A new update to an ICRC commentary on an iconic international treaty such as the Fourth Geneva Convention (GC IV), after almost 70 years is certainly a noteworthy event beyond the doctrinal interest that a project of this scope and importance evokes. It is also helpful to reflect on the function of treaty commentaries as time passes. When a commentary is written on a new treaty, its role is mainly prospective: to clarify to parties what is expected of them. After almost 80 years of State practice under the treaty, however, a commentary is naturally also reactive, responding to practice and controversies that have accumulated across the decades.

In 1958, when the previous Commentary (the Pictet Commentary) was published, the Geneva Conventions were young. The authors could rather optimistically write sentences such as “supreme rights are not, generally speaking, in any danger” as a result of administrative security measures taken by occupying powers. The new Commentary on the same provision generally avoids such language. Unfortunately, in 2026, we have decades of experience in which many GC IV articles, and their exceptions, were read in a power-legitimating way, or in a way that aimed to exclude the application of the Convention altogether.

The 1949 Conventions were meant to close significant gaps in the previous normative framework regulating conduct in war – and the Pictet Commentary was written shortly after. The function of the 2025 Commentary, as I will show in this piece, is not only to provide a positive interpretation of international humanitarian law (IHL), but also to respond to loopholes utilized by belligerents through decades of exceedingly permissive IHL interpretations. Its key responsive function is thus an attempt to narrow the gap between GC IV’s prohibitive ideals and its sometimes facilitative reality. I will accordingly focus on several issues that are not only central to contemporary IHL interpretation debates but also exemplify the new Commentary’s loophole-closing function. In this context, “loopholes” are not necessarily “legally correct” interpretations of the treaty’s text, or even plausible ones, but rather argumentative tactics that create loopholes, and in turn destabilize a protective rule.

The three examples of loophole-closing interpretations I will discuss are: first, the definition of “protected persons” in the Fourth Convention, an issue that has been central to some policies during the “global war on terror;” second, evacuations in and from occupied territories, a predominant practice during the Gaza War; and third, select issues relating to security powers under occupation, including the indeterminate definition of “collective punishment.”

Of course, I do not argue that these three exhaust all important, loophole-closing aspects of the Commentary. For example, one issue that I will not address here in detail concerns the law on humanitarian access during hostilities (Article 23 of GC IV), and the duty to provide essential needs in situations of occupation (Article 55). Concerning this framework, the Commentary addresses loopholes that do not necessarily emanate from permissive interpretations by States, but also from the self-defeating language of GC IV itself (paras. 2055-2058). This in turn raises interesting jurisprudential questions. As it stands, it could be that parties that nowadays claim to adhere to the language of GC IV, as it was drafted in 1949, would be committing war crimes – which is an intriguing legal turn of events (to say the least). Be that as it may, the question of humanitarian aid has received much recent attention in scholarship and ICJ rulings, and is also addressed in a separate piece in this symposium.

Last, I will not address the new Commentary concerning the Common Articles. These articles are identical in all four Geneva Conventions and include some of their most important provisions. However, the Common Articles were already addressed in the recent commentaries on Conventions I, II, and III. Although there are some updates in the new Commentary as well, these are generally not as consequential.

Definition of Protected Persons

One of the issues that has haunted GC IV from its inception concerns the definition of “protected persons.” This is important because while some of GC IV’s Articles refer to the protection of “civilians,” many of its provisions apply to “protected persons” specifically. GC IV’s Part III applies to the “Status and Treatment of Protected Persons” as such, whether in territories of the parties to the conflict and occupied territories (Part III.1), or in occupied territories alone (Part III.2). For example, protected persons are entitled to humane treatment (Article 27); they cannot be deported from occupied territory (Article 49); and they enjoy a range of protections when detained (Part III.4).

While the sting was arguably taken out of this distinction by the extra-territorial application of international human rights law (IRHL), or the Fundamental Guarantees regime introduced in Additional Protocol I (API), States that adopt a narrow interpretation of the relationship between IHL and IHRL, or that are not parties to API, may still advance narrow interpretations of “protected persons” in order to deprive certain individuals of protections. 

Article 4(1) of GC IV defines protected persons quite broadly. The term includes all non-nationals of a belligerent who “at a given moment and in any manner whatsoever, find themselves, in case of a conflict or occupation, in the hands of a Party to the conflict or Occupying Power of which they are not nationals.” Nationals of neutral States found in a belligerent’s own territory (i.e., not in occupied territory) are generally not protected persons (Article 4(2)).

The “invasion stage” loophole:

One key controversy about the status of protected persons concerns the status of people who are neither in the territory of a belligerent nor in occupied territories. During the early stages of its “Global War on Terror,” and later in the Department of Defense (DoD) Law of War Manual, the United States advanced an interpretation of protected persons that required that a person “be located in either (1) occupied territory or (2) the home territory of a party to the conflict” (Section 10.3.2.2. DoD Manual) (a geographic nexus requirement). This issue has also arisen in the context of the ICC’s investigation concerning treatment of Georgian detainees during Russia’s invasion in 2008.

In a 2024 Just Security piece on this issue, I elaborated on the roots of this interpretation and why it is wrong both in terms of positive law and policy. I will not repeat that here; I will only highlight the major loophole that the geographic nexus requirement allows: namely, it can be used to claim that civilians detained by armed forces during the “invasion phase” (i.e., pre-occupation) are not protected persons; that such people could be removed to third States; and that they would not enjoy protection under GC IV while in these third States.

The Pictet Commentary made passing reference to cases that fit within the geographic nexus as the “main classes” of protected persons, and did not provide much detail on this specific interpretive problem. The new Commentary clarifies that “[t]he definition of protected persons under Article 4(1) does not mention any geographical limitations” and that the decisive factor is that an individual is “in the hands of a Party to the conflict”. This, alone, is a helpful clarification. 

The new Commentary still leaves some ambiguity by stating that regardless of this broad definition, certain parts of GC IV “only apply in the territory of a Party to the conflict and/or in occupied territory.” Since these parts include some of the most crucial protections, the precise protections that apply in other situations remain somewhat unclear. That said, this problem is alleviated when read in conjunction with other parts of the Commentary that narrow another loophole by reaffirming that some of the rules of the law of occupation apply during the invasion stage (paras. 371, 1069). That view is consistent with its embrace of the “functional” approach to occupation more generally (para. 381). Crucially, this leads to the conclusion that a civilian captured during the invasion stage would enjoy similar protections to one detained under occupation.

The “unlawful combatant” loophole:

Even without insisting on a geographic nexus, a further controversial loophole invoked during the “global war on terror” concerned the status of “unlawful combatants” – meaning, persons who fight but do not fall under any of the categories that grant prisoner of war status. One argument, advanced for a time by the United States, was that since such persons were neither “civilians” nor lawful combatants they are not protected under the Geneva Conventions (paras. 1058-1059). 

The Commentary convincingly shows that even persons directly participating in hostilities (without “entitlement to do so”) retain their status as protected persons if they fulfil the nationality criteria (i.e., not belonging to the belligerent that exercises power over them). However, the Commentary could be clearer on situations in which an “unlawful combatant” actually loses their civilian status by becoming a member of an organized armed group (a doctrinal possibility the ICRC itself supports), and how this shift interacts with protected persons status. This problem frequently arises in transnational armed conflicts, where the laws of international armed conflict and non-international armed conflict may apply in parallel (for example, the ICC’s classification of the Gaza conflict).

Article 49 and Evacuations

Article 49 of GC IV prohibits forcible transfers and deportations from occupied territory. However, it provides an exception that “the Occupying Power may undertake total or partial evacuation of a given area if the security of the population or imperative military reasons so demand.” The provision further stipulates that such evacuations cannot be to places outside of the occupied territory, “except when for material reasons it is impossible to avoid such displacement;” it requires that evacuation must be temporary and tied to the original necessity; and that the occupying power ensure the evacuees have basic needs. 

The scope of this provision was central to debates during the recent Gaza war concerning the line between evacuation and forcible transfer of deportation. The key loophole here, unsurprisingly, is the danger that the security of the population, or imperative military reasons, be interpreted widely to legitimate what would otherwise be prohibited forcible transfer. In this way, the protective rules of IHL can be used to justify pressure on the civilian population, or even illegitimate political goals.

Interpreting the “imperative military reasons” exception:

While the notion of the “security of the population” is at least theoretically clear (even if not in practice), the idea of “imperative military reasons” that would justify evacuations immediately raises two questions: first, whether these reasons can be disconnected from the safety of the local population, and second, whether they can relate to the military interests of the occupying power alone, and if so, whether those military interests are to be conceived of narrowly or broadly. The Commentary seems uncomfortable with the “imperative military reasons” part of the exception, and refrains from elaborating on it beyond emphasizing that the need must arise directly from ongoing military operations and that it must be overriding and imperative (paras. 3194-3195). In this context, the Commentary only cross-references its definition of “military operations” in relation to destruction of property under Article 53 of GC IV, where the Commentary states that:

 “[T]he expression ‘military operations’ must be construed to mean the movements, manoeuvres and other action taken by the armed forces with a view to fighting. Destruction of property… cannot be justified under the terms of that article unless such destruction is absolutely necessary – i.e. materially indispensable – for the armed forces to engage in action, such as making way for them (para. 3370).”

From this analogy it can be understood that persons can be evacuated from property for “imperative military reasons,” in the very narrow circumstances that this property would also be liable to be destroyed. This would clearly exclude situations, for example, in which entire areas are evacuated to achieve the overall goal of driving out members of organized armed groups. At bottom, the Commentary seems to contemplate that the “military reasons” be connected concretely to specific operations, rather than tied in a more ambiguous way to the aims of the belligerent.

Helpful clarity on evacuations:

Beyond this issue, and recognizing the potential loopholes that inhere this provision, the Commentary provides a substantially more detailed analysis of evacuations than its 1958 predecessor. For instance, it clarifies that evacuation for the safety of the local population can relate only to the security of the evacuated population, and not the security of illegal settlers or the occupying power’s armed forces. It stresses that evacuations to an area outside of the occupied territory can only take place when the whole territory is dangerous. And it also provides a much more developed understanding of the obligation to provide essential needs in evacuation sites. Regarding temporariness, the Commentary emphasizes that the provision does not allow any discretion and that once the original necessity (“hostilities”) lapses, evacuees must be allowed to return.

Where further elaboration is needed:

Three issues in my view could have received more attention. First, it could have been helpful to discuss the relationship between evacuation orders under Article 49 of GC IV and advance warning under Article 57 of API (which provides that “effective advance warning shall be given of attacks which may affect the civilian population, unless circumstances do not permit”). For example, in Gaza, Israel treated its evacuations as warnings under Article 57, rather than orders under Article 49, because it did not concede that it was an occupying power. But if this is the case, it is unclear on what legal basis it could condition or enforce non-return to the evacuated areas (currently formalized through the “yellow line”). In a sense, the mere assertion of the power to enforce evacuation is an admittance of occupation. Furthermore, it is worthwhile to consider whether persons that flee in the “invasion stage” become evacuees “sur place” under Article 49 – with all that it entails – once their area of residence is occupied. 

Second, and relatedly, another crucial issue concerns the means of enforcement of evacuation orders. Article 57 of API clarifies that the mere fact that precautionary measures mandated by that Article were taken does not change the protected status of civilians (it does not mean, for example, that a belligerent can assume that persons who do not heed to warnings are not civilians, or that they can be somehow discounted in a subsequent proportionality analysis). The same logic should of course apply to evacuations under Article 49. To the extent that evacuations are enforced by the belligerent party, it should be clear that any enforcement in such cases is ruled by the law enforcement paradigm rather than that of the conduct of hostilities, even if hostilities are still conducted in the area.

Third, the question arises whether evacuation orders should be subject to a harm-benefit proportionality analysis. This ties into a broader question as to whether proportionality applies in situations beyond attacks (see the discussion here, pp. 142-145). Arguably, when evacuation – supposedly for their own safety – leaves people worse off than they would be by staying put, evacuation would be disproportionate. The Commentary comes close to that by noting that since evacuation “is an extreme and exceptional measure, the obligation to have recourse to it must be construed narrowly and justified by valid and compelling reasons” (para. 3196), but it stops short of adopting such a standard explicitly.

Security Powers under Occupation

Under the law of occupation, the occupying power is mandated to ensure public order in the occupied territory, which includes the power to adopt security measures on the condition that they are essential for the purposes recognized as legitimate by the law of occupation (i.e., the occupying forces’ security needs and the needs of the local population). The scope of permitted measures, whether in terms of objectives or means, has been heavily contested over the years. Unsurprisingly, occupying States often advance broad understandings of “security,” which might render the GC IV’s protective norms ineffective. Here, I want only to highlight two key aspects addressed in the Commentary in which it seeks to close key loopholes: the notion of collective punishment and the destruction of property (for a broader treatment of security powers, see here).

Security measures and collective punishment:

The general framework concerning security measures under the law of occupation is as follows: Article 27(2) of GC IV provides that occupying powers “may take such measures of control and security in regard to protected persons as may be necessary as a result of the war.” Article 78 provides the upper cap for such measures by holding that if the occupying power considers it necessary “for imperative reasons of security, to take safety measures concerning protected persons, it may, at the most, subject them to assigned residence or to internment.” 

Between the general authorization under Article 27 and the upper limit under Article 78, there is a wide range of possible security measures that are not specifically mentioned (for instance, imposition of checkpoints). Additionally, some specific prohibitions concerning the law of occupation contain “military necessity” exceptions (e.g., Article 53). In any case, GC IV subjects all possible security measures to requirements of humane treatment and non-discrimination (Article 27(1)) and to the principle of individual responsibility (which includes the prohibition on collective punishment, Article 33). Israel’s High Court of Justice has also subjected security measures to a harm-benefit proportionality test, while simultaneously adopting a broad reading of both what constitutes “legitimate security considerations” and the lawful means to achieve them.

In this context, the notion of “collective punishment” suffers from significant indeterminacy, and raises a myriad of theoretical and practical questions. For example, what constitutes “punishment”? Does it refer to measures that resemble criminal punishment, or to any sanction or inconvenience? Does it only include measures that infringe on a legally recognized right, or also those that harm lesser interests? The 1958 Commentary simply referred to “penalties of any kind” (without clarifying what “penalties” are). The 1987 Commentary to API offered a more functional definition in stating that the term “must be understood in the broadest sense: sanctions and harassment of any sort, administrative, by police action or otherwise” (para. 3055).

The new Commentary adopts the broad view suggested in the commentary to API, but adds significant elaboration on the scope and nature of the prohibition, as well as examples of actions taken over the years that would constitute such punishment (including, for example, suspension of family visits for detained persons; restrictions on movement; home demolitions, revocation of residency permits, etc.). It furthermore clarifies that for a measure to be considered prohibited collective punishment, it does not have to be unlawful per se. For example, revoking privileges such as various permits from a group – even if these are not mandated by law – could amount to collective punishment (para. 2545).

The centrality of purpose and centering human dignity: 

Nonetheless, a persistent remaining problem is how to differentiate between prohibited collective punishment and “incidental” effects of security measures that might be lawful under certain circumstances (for example, imposition of a curfew, restrictions of movement, etc.). The ease with which punishment can be recast as a preventive security measure is indeed one of the key loopholes concerning security powers. 

In our work on the law of occupation, Eyal Benvenisti and I suggested that the core distinction between collective punishment and incidental effects of security measures revolves around the measure’s purpose, and specifically, whether it is intended to make opportunistic use of group suffering for retribution or deterrence (p. 128). Consistent with this view, the new Commentary states that the purpose of the measure is indeed the “decisive criterion” para. 2547). This emphasis on purpose also leads to the conclusion that the measure does not have to in fact cause significant harm to be considered collective punishment (para. 2548). By adopting this view, the Commentary advances a strong human dignity-based rationale that it is the collective and opportunistic element that makes such measures unlawful, rather than the gravity of the harm they cause.

Of course, if intent is so central to the prohibition, the difficulty is how to determine whether a measure is intended to cause opportunistic harm. The Commentary alludes to statements of officials and the “circumstances surrounding [the measures’] imposition,” as well as whether the measure was still applied after its alleged security purpose lapses (para. 2549). I would add to these indications of intent a presumption that when a measure is disproportionate in its actual harm in relation to the stated security purpose, it is possible to infer that the measure is in fact intended to inflict collective punishment.

Protection of property:

Another context in which the Commentary addresses exceedingly permissive interpretations concerns the protection of property under occupation. Article 53 of GC IV prohibits destruction of public and private property in occupied territories “except where such destruction is rendered absolutely necessary by military operations.” For decades, Israel has argued – and its Supreme Court generally agreed – that demolishing homes belonging to militants or their families is not an act of collective punishment since it does not aim to punish but rather to “deter” (an issue addressed above) and that furthermore, acts of deterrence fall under the exception in Article 53. The Commentary clarifies that “military operations refers to fighting or combat and activities directly related to it” and does not include general security considerations “or destruction of property as punishment or deterrent” (para. 3370).

Importantly, the Commentary suggests that such destruction, even when otherwise lawful, “must be proportionate to the anticipated military advantage” (para. 3371). This makes sense, as it aligns Article 53 with provisions regulating the protection of objects during active hostilities. Indeed, it would make no sense that attacks against objects during hostilities would be subject to proportionality, while destruction in occupied territories would not. I would argue more broadly that all security measures under occupation must be subjected to a similar proportionality analysis.

Reasserting IHL’s Core Protective Functions

The new Commentary demonstrates a loophole-closing frame of mind throughout much of the project. From a wider perspective, it sheds light on the extent to which some IHL norms have been utilized, throughout the years, to facilitate actions that unduly harm civilians despite their formal protective purpose. Reasserting IHL’s core protective functions, as the Commentary does, is an entirely warranted response – not only as a matter of policy or ethics, but also in the more basic sense that rules should be interpreted in a way that gives them actual effect.

Of course, closing loopholes is like a game of legal Whack-A-Mole. Each attempt may give rise to the exploitation of other potential loopholes. But that’s life in law. The new Commentary will definitely play a central role in this process of interpretation and contestation that will surely continue.

FEATURED IMAGE: Geneva, Switzerland – September 3, 2020: Close-up of the facade of the headquarters of the International Committee of the Red Cross (ICRC), a private humanitarian organization created in 1863.​ (Via Getty Images)