another view, a failed launch of this type would still constitute an attack, since what matters is that PIJ initiated a violent act against its adversary, that is, it launched a rocket designed for targets in Israel, thus ‘[setting] upon with hostile action’. This latter approach is to be preferred. First, the text of the definition does not suggest a result element. Second, during the Preparatory Commission for the International Criminal Court, a majority of delegations suggested that a result requirement was consciously left out from the prohibition of intentionally directing attacks against civilians under art. 8(2)(b)(i) Rome Statute during the Diplomatic Conference negotiations in Rome (p. 380-381). Under this view, the prohibition would be engaged even where, due to the failure of a weapons system, the intended target was not hit. As this prohibition is built on the IHL notion of attack, it suggests that a hostile operation need not reach its target to qualify as an attack. And third, the first view would tie the definition of attack to consequences which may be the result of external and unforeseen factors, while the second one closely tracks the decision-making of the party to conflict.

An argument can thus be made that, even though PIJ’s rocket fell short, it still constituted an attack, as it was an act of violence against the adversary, that is, against IDF positions and objects and persons under Israel’s control.

Second condition – content of the prohibition of indiscriminate attacks

If the rocket launch qualifies as an attack, the next question is whether this attack was indiscriminate. A type of indiscriminate attack relevant to the scenario at hand is one which employs a means of combat which cannot be directed at a specific military objective and is therefore of a nature to strike military objectives and civilians or civilian objects without distinction (art. 51(4)(b) AP I; ICRC Customary IHL, Rule 12). Here we need to distinguish two scenarios: (1) an attack which uses weapons that are indiscriminate by nature (that is, incapable of being used lawfully in any circumstances, such as ‘blind’ weapons (para. 5)) or (2) an attack using weapons that are incapable of complying with the principle of distinction in a particular battlefield environment (even if the weapons could be lawfully used in other circumstances) (para. 8).

Under (1), if it can be determined that PIJ used an inherently indiscriminate weapon, this would be enough to establish a violation. The question here is therefore a definitional one – what is an inherently indiscriminate weapon? Under (2), PIJ will be responsible for launching attacks where its means fall short of the required discrimination for a particular battlefield environment. For many years, armed groups operating from Gaza have used unguided rockets that impact Israeli population centres. These attacks are either done without care for distinguishing between civilians and military objectives in Israel, or the weapons used cannot adequately be aimed. In either case, a strong argument can be made that these attacks are indiscriminate. The harm from these rocket launches, however, due to a combination of Israeli defences and weapon performance, has also led to significant harm on the Palestinian side. As CNN reported in May 2023, ‘more than 500 rockets were fired from Gaza toward Israel in the recent barrage […] Of those, 154 were intercepted by Israeli missile defenses and 110 fell short, landing in Gaza.’

This raises an important question on how we assess whether a particular launch is indiscriminate to a particular battlefield environment. More specifically, the question is whether, in determining what would be indiscriminate, we take account of potential civilian harm under the control of the adversary only, or of all civilian harm that may be on the trajectory of the attack, including civilians under the control of the launching party. Imagine, for instance, the launch of a Qassam unguided rocket from a densely populated area controlled by the attacker towards an area controlled by the adversary where there are only lawful military objectives. Imagine also that it has a high launch failure rate. Could the potential harm to civilians under the attacker’s control make the attack indiscriminate? The harm to Palestinian civilians at the premises of Al-Ahli Hospital can only feature in the analysis of discrimination if we accept the proposition that any foreseeable civilian harm potentially implicated by the attack – from launch to impact – bears on its legality.

A significant complication under both (1) and (2) is that no set benchmarks of precision and reliability divide discriminate and indiscriminate weapons. There are similarly no clear prescriptions on the types of weapons that are indiscriminate in particular battlefield contexts. At the extreme, it is clear that weapons whose flight path cannot be managed and that can only be directed at a general area, such as modified air bombs, are either indiscriminate by nature or at the very least indiscriminate when used in densely populated urban environments. Beyond that, how one specifies tolerance levels for discrimination in weapons is less clear. Typically, the contours of the categories of weapons that are indiscriminate by nature or in particular environments are drawn by examples (Katyusha rockets, Scud missiles, V2 rockets) rather than by particular factors and thresholds. As noted by Brehm, indeterminacy in the rules and the case-by-case nature of legality assessments complicate the application of the prohibitions.

PIJ’s launch will likely be considered indiscriminate given its lack of direction at a specific military objective in Israel. That said, one has to take a wide view of the rule’s interpretation and application to be able to factor in potential harm to civilians in Gaza. We would have to accept that discrimination is judged by any potential civilian harm from launch to impact. If this view is taken, relevant factors will be the history and frequency of PIJ weapons falling short. Incidents involving munitions fired by Gaza-based armed groups falling short, killing and injuring Palestinians have been reported for years.


A second version of events is that the explosion was the result of a mistaken IDF airstrike. Two obligations will be examined under this version: first, the precautionary obligation to verify targets, and second, the prohibition of attacking civilian objects.

The obligation to take precautions in attack

IHL requires parties to conflict to take precautions in attack. Of particular relevance here is that parties to conflict must do everything feasible to verify that their targets are military objectives (art. 57(2)(a)(i) AP I; ICRC, Customary IHL, Rule 16). The goal is to form an accurate representation of a person or object’s status through the means available to the decision-maker (such as intelligence, surveillance and reconnaissance).

The obligation to verify targets is an obligation of means, not of result. It is assessed in light of the information available at the time of decision-making and is conditioned by a standard of feasibility. Many determinants have been argued to feed into the notion of feasibility: the capabilities possessed by a party to conflict, the risk to a party’s own forces, legal considerations (for instance, whether a possible precautionary measure would breach another rule of international law). Particular complications arise where the adversary employs methods of deception to mislead the decision-makers of the attacking party.

To determine whether the IDF complied with its precautionary obligation to verify targets in this scenario, one would need more information on the steps they took or could have been expected to take given their available capabilities, the battlefield environment and the information available at the time of the attack.

The prohibition of attacking civilian objects

Once a party has taken all feasible precautions to verify a target, it may still be that the party would be legally required to abstain from launching an attack against that target. This transitions the analysis to negative obligations, which, as in the PIJ scenario, pose more interpretative hurdles than the relevant positive ones.

According to treaty and customary IHL, civilian objects shall not be the object of attack or of reprisals (art. 52(1) AP I; ICRC Customary IHL, Rule 7). The key question here is: what does it mean to make something ‘the object of attack’?

This phrase can be interpreted in a number of ways. It can be seen as containing some type of fault element and may refer to the following possibilities of fault – attacks: (i) where the attacker wants to target a civilian object because of its status; (ii) where the attacker knows or believes that the object is civilian; (iii) where the attacker is reckless regarding the status of its designated object; or (iv) where the attacker is negligent towards its status. Alternatively, the phrase may be seen as denoting a purely objective standard: (v) attacks where the attacker, without any element of fault, as a matter of fact makes a civilian object its target (on this last option, for the prohibition of attacking civilians, see here). Where one draws the line is of particular importance to mistaken strikes. If the prohibition only captures attacks against civilian objects that are characterised by intention to strike them as civilian objects (that is, with purpose or knowledge vis-à-vis their status), then being mistaken about the status of the object would mean falling outside the scope of the prohibition. As a reminder, this is an analysis of an obligation of parties to conflict under IHL, not of individual criminal responsibility for the war crime of attacking civilians under ICL.

To identify the range of plausible interpretations of this IHL obligation, it is necessary to resort to the customary rules on the interpretation of treaties. Critical to the analysis is the interpretation of the phrase ‘object of attack’ in its context, and more specifically in its relation to the provision of grave breaches (art. 85 AP I) and the rule of doubt (art. 52(3) AP I).

Making someone or something ‘the object of attack’ is language used in a number of obligations under AP I (art. 12(1); art. 41(1); art. 42(1); art. 51(2); art. 52(1); art. 56(1); art. 85(3)). The comparison between these provisions shows that the phrase itself does not imply an element of intent vis-à-vis the status of the object. This is because the grave breaches provision elevates wilfully (1) making civilians the object of attack and (2) making a person the object of attack in the knowledge that he is hors de combat to the category of grave breaches. At the very least then, this contextual analysis seems to exclude interpretations (i) and (ii) above.

Further, under art. 52(3) AP I, ‘in case of doubt whether an object which is normally dedicated to civilian purposes, such as a place of worship, a house or other dwelling or a school, is being used to make an effective contribution to military action, it shall be presumed not to be so used.’ The ICTY, in examining the content of the IHL prohibition on making civilian objects the object of attack, interpreted the prohibition in light of the rule of doubt (Galić, Trial Chamber Judgement, para. 51). Reading the presumption of civilian status into the prohibition of attacking civilian objects would confirm the view that the prohibition covers mental states lower than intentionality towards status. If doubt is interpreted subjectively (the attacker had doubts), then the prohibition would capture recklessness. If doubt is interpreted objectively (the attacker should have had doubts in the circumstances), it would further capture negligence in determining the status of an object of attack. Following this line of argument, if the attacker launches their strike when they (1) subjectively harbour doubt or (2) are negligent in determining the absence of doubt (ie fall short of the requisite standard of confidence on status), they will violate the prohibition. A similar analysis can be made for the prohibition of attacking civilians (here, here and here).

It is important to note that doubt is not necessarily the flipside of absolute certainty, which is a standard that could hardly ever be met. It sits somewhere on a spectrum of confidence, though where exactly it sits has not been specified in any detail in the national practice of states.

Applied to the Al-Ahli Hospital explosion, the IDF would be in breach of the prohibition of attacking civilian objects if it engaged the Hospital premises in circumstances of doubt, that is, without sufficient affirmative information to indicate their status of military objective. This could be the case if they had, for example, conflicting intelligence, or intelligence they knew or ought to have known was outdated or unreliable, or simply insufficient data to confidently determine status.

Finally, it bears mentioning that, had the IDF claimed it had launched a strike against a lawful military objective (say Hamas militants on Al-Ahli premises or nearby), a key question would have been its proportionality (art. 51(5)(b) AP I; ICRC, Customary IHL, Rule 14). If the IDF was unaware that so many Palestinians had been sheltering on the premises of the Hospital, this would again lead to an inquiry of mistakes of fact. What was expected and anticipated at the time of attack lies at the heart of the proportionality analysis, rather than the actual extent of incidental civilian harm and achievement of a military advantage. It has been argued that the ‘expected to cause’ standard is one of reasonable foreseeability (p. 16), that is, ‘assessed from the commander’s perspective based on what he/she knew or should have known at the time of the attack, and not on the basis of hindsight’ (p. 13). According to the ICTY, that perspective is one of a ‘reasonably well-informed person in the circumstances of the actual perpetrator’ (para. 58). Under the reasonable foreseeability of incidental harm view, even if the strike would have been proportionate given what the attacker believed, it could still be unlawful if that belief was unreasonable in the circumstances.


Incidents such as the one at the Al-Ahli Hospital are not mere ‘accidents’ – a label that neither captures the suffering of victims nor bears a particular meaning under IHL. In most cases, such tragedies occur when a party to conflict does not care, or does not care enough, in circumstances where they are required to care.​
Top Bottom