New Challenges and Old Problems for International Humanitarian Law

This post provides some reflections based upon the experience of writing the second edition of the book International Humanitarian Law: Rules, Controversies, and Solutions to Problems Arising in Warfare.

What is New

The second edition of my international humanitarian law (IHL) textbook was written less than five years after the first edition, while the international armed conflict (IAC) between Russia and Ukraine, which monopolized most of the attention of Western public opinion and governments, was ongoing. It is not yet clear whether this conflict implies a fundamental change for IHL or whether it simply constitutes an outlier in the fundamental trend towards non-international armed conflicts (NIACs). In many respects, the conflict also prolongs past trends. In the second edition, I refer to problems in Ukraine much more frequently than any other armed conflict, which deserve as much attention and indeed, often have had even greater human cost. This is because the Ukraine conflict has shaped Western perception of armed conflicts and IHL more than any other armed conflict since the Second World War. The price to be paid is the risk that (hopefully) parts of the book will be outdated in a few years.

While the book was already in production, another major armed conflict in and around the Gaza Strip broke out, which attracted even greater attention in Western countries and the Arab world. It may increase or counterbalance certain tendencies noticed in Ukraine. Today, I would certainly have gone into more detail about the possibility of hostilities being conducted in occupied territory. And I would have emphasized my understanding of occupation and of the applicability of the law of military occupation during an invasion phase.

The focus has therefore shifted further towards IACs, which are in any case the starting point for understanding IHL. At the same time, I only marginally shortened parts dealing with NIACs and the “fight against terrorism.” I remain convinced that those, particularly conflicts involving armed groups, remain the main conceptual challenge facing IHL.

I added genuinely new text partly because of recent developments, partly also because age-old problems were not adequately covered in the first edition. These include IHL in outer space, persons with disabilities, sieges and humanitarian corridors, investigations, the concept of meaningful human control over lethal autonomous weapons systems, and the role of the media. The subchapter on the nexus between an armed conflict and certain conduct, necessary for IHL to apply, has been profoundly refined based upon arguments made in a PhD thesis on the subject, which I supervised and that will soon be published. Finally, due to current developments and arguments made in scholarly writings, the subchapters on neutrality and the law of naval warfare have been considerably revised.

In this post I address two additions to the text, one of which has been strongly influenced by what others have written since the first edition was published: sieges; and meaningful human control over lethal autonomous weapon systems. The question of whether the “rights” foreseen in the law of naval warfare, including towards neutral ships, can still exist for an aggressor, will be discussed in a later post.

Sieges

In both Ukraine (e.g., Mariupol) and Gaza, sieges have been the focus of much international attention and humanitarian concerns. We were already confronted with the revival of this medieval practice in Bosnia and Herzegovina at the end of the last century and in Syria during the last 10 years (e.g., in Aleppo). It therefore seemed important to me to add a new section on sieges, presenting the possible specific IHL rules on the subject, but mainly applying the general rules on the conduct of hostilities, humanitarian assistance, and treatment of persons in the power of a party to the specificities of sieges. This has allowed me to deconstruct claims that, though certainly desirable from a humanitarian point of view, do not correspond to the existing law and therefore weaken IHL by leaving the public with the erroneous impression that IHL is always violated. Second, it allowed me to cautiously explore possible interpretations of the existing IHL rules, solving some, but not all, the humanitarian problems that appear in actual armed conflicts and to pinpoint where the existing law is simply inadequate.

Specific Rules on Sieges, Arguably Prohibiting Them

Besieging enemy armed forces not willing to surrender in an inhabited town heavily affects the civilian population. Such sieges are often accompanied by bombardment and fighting between besieging and besieged forces, creating a constant danger for the civilians trapped in the besieged area. Those civilians are often cut off from water, electricity supplies, heating, food, and health services.

Under IHL, there is no prohibition against besieging an area where there are only enemy forces or against blocking their reinforcement or resupply for the purpose of seeking to achieve their surrender through starvation. A problem arises, however if (as nearly always) civilians remain present in the besieged area. Intentionally using starvation of civilians as a method of warfare violates IHL (Additional Protocol (AP) I, art. 54(1)); International Committee of the Red Cross (ICRC) Customary IHL Study, rule 53). Several authors (for example, Sean Watts and Beth Van Schaack) consider that this prohibition makes sieges virtually impossible today. Indeed, one may argue that the intent to starve the civilian population is demonstrated if the besieger allows no evacuation and does not let humanitarian assistance in (although the crux in the latter case is that it may refuse such supplies if there is no control over the distribution).

Furthermore, some argue that even the encirclement aspect of a siege (not just each act of violence against targets in the besieged town) constitutes an attack, because the end goal would be to starve at least the combatants to death, which is an act of violence. If encirclement as such constitutes an attack, the proportionality rule (AP I, art. 51(5)(b); ICRC Customary IHL Study, rule 14) would apply in relation to the expected effects on the lives and health of civilians (see here, para. 748). In this case, feasible precautions would also have to be taken to avoid such effects on civilians.

I would object to the idea that a siege always constitutes an attack, because the aim of the besieger may also be seen as not aiming at the death of combatants but at obtaining a surrender of the besieged forces, which is not an act of violence that falls under the definition of an attack. Nevertheless, I would agree with the U.S. Department of Defense (DoD) Law of War Manual (§5.20.2) that the proportionality principle, as a general principle, also applies to the prohibition of starvation as a method of warfare against civilians, if it is impossible to starve combatants without simultaneously starving civilians, which is often the case in a siege. However, the effects on civilians are difficult to evaluate because they only realize themselves over time, are difficult to anticipate, and also depend on the measures taken by the besieged.

As for precautions, constant care to spare the civilian population must be taken in case of an attack, but also, and more broadly, in military operations (AP I, art. 57(1)), which clearly include the encirclement aspect of a siege. Such care may be considered to include an obligation to continuously monitor the effects of the siege on civilians (Nijs, p. 701-04), in order to become aware if or when the effects on them become excessive.

A passive precaution against the effects of attacks specific to sieges is prescribed by Article 27(2) of the Hague Regulations: “It is the duty of the besieged to indicate the presence of buildings or places [dedicated to religion, art, science, or charitable purposes, historic monuments, hospitals, and places where the sick and wounded are collected] by distinctive and visible signs, which shall be notified to the enemy beforehand.” At least for buildings other than hospitals, medical units and cultural property this goes clearly beyond the general rules of IHL. In a small and congested besieged area, only a few other passive precautions are feasible (Sean Watts, p. 13).

Furthermore, the International Court of Justice (ICJ) in the Nuclear Weapons Advisory Opinion (para. 95) has affirmed that “methods and means of warfare, which would preclude any distinction between civilian and military targets . . . are prohibited,” without limiting this prohibition to attacks, and sieges may be considered as an indiscriminate method of warfare. A difficulty for invoking the right to food of the civilian population, which would otherwise render most IHL related controversies moot, is that it is difficult but not impossible (see here, paras. 41-47) to argue that the civilian population is under the jurisdiction of the besieger.

One of the rare IHL treaty provisions explicitly mentioning sieges (Hague Regulations, art. 27(1)) prescribes that “[i]n sieges and bombardments all necessary steps must be taken to spare, as far as possible, buildings dedicated to religion, art, science, or charitable purposes, historic monuments, hospitals, and places where the sick and wounded are collected, provided they are not being used at the time for military purposes.” Due to the place where this rule appears, it can be argued that it applies to more than simply attacks. However, as it only protects objects and not persons, it is not clear how such an object could be affected otherwise than by an attack. In that case, the rule only emphasises the general rule of distinction. Nevertheless, one may object that in order to “spare” such buildings, in particular hospitals, it is not sufficient not to destroy them; the encirclement aspect of a siege must permit their functioning.

Application of General IHL Rules to Siege Warfare

General IHL rules reconcile, to a certain extent, the lawfulness of besieging enemy armed forces with the need to respect and protect civilians in three ways.

First, if the besieger respects the rules on distinction, proportionality, and precautions, and the defender does not use civilians as human shields and keeps military targets and civilians separate, civilians should be able to stay in their homes and not be obliged to flee. The fundamental approach of IHL is not to evacuate civilians so that belligerents can conduct hostilities with minimal restraint, but rather to oblige the military to conduct hostilities taking the presence of civilians into account. However, it is particularly difficult for the besieger to identify military objectives in a besieged town and to target them individually. Furthermore, at the shifting margins of the besieged area, every house becomes a military objective because the besieged forces must, as a matter of course, defend them, if they want to prevent the besieger from taking control of the besieged area. The specific rules governing attacks against objects indispensable for the survival of the civilian population apply (AP I, art. 54; ICRC Customary IHL Study, rule 54). However, this does not prohibit the besieger from hindering the resupply of such objects.

Any proportionality calculation must consider the particular impact of reverberating effects in a besieged area, the difficulties of moving within a besieged area, e.g., to a hospital or an alternative shelter, and the increased difficulty of treating injuries. Among the precautionary measures in attack, Article 26 of the Hague Regulations specifically mentions that the besieger “must, before commencing a bombardment, except in cases of assault, do all in his power to warn the authorities.”

Whether civilians were offered the opportunity to leave, or are even used by the besieged as human shields, does not modify the attacker’s obligations in terms of distinction, proportionality and precautions (AP I, art. 51(7)).

Second, to the extent prescribed by the general rules of IHL and subject to many conditions, the besieger must allow humanitarian assistance for the civilian population in need to enter the besieged area. This interpretation, rejected by some (for example, Sean Watts, p. 47-48), may be deduced from the prohibition on using starvation as a method of warfare against civilians. If the necessary guarantees that the assistance will only benefit civilians are in place, then a refusal is necessarily aimed at starving civilians. The prohibition against intentionally starving civilians is also, in my view, necessarily subject to the proportionality principle if both combatants and civilians are starved. In addition, a denial of consent to the delivery of humanitarian assistance to civilians may often be considered as arbitrary and therefore prohibited (Oxford Guidance, paras. 48-54). However, I do not think that a besieged area may be considered as occupied by the besieger, in which case the latter would have a clear obligation to consent to the delivery of humanitarian assistance (Geneva Convention (GC) IV, art. 59).

The crux of the matter is that the besieger may make its consent conditional upon local supervision of the distribution of the relief in the besieged area (GC IV, art. 23(3); AP I, art. 70(3)(b)). This is nearly always unrealistic, and it is even more so in the case of a siege, because it is more difficult to expect no diversion to combatants and fighters in a situation as extreme as a siege.

Third, the plight of civilians affected by a siege can be avoided by allowing them to leave the besieged area. It is argued (Sean Watts, p. 16) that the prohibition of starvation as a method of warfare used against civilians implies for both the besieger and the besieged an obligation to evacuate civilians who cannot be adequately provided with objects indispensable for their survival. Nevertheless, such evacuation meets several obstacles. True, IHL provisions (GC I, art. 15; GC II, art. 18; GC IV, art. 17) specifically foresee that parties “may” conclude arrangements for the removal of wounded and sick combatants from besieged and encircled areas, and that they even “shall endeavour” to conclude such arrangements for the benefit of civilians, not only to remove those who are wounded and sick but also the “infirm, and aged persons, children and maternity cases.” I do not think that the absence of such an agreement can be used to infer an absence of an obligation to let civilians leave. At least a besieger who does not let humanitarian assistance into the besieged area because the distribution is not controlled is, in my view, obliged to consent to such an arrangement for the benefit of all civilians.

However, because of the weapons used in siege operations the corridor through which civilians are allowed to leave must be large, as it may be shelled from afar. Humanitarian organizations involved in conducting evacuations out of the besieged area will not proceed if shells fall even in the vicinity of the corridor. In addition, there must also be a minimum level of trust between the parties that one party will not take advantage of the cease-fire for military purposes (i.e., the besieger to use it as a breach permitting to enter the besieged area or the besieged forces to escape from the besieged area).

Finally, humanitarian organizations involved in evacuations in such a coercive environment may fear that they thus contribute to unlawful deportations and participate in the war crime of forced displacement (see UN Commission of Inquiry Syria Report, para. 93) even if they only evacuate those who so wish. Such forcible transfer is prohibited (GC IV, art. 49; AP II, art. 17(1); ICRC Customary IHL Study, rule 129) in or from occupied territory and in NIACs, while for IACs IHL contains no rule on such practices by a party on its own territory. Even when forcible transfer is prohibited, we should remember that agreements to evacuate civilians from besieged areas are encouraged by IHL (GC IV, art. 17). The conclusion of such agreements, their implementation, and the involvement of humanitarian organizations should therefore normally not be qualified as implying forced displacement.

The besieged forces obviously have no military interest in seeing all civilians leave. This would greatly facilitate the capture of the besieged place, because everyone who remains there would be a legitimate target of attacks. Nevertheless, the defender has an obligation to remove, to the maximum extent feasible, civilians from military objectives as a passive precaution against the effects of attacks (AP I, art. 58(a); ICRC Customary IHL Study, rule 24). Preventing civilians from leaving with the intent to shield combatants and military objectives would constitute a clear violation of IHL (AP I, art. 51(7); ICRC Customary IHL Study, rule 97). As for the besieger, while in theory it has an interest in civilians leaving, in practice it will not be willing to agree to the modalities required to realise this. The continued presence of civilians puts additional pressure on the resources of the besieged. However, contrary to what was accepted in the past, a besieger may no longer attack civilians in order to prevent them from leaving the besieged area (U.S. DoD Law of War Manual, §5.19.4.1).

Whoever wants to evacuate civilians from a besieged area must organize food, shelters, health care, and education for the civilians thus displaced, whereas they are more easily, autonomously, and resiliently able to organize their lives in the place where they live. Both the besieger and the besieged may finally fear that civilians leaving a besieged area may interfere, deliberately or unconsciously, with their military operations. The besieger may also fear that they are not really civilians.

The Concept of Meaningful Human Control Required over Lethal Autonomous Weapon Systems

In the discussions among States (see here, here, and here) and experts about lethal autonomous weapon systems (LAWS), there is today a large consensus that those systems may only be used under meaningful human control (although the exact wording still varies). One may obviously fear that this is a façade consensus. Humans could become an alibi if they must merely confirm or interrupt computerized and artificial intelligence driven “suggestions” or “decisions.” Indeed, those “suggestions” or “decisions” would be based on data processed by machines thousands of times more rapidly than any human could process them. The human would in addition be influenced by what has been called “automation bias.”

Studies have indeed shown that humans are overly confident about the reliability of computer systems and the accuracy of their outputs, that they tend to accept such recommendations without seeking any disconfirming evidence. The absence or presence of human intervention is also a relative distinction as is the distinction between humans “in,” “on,” or “out of the loop”, all of which are therefore considered by some as being not very helpful. All this is, however, not a reason to give up attempts to define “meaningful human control” but rather to do so to avoid those traps. Here as elsewhere, the devil is in the detail, and I doubt whether States will soon accept a meaningful operational definition.

In my view, based on the suggestions of Jonathan Kwik (see here and here), there are different facets of meaningful human control, each of which balances the other and can compensate for weaknesses in another facet. These are ensuring sufficient awareness, exercising proper weaponeering, and implementing context controls.

Awareness includes the understanding of, first, technical aspects of the system. Obviously, the user does not have to understand how it functions, but what it is able to do and what risks the “black box” phenomenon implies. In technical discussions about artificial intelligence this phenomenon designates the fact that we do not know why particular outputs are generated. Second, the environment in which the system will operate and how it will interact with that environment must be known to a human. Moreover, a human must make a post-execution assessment to learn more about the system’s past behaviour and failures and to overcome the possible “automation bias” of the humans actually using the LAWS.

The weaponeering aspect of human control pertains to the ultimate decision of whether to use a LAWS for a mission. A human must also exert context control, by making active decisions concerning the system’s internal settings and parameters, imposing restrictions in space and time. What is most controversial is whether humans must also have the means to intervene mid-operation. If this is the case, a permanent communication link between the human and the system is needed. Moreover, the system must be able to explain to the human why it is doing or recommending something. If such mid-operation human control is a requirement, lethal functions must be deactivated in case of loss of communication. Compliance through (human) design without any human mid-operation intervention may be an alternative but presupposes stronger control under the other facets. Indeed, a purely passive “veto” power of a human diminishes their attention. If humans have a role mid-operation, their workload must neither be too high nor too low.

As Jonathan Kwik points out, implementing those facets equally solves two problems, which are widely discussed: they ensure predictability (by constraining the system’s avenues of action) and make accountability of a human possible (by making the human both cognitively involved and causally responsible for the result).

Conclusion

The first example developed in this contribution demonstrates that IHL contains rules covering age-old problems arising in warfare. Those rules must be interpreted in light of contemporary realities and other rules and principles of international law. There are still many unresolved controversies concerning the interpretation of IHL rules and their interaction with the rules of other branches of international law. Many are pursued in good faith, some in bad faith. The second example shows that new problems can be solved, pending the adoption of specific rules, by logical reasoning.

We may expect that the development of IHL rules will continue to confront dilemmas along different parameters. In international law, where there is no watertight distinction between law-making and interpretation of existing rules, those dilemmas also largely arise when interpreting IHL. First, the need to make a compromise between humanitarian aspirations and realism will continue to exist. Second, a good balance will have to be found between rules meeting new, in particular, technological, challenges and rules on the existing, traditional forms of armed violence, which still affect the greatest number of victims. Third, the tension will persist between stating timeless general principles (which will only protect if belligerents act in good faith) and detailed regulations, which will be quickly outdated and will inevitably turn out to be unrealistic in certain situations.

Whatever the perspective may be, the study of IHL is still a necessary and worthwhile endeavour. I hope that my book helps in this respect, and I am certain that the Articles of War do so. An important number of (but by far not all) IHL violations result from a lack of understanding of its rules, including two erroneous ideas:

– That IHL rules make the successful conduct of an armed conflict impossible;

– That wars could be “humane” or, that at least civilians would not be killed or injured or civilian objects would not be destroyed, if IHL were respected.

Such misunderstanding is equally an obstacle to better respect for IHL and to the development of new rules and mechanisms because States may fear that they will hinder their ability to win wars.

***

Marco Sassòli is Professor of International Law at the University of Geneva Faculty of Law.

 

 

 

Photo credit: Pexels, Elgar