Minding the Gap, Properly: Conflict Classification, Protected Persons, and What the Geneva Conventions Already Require (Part I) – A Reply to Alexandria Virginski

[Yousuf Syed Khan is a Manager, Law and Policy at Legal Action Worldwide in Geneva. He is also a Nonresident Senior Fellow at the Atlantic Council (Strategic Litigation Project), and an Associate Fellow at the International Centre for Counter-Terrorism – The Hague.

Jérôme de Hemptinne is an Assistant Professor at Utrecht University.]

We read with real interest Alexandria Virginski’s recent post, Behind Enemy Lines: The Protection Gap Between POWs and Captured Nonstate Actors Under International Humanitarian Law. The subject is among the most consequential in contemporary international humanitarian law (IHL), and her closing intuition – that the humane treatment of a person in enemy hands ought not to turn on the accident of which flag that person fought under – is one we share. This reply is not meant to dispute that intuition. We find, however, that the doctrinal route Virginski takes misstates the law at several points, and does so in a more permissive direction.

This reply proceeds in two parts. Part I addresses the legal framework governing the classification and protection of persons in enemy hands, the relationship between the Third and Fourth Geneva Conventions, and the scope of protected person status. Part II turns to what those legal regimes require in practice, particularly in relation to internment and interrogation, before considering Virginski’s reform proposal and identifying where the more consequential gaps in the existing framework actually lie. We conclude with a few brief remarks on why getting that legal baseline right matters to the case for reform.

The Threshold Problem: POW Status is a Concept of International Armed Conflict

The post opens by establishing its subject as non-international armed conflict (NIAC), which it rightly identifies as the dominant form of contemporary warfare. It then conducts its legal analysis through Art. 4 of the Third Geneva Convention (GC III) and Art. 4 of the Fourth (GC IV). But both provisions, and the status regimes they create, operate only in international armed conflict (IAC). Prisoner of war (POW) status does not exist in NIAC, so it is not withheld from members of armed groups but is unavailable to everyone. A government soldier taken prisoner by an armed group in a purely NIAC context is no more a POW in law than his captors would be.

A second problem is with the post’s treatment of Art. 4(A)(2) GC III. That provision governs certain militias and volunteer corps, including organised resistance movements, “belonging to a Party to the conflict,” within the Convention’s POW regime. Its four cumulative conditions – responsible command, a fixed distinctive sign, arms carried openly, and compliance with the laws and customs of war – therefore become relevant only once that threshold framework applies. In a purely NIAC, however, Common Art. 3 does not confer combatant or POW status, and an independent non-State armed group would not be assessed under Art. 4(A)(2). The difficulty is therefore not, as Virginski assumes, whether such a group satisfies the four conditions – it is the logically prior point that the Art. 4(A)(2) POW regime is inapplicable, so neither “belonging” nor compliance with those conditions needs to be assessed.

This matters because it dissolves the asymmetry Virginski’s post is built around. The “protection gap between POWs and captured nonstate actors” cannot arise in a NIAC because there are no POWs there to compare anyone to. The gap the post describes – between a privileged combatant entitled to the full GC III regime and an unprivileged one entitled to less – is a feature of IAC, and specifically of the case where a person takes direct part in hostilities in an interstate conflict without satisfying the conditions for combatant status. That is a genuine and much debated problem, but it is a different one from what the post’s opening paragraphs promise to address, and conflating the two makes the analysis that follows difficult to apply to any actual conflict.

The Authorities that Decide These Questions are Absent

Read as a discussion of IAC, the post does not engage the authorities that govern the relationship between GC III and GC IV. It frames the question of whether a captured fighter who fails the GC III conditions falls under GC IV as unresolved and cites the 1958 Pictet Commentary’s statement that the drafters intended to leave no one uncovered, dating it to 1952. It then suggests a counterargument: because the IRA, al-Qaeda, ISIS, and Hamas did not exist when the Conventions were drafted, such groups “should remain excluded.” That is, regrettably, not a legal argument.

Treaty terms are general; a rule addressed to persons in enemy hands applies to them whenever the organisation they belong to was founded. The question has, in any event, been squarely addressed in international jurisprudence. In Delalić (Čelebići), the ICTY Trial Chamber held that there is no gap between the Third and Fourth Conventions: a person in enemy hands who is not entitled to POW treatment falls within the Fourth Convention, provided its Art. 4 requirements are satisfied (para. 271) – requirements which that Chamber read by reference to effective link rather than formal citizenship (paras. 257-265), and which the Tadić Appeals Chamber later read in terms of allegiance (paras. 163-169).

We do not claim the question is universally closed, and the ICRC’s updated Commentary on GC IV acknowledges two competing approaches, although it characterises protection under GC IV as the predominant approach and expressly endorses the broader interpretation (paras. 1058-1067). The United States has, in particular contexts, advanced narrower interpretations of the Fourth Convention’s personal scope. The predominant view, however, strongly supported by the text, object, and purpose of the Fourth Convention, by the Commentary the post itself invokes, and by Delalić, is that unprivileged participation in hostilities does not by itself place a person outside the Convention. That body of authority is the centre of the debate the post describes, and none of it is cited.

Moreover, Virginski never mentions Art. 75 AP I, although it supplies exactly the status-independent guarantees the post’s conclusion proposes creating from scratch, and although the United States, not a party to the Protocol, stated in 2011 that it would, out of a sense of legal obligation, treat the principles of that article as applicable to any person it detains in an IAC. Nor is Art. 45(3) AP I cited, under which anyone who has taken part in hostilities, is not entitled to POW status, and does not benefit from more favourable treatment under GC IV, has at all times the protection of Art. 75.

Read as a discussion of NIAC, the post is missing the principal treaty on the subject, Additional Protocol II (AP II). Its Art. 5 sets out detention conditions (food, water, health, correspondence, medical care, relief) that closely resemble the list the post proposes as a novel reform. The Protocol binds only where its Art. 1 threshold is met, which requires fighting between a State party’s own armed forces and dissident forces or an organised group that, under responsible command, controls territory sufficiently to sustain concerted military operations and to implement the Protocol. A great many NIACs fall below that line, and in those, Common Art. 3 and custom carry the load.

Rule 118 of the ICRC Customary IHL Study on treatment of persons deprived of liberty also goes uncited, as does Hamdan v. Rumsfeld, in which the US Supreme Court held Common Art. 3 applicable to the United States’ conflict with al-Qaeda. Nor is the Tadić Jurisdiction Decision of 2 October 1995 cited, which is presumably what the post has in mind when it says that such provisions were not applied to these conflicts “until the mid-1990s.” That may describe a turning point in international criminal adjudication, but not the emergence of the governing law: Common Art. 3 had regulated NIAC since 1949, and AP II, where its threshold was satisfied, since its entry into force in 1978.

The Nationality Analysis Runs the Wrong Way

Art. 4 GC IV protects those in the hands of a Party of which they are not nationals, and excludes: nationals of a State not bound by the Convention; nationals of a neutral State in a belligerent’s territory, and of a co-belligerent State, where their State has normal diplomatic representation in the detaining State; and persons already protected by GCs I, II, or III. Virginski applies the co-belligerent exclusion to conclude that captured Hamas members held by Israel would fall outside the Convention because “they are nationals of a co-belligerent state that has ‘normal’ diplomatic relations with Palestine.”

A co-belligerent is an ally – a State fighting on the same side as the detaining power – and Palestine and Israel are on opposite sides of the conflict the post is discussing. The exclusion has no application, and the reasoning also treats Palestine as both the detaining power’s co-belligerent and the State whose diplomatic relations are assessed, which cannot both be so.

The more important point is that the post reaches the opposite of the prevailing position, and on one component of it, the opposite of a position about as settled as anything contested in this field: the de jure applicability of GC IV to the occupied Palestinian territory, and the protected person status of Palestinian inhabitants who satisfy Art. 4. That is the conclusion of the International Court of Justice (ICJ) in its Advisory Opinion on the Legal Consequences of the Construction of a Wall, the longstanding view of the ICRC, and the repeated position of the Security Council and General Assembly. It follows from the plain text that they are in the hands of a Party of which they are not nationals, and none of the exclusions applies to them. The only other exclusion that might be reached, that for neutral State nationals, is confined by its terms to the territory of a belligerent and does not extend to occupied territory.

Whatever one thinks of how that framework has been applied in practice, the post cannot get to its answer without addressing the Wall opinion, which it does not cite. The harder question of whether a member of an organised armed group retains protected person status, and what rights may be derogated from if they do, is a real one, and we return to it below – but it is answered under Arts. 4 and 5 GC IV, rather than by reading the person out of Art. 4 at the threshold. In occupied territory, any assigned residence or internment is separately governed by Art. 78.

Art. 5 GC IV is a Derogation Clause, not an Exclusion Clause

The post then argues that “neither the drafting history nor the Fourth Geneva Convention itself suggests that the drafters would extend protected person status to those who did not qualify for prisoner of war status,” and offers Art. 5 GC IV as evidence that “the opposite” is true. Article 5 actually does the reverse of what Virginski cites it for.

Article 5 permits a State to withhold from a protected person suspected of hostile activity those Convention rights whose exercise would prejudice its security. To withhold rights under the Convention from a person, that person must be within the Convention, and Art. 5 presupposes protected person status. Far from showing that suspected fighters are outside GC IV, it confirms that the drafters contemplated exactly such people inside it and legislated for how they should be handled.

Two further aspects of Art. 5 are worth noting. The post quotes only Art. 5(1) and omits Art. 5(2) and (3). Article 5(3) expressly provides that persons subject to derogation “shall nevertheless be treated with humanity and, in case of trial, shall not be deprived of the rights of fair and regular trial prescribed by the present Convention,” and that their full rights are to be restored at the earliest date consistent with the security of the State or Occupying Power. Moreover, Art. 5(1), which the post quotes, governs persons in the territory of a Party to the conflict, whereas Art. 5(2) governs occupied territory, which is the provision relevant to the Israel/Palestine context used as the post’s principal example. Article 5 is a narrow, time-limited, humanity-preserving derogation, not a trapdoor.

“Protected Persons” Does not Mean Medical Personnel

The post goes on to state that protected persons “typically include doctors, nurses, technicians, and other personnel like staff of the National Red Cross Societies that are caring for the wounded and sick.” This is not what the term means in GC IV.

Art. 4 GC IV defines protected persons as those who find themselves, in a conflict or occupation, in the hands of a Party or Occupying Power of which they are not nationals, subject to specified exclusions. The operative criterion is the person’s nationality relationship to the Power in whose hands they find themselves, not civilian status as such – which is why, as we discuss above, a person who has taken up arms may nonetheless qualify.

Medical personnel, for their part, are not a single category. Permanent medical personnel of the armed forces under Art. 24 of the First Geneva Convention (GC I), and the personnel of recognised National Societies and other voluntary aid societies employed on the same duties under Art. 26, may, once in enemy hands, be retained under Art. 28 GC I and Art. 33 GC III. Auxiliary medical personnel under Art. 25 occupy a different position; if captured, they become prisoners of war under Art. 29 GC I. Civilian medical personnel receive specific protection under Art. 20 GC IV and, for States party to it, Art. 15 of Additional Protocol I (AP I). Where they hold protected person status, they hold it because of their relationship to the detaining power, not their profession. The phrase “protected persons,” as Virginski uses it, appears to have been drawn from a general ICRC explainer that employs the term more loosely, and then treated as though it were a definition under GC IV.

We would not dwell on a terminological slip except that this one does real argumentative work. Having defined protected persons as medical staff, Virginski uses that definition to defeat the scholarly position it just set out: “This fact undermines some scholars’ argument that nonstate armed forces should fall into this category, as these groups take up arms and are actively engaged in the conflict, which both harms—rather than cares for—those invocated [sic] in the conflict.” Yet the rebuttal only works if protected persons are care providers, which they are not. It is also pitched at the wrong level – Art. 4’s criteria are nationality and the listed exclusions, not membership of an organisation. The argument the post rejects is therefore left standing, and it is a serious one.

Part II of this post turns to the protections that govern detention in practice, before considering Virginski’s reform proposal and where the more consequential gaps in the existing framework actually lie.