I’d like to turn our readers’ attention to the comment thread of Constantin’s post, which has raised a fascinating issue – when does an internal armed conflict become internationalized? I’d like to add a few thoughts of my own, first on some matters of definition.

We first need to agree on what the ‘internationalization’ of an internal armed conflict actually means. To my mind, that concept is only legally useful if it denotes the transformation of a prima facie non-international armed conflict into an international one, thereby rendering applicable to the said conflict the more comprehensive IAC legal regime. As is well known, there has been a long-standing trend – promoted, for example, by the case law of the ICTY and the ICRC customary law study – of arguing that most of IAC rules now apply to NIACs as well.

Crucially, however, at least one distinction between the two legal regimes remains. In IACs, the parties to the conflict are (at least) two equal sovereigns. Lawful participants in the hostilities who in effect represent those sovereigns thereby have combatant status, and enjoy the privilege of belligerency. They cannot be prosecuted by the other party for their mere participation in the hostilities, but solely for violations of IHL. In NIACs, however, the parties are fundamentally different – most commonly a government and a rebellious non-state actor. Because governments have every right to suppress rebellions against them, no combatant status or privilege exists in NIACs. A rebel can be prosecuted for the mere fact that he is a rebel, even if he has been completely observant of the rules of IHL. Thus, for example, the government of Afghanistan has every right to imprison a Taliban soldier, even if that soldier committed no war crime.

Note that this distinction is based on party structure to the conflict and is therefore here to stay. Note also that because the distinction between IACs and NIACs is based on party structure, one cannot logically first ask the question (as Federico does in the comments) whether there is an armed conflict simpliciter, and the ask further whether that conflict is international or non-international. Rather, IACs and NIACs are separate legal categories, neither of which is residual in nature, as it is impossible to establish either without making an inquiry into party structure. In other words, an ‘armed conflict’ exists when there is an IAC or a NIAC, not the other way around.

Per Common Article 2 of the Geneva Conventions, IACs are defined as conflicts between states. There are thus two basic ways of ‘internationalizing’ a NIAC: (1) for treaties and/or custom to exceptionally expand the definition of an IAC to include as parties some sufficiently state-like entities, or (2) for the non-state actor which is a party to a NIAC with a state to be considered as acting on behalf of a third state.

As for option (1), I know of only two such possible exceptions: Art. 1(4) AP I, which renders international a conflict between a state and an oppressed people, which would otherwise be considered a NIAC; and the recognition of belligerency, which has now fallen into disuse, but perhaps not desuetude. As for the former, the idea behind Art. 1(4) AP I is that a people entitled to self-determination, whose right thereto is being denied by a state, exhibits a form of proto-statehood or sovereignty, an innate legitimacy that requires the application of the IAC regime, the privilege of belligerency and all. However, Art. 1(4) is not widely accepted as reflecting customary law, and it has to my knowledge never actually been applied. As for the recognition of belligerency, it allows the state fighting an insurgent and/or third states to recognize the fact that the magnitude of the insurgency is such that it would be appropriate to treat it as a belligerent. It has been used in the 19th century, when the operational concept of the law of war was, well, war, most notably during the American Civil War. It has not been used since, but it could conceivably still be used to ‘upgrade,’ as it were, a NIAC into an IAC, in case of particularly protracted civil war involving large-scale hostilities and stable insurgent control over large areas of territory (i.e. a conflict on an AP II scale).

Conceivably, one could demonstrate sufficient state practice and opinio juris to expand option (1) and the definition of an IAC even further, to include other situations. However, I don’t believe that such a case can convincingly be made for any currently relevant situation. It is thus option (2) which is of practical relevance – a NIAC will become internationalized if it is shown that a non-state actor fighting a state is actually doing so on behalf of a third state. In other words, if the definition of an IAC that we are operating under is solely the CA2 one, i.e. if option (1) is excluded, a conflict will become international only if there are two states parties to it.

This, I think, is an answer to Federico’s argument in the comments that the current conflict in Afghanistan has become internationalized because of the presence of foreign forces there. As other commenters pointed out, this cannot be the case, as there are no two states fighting each other. The US, the UK, and other NATO states are fighting the Taliban, a non-state actor, jointly with the government of Afghanistan. In effect, they are acting on its behalf, with its consent, and in alliance with it. This conflict cannot be international, because if it was, the Taliban would have the right to fight, and so long as they respected the rules of IHL they could not be punished merely for rebelling against the lawful government of Afghanistan.

To successfully argue otherwise, Federico would need to show that the definition of an IAC has changed through custom, and this is not something that I think he can do. ‘International’ armed conflict means precisely that – a conflict between nations – unless an exception to that definition is established, as under my option (1). In that regard, I think Federico’s reliance on the Israeli Supreme Court’s Targeted Killings judgment is misplaced. There the Court wrongly considered an IAC to be any armed conflict which crosses the state boundary, but that is not the CA2 definition, nor is it supported by state practice. Likewise, that the Israeli-Palestinian conflict is taking place in an occupied territory (as argued, e.g., by Cassese) is also insufficient to internationalize a NIAC, as again there are no two states parties to the conflict, while Art. 1(4) AP I is unavailable since Israel is not a party. For more, see my article on Hamdan and Targeted Killings in the 2007 International Review of the Red Cross.

Thus, both Iraq and Afghanistan after the overthrow of Saddam and the Taliban, as well as the Israeli-Palestinian conflict, can to my mind only be qualified as NIACs. They cannot be internationalized, because there are no two opposing states, and because no other exception to the definition of an IAC as an inter-state conflict can be proven.

Hence, there are in my view only two possible scenarios of the internationalization of prima facie NIACs involving a state and a non-state actor:

(i)              State A intervenes into an internal conflict in state B, in support of the non-state actor and against state B. This is the Bosnian scenario, where Serbia and Croatia supported the Bosnian Serbs and Croats against the internationally recognized government of Bosnia. This is likewise the scenario of the coalition attack on Afghanistan post 9/11, when they acted jointly with the Northern Alliance against the Taliban who were then the de facto government of Afghanistan, before the new government of Afghanistan was set up.

(ii)            State A attacks a non-state actor located in state B, without B’s consent. This is the scenario of the 2006 Israel-Hezbollah conflict in Lebanon, and this is the scenario to which Dapo refers in the comments.

What, then, would be the exact mechanism whereby the prima facie NIACs are turned into IACs? Commenter Julian argues that this mechanism is attribution as a matter of the secondary rules of state responsibility. This, of course, was famously the approach of the ICTY Appeals Chamber in Tadic, where it considered that the acts of the Bosnian Serbs had to be attributable to the FRY/Serbia in order for the conflict to become international, and fashioned the ‘overall control’ test of responsibility in order to do so, thereby rejecting the ICJ’s approach to attribution in the Nicaragua case.

There is an intuitive appeal to the Tadic approach. After all, what else could it mean for a non-state actor to be acting on behalf of state than for its acts to be attributable to the state?

However, I have argued at length in my article on State Responsibility for Genocide, (2006) 17 EJIL 553, at 575 ff, that the Appeals Chamber’s approach was misguided, for two basis reasons. First, it actually misinterpreted the ICJ’s Nicaragua judgment as setting out only ONE test of attribution, that of effective control, and thought that this single test was unreasonable and impracticable. Indeed, it would be so, had the ICJ not set out TWO tests of attribution – that of complete dependence and control, operating at a general level and seeking to attribute ALL of the acts of a non-state actor to a state, and that of effective control, seeking to attribute SPECIFIC acts controlled by the state.

Second, and more importantly for our purposes, I have argued that it is conceptually inappropriate for secondary rules of attribution to determine the scope of application of the primary rules of IHL. Rather, it is upon IHL to fashion a test which determines when the relationship between a state and a non-state actor is such that a NIAC is to be internationalized – and that test may well be that of overall control (at 584-585).

Similarly, in its 2007 Bosnian Genocide merits judgment, the ICJ held that (paras. 404-405):

This is the case of the doctrine laid down in the Tadić Judgment. Insofar as the “overall control” test is employed to determine whether or not an armed conflict is international, which was the sole question which the Appeals Chamber was called upon to decide, it may well be that the test is applicable and suitable; the Court does not however think it appropriate to take a position on the point in the present case, as there is no need to resolve it for purposes of the present Judgment. On the other hand, the ICTY presented the “overall control” test as equally applicable under the law of State responsibility for the purpose of determining ⎯ as the Court is required to do in the present case ⎯ when a State is responsible for acts committed by paramilitary units, armed forces which are not among its official organs. In this context, the argument in favour of that test is unpersuasive.

It should first be observed that logic does not require the same test to be adopted in resolving the two issues, which are very different in nature: the degree and nature of a State’s involvement in an armed conflict on another State’s territory which is required for the conflict to be characterized as international, can very well, and without logical inconsistency, differ from the degree and nature of involvement required to give rise to that State’s responsibility for a specific act committed in the course of the conflict.

Again, I do not disagree that there is an intuitive appeal to the Tadic approach. In that regard, the ICJ’s holding to the contrary was criticized quite ably by Marina Spinedi in her excellent article ‘On the Non-Attribution of the Bosnian Serbs’ Conduct to Serbia,’ (2007) 5 JICJ 829.

However, in my view, Tadic is right only to the extent that if the acts of the Bosnian Serbs were attributable to Serbia, then the conflict would surely have been international in nature, since the Bosnian Serbs would not have been a non-state actor at all, but agents of Serbia. Likewise, if the acts of Hezbollah were attributable to the state of Lebanon, for example because Hezbollah acted in the absence or default of official authorities in Southern Lebanon, then the 2006 Israel-Hezbollah conflict would in fact simply have been an international conflict between Israel and Lebanon. However, IHL, as a distinct body of primary rules, can adopt its own solution regarding the link between a state and a non-state actor that would suffice for internationalization of a conflict, and that link need not be attribution as a matter of state responsibility.

As for the nature of that link, I would like to draw the readers’ attention to another excellent article, this time in the most recent issue of the EJIL – Katherine Del Mar, ‘The Requirement of ‘Belonging’ under International Humanitarian Law,’ (2010) 21 EJIL 105, focusing on the ‘belonging to a party’ test under Art. 4(A)(2) GC III. Katherine argues – and I agree – that a non-state actor militia group may well ‘belong’ to a state party to an IAC, on the basis of nothing more than a de facto agreement, but at the same time its acts might not be attributable to the state under the general rules of state responsibility.

In short, the internationalization of a prima facie NIAC of the first type that I have set out above – an intervention into a civil conflict by a third state – depends on the nature of the relationship between the intervening state and a non-state actor. That relationship may, but need not be, attribution. A relationship of overall control, which does not suffice for attribution, or de facto agreement may also suffice.

What, then, of the second possible type of internationalization, that of the Israel-Hezbollah example, assuming, arguendo, that the acts of Hezbollah were NOT attributable to Lebanon? My view is the same as Dapo’s – it is the very fact that state A uses force against the people of state B without its consent that creates an international armed conflict. Israel’s invasion of Lebanon initiated a conflict with Lebanon, not just with Hezbollah. This would remain the case even if Israel limited itself strictly to attacking Hezbollah targets (which it did not). To my mind, Hezbollah could then be classified either as a militia fighting on behalf of Lebanon, whose members would have to satisfy the additional requirements of Art. 4(A)(2) GC III in order to earn combatant status, or as civilians taking a direct part in hostilities.

The alternative to this approach would be to limit internationalization strictly to those cases where the acts of the non-state actor could be attributed to the state. Since the general rules of attribution are strict precisely because they are general, this would either greatly limit the number of possible cases of internationalized NIACs, or would lead to jurisprudential conflicts on attribution of the Nicaragua/Tadic/Genocide variety. Likewise, this would necessitate accepting the possibility of parallel conflicts, as was of course done by the ICTY in a number of cases. Thus, there would have been an IAC between Bosnia and Serbia only when the acts of the Bosnian Serbs were attributable to Serbia, while a NIAC between the Bosnian government and the Bosnian Serbs would be running in parallel. Likewise, there would have been an IAC between Israel and Lebanon and a parallel NIAC between Israel and Hezbollah, unless the acts of Hezbollah could be attributed to Lebanon.

To my mind, this alternative is not only messy, but is needlessly so. Such a patchwork of overlapping regimes might be unworkable, and is to my mind undesirable. But, of course, reasonable people can disagree.