Self-Defence and Non-State Actors: An Inquiry into the Morphology of the Right of Self-Defence in International Law

Abstract

This article examines whether states may use force in self-defence against non-state actors operating from the territory of another state. It first maps the prevailing approaches to self-defence in response to non-state armed attacks, focusing on attribution, Article 3(g) of the United Nations General Assembly’s Definition of Aggression, and the ‘unable or unwilling’ doctrine. It argues that these approaches give rise to normative, methodological and practical difficulties and often leave states without an effective right to defend themselves. The article then goes on to explain the morphology of the right of self-defence. It contends that self-defence is a primary rule of international law permitting the use of force in response to an armed attack as a factual occurrence, irrespective of its perpetrator. It further argues that self-defence is neither an exception to Article 2(4), nor a legal sanction. On this basis, the right of self-defence extends to armed attacks by non-state actors, even when they operate from within the territory of another state. In such cases, self-defence as a circumstance precluding wrongfulness excuses the responsibility of the defending state for incidental violations of the territorial state’s sovereignty in the course of defensive action.

1. Introduction

The attacks carried out by Hamas on 7 October 2023, followed by subsequent attacks by Hezbollah and the Houthis, against Israel have once again brought to the forefront the question of the legality of using defensive force against non-state actors (NSAs).¹ Although this debate is not new, the challenges posed by NSAs, however, have evolved considerably as a result of the increasing frequency, scale and sophistication of their attacks. Also, while some NSAs operate as proxies for states – advancing their political, military or strategic objectives – others act with a significant degree of independence from state sponsorship, even while conducting operations from within the territory of states.

Non-state armed attacks on states raise complex legal questions concerning the applicability of the right of self-defence as stipulated in Article 51 of the UN Charter² and customary international law, particularly when defensive action is carried out within the territory of the state from which the NSA operates.

In this article, I argue that a state may lawfully resort to force in self-defence on the territory of another state against a non-state actor that has carried out an armed attack. This argument rests on an analysis of the morphology of the right of self-defence in international law. By morphology, I refer to the structural composition, internal logic, features and relational dimensions of this right within the broader legal system to which it belongs. In my view, it is only through such understanding that the application of the right of self-defence to attacks by non-state actors can be properly assessed and normatively justified.

To support this argument, I first present existing approaches to self-defence as reflected in international jurisprudence and academic commentary. More specifically, I start with the traditional approach, which conceives of self-defence as an interstate affair, permitting the use of defensive force only against another state either when a non-state armed attack is attributable to that state (Section 2.1) or when a state is regarded as the constructive author of the non-state armed attack (Section 2.2). I will then turn to the ‘unable or unwilling’ doctrine, which permits the use of defensive force against non-state actors operating from the territory of another state, provided that the territorial state is either unable or unwilling to prevent or suppress their attacks (Section 3). I will argue that they suffer from normative and methodological weaknesses and, in most cases, they deny the victim state a lawful means of defending itself.

To answer the challenges posed by NSAs and to present a normatively and methodologically coherent justification for the use of defensive force, I will proceed to explain the morphology of the right of self-defence (Section 4). As I will explain, self-defence operates both as a primary and as a secondary rule of international law. As a primary rule, enshrined in Article 51 of the UN Charter and customary international law, it establishes the entitlement of a state to use force in response to an armed attack regardless of whether that attack is carried out by another state or by a non-state actor. I will also argue that, as a primary rule, it is not normatively or methodologically dependent on the prohibition of the use of force.

As a secondary rule, self-defence functions as a circumstance precluding wrongfulness as stipulated in Article 21 of the International Law Commission’s (ILC) Articles on Responsibility of States for Internationally Wrongful Acts (ARSIWA). In this capacity, its role is distinct: it excuses the wrongfulness of incidental violations of international obligations that may occur in the lawful exercise of the primary right of self-defence.

Before proceeding, three caveats must be noted. First, this article does not address all aspects of the right of self-defence, such as the requirements of necessity and proportionality. Second, the analysis is confined to the jus ad bellum framework and does not examine jus in bello considerations arising from the exercise of self-defence. Third, the article does not provide a detailed examination of all state practice involving the use of defensive force against NSAs.

2. The interstate paradigm of self-defence and non-state armed attacks

As was noted, the traditional construction of the right of self-defence conceives it as operating exclusively within an interstate context.³ This conception of self-defence aligns self-defence with Article 2(4) of the UN Charter, which confines the prohibition of the use of force to interstate relations, and with the understanding of self-defence either as an exception to that prohibition or as part of a broader normative ensemble with Article 2(4).

When it comes to non-state armed attacks, the interstate paradigm of self-defence is maintained in two principal ways: first, through the mechanism of attribution whereby a non-state armed attack is imputed to a state according to the secondary rules of attribution found in the law of state responsibility; and, secondly, through the notion of constructive state authorship of an armed attack on the basis of Article 3(g) of the UN General Assembly’s Definition of Aggression.

Although these two approaches overlap in that they both treat the state concerned, rather than the non-state actor, as the author of the armed attack and the proper target of self-defence, and both require some degree of state involvement in the non-state armed attack, they differ in the degree of state involvement required, as well as in their normative and methodological premises as is explained below.

I will now discuss these two approaches in more detail.

2.1. Attribution of non-state attacks to a state

According to this approach, which is prevalent in scholarship, a state may exercise the right of self-defence only against the state to which a non-state armed attack has been attributed under the attribution criteria of the law of state responsibility. Attribution refers to the legal process of imputation under which conduct that is prima facie private – such as an armed attack carried out by a non-state actor – becomes state conduct. If a non-state armed attack cannot be attributed to a state, the right of self-defence is considered unavailable to the victim state.

Attribution as a mechanism of legally imputing armed attacks to a state is reflected in Articles 8, 9 and 11 ARSIWA. These provisions are distinct from Article 4 ARSIWA, which encompasses armed attacks by de jure or de facto state organs,⁴ and Article 5 ARSIWA, which covers armed attacks by entities empowered by a state to exercise elements of governmental authority.⁵ In the latter two cases, the armed attacks are authored ipso jure and directly by a state because they are carried out by its own organs or entities that are part of its institutional structure, and not because of legal imputation of armed attack in a secondary sense, which is what attribution is about. In what follows I will thus focus on how non-state armed attacks can be attributed to a state under Articles 8, 9 and 11 ARSIWA.

Article 8 ARSIWA attributes to a state armed attacks committed by NSAs acting under its instructions, direction or control. Instructions require specific state orders whereas direction requires state guidance short of orders. Regarding the criterion of control, the International Court of Justice (ICJ) requires ‘effective’ control.⁶ Although the ICJ has not clarified precisely what this entails, its rejection of alternative standards such as ‘overall control’ suggests that effective control requires indispensable state input into the non-state armed attack, amounting in effect to its causation. Applying this attribution criterion to the October 7 attacks, attribution on the basis of Article 8 ARSIWA is not tenable because the available information does not indicate that Hamas acted under the instructions, direction or effective control of any state.⁷

Turning now to Article 9 ARSIWA, an armed attack by a non-state actor exercising elements of governmental authority in the absence or in default of state authorities will be attributed to that state. This provision addresses incidental and temporally limited exercise of governmental powers by NSAs in lieu of the government when the state apparatus fails to function.⁸ Although Hamas acted as the de facto government of Gaza, it was not temporary and, more critically, it was not ‘called for’ in the sense of arising from necessity. On the contrary, Hamas fought and deliberately excluded the Palestinian Authority from Gaza. Consequently, the October 7 attacks cannot be attributed to Palestine.⁹

Finally, under Article 11 ARSIWA, an armed attack by a non-state actor will be attributed retroactively to a state if the state acknowledges and adopts it as its own.¹⁰ These are cumulative conditions requiring the state to identify itself with the armed attack. Applying this criterion to the October 7 attacks, they cannot be attributed to any state. Although the Palestinian Authority remained silent and ‘blatantly refuse[d] to condemn the barbaric Palestinian terror attacks against Israeli civilians’,¹¹ silence alone does not amount to acknowledgement and adoption under Article 11, even if it may be politically meaningful. Similarly, reports that Iran’s president congratulated the Hamas leader¹² would also not amount to attribution under Article 11 because the latter requires clear and unequivocal endorsement of the act rather than political or symbolic support.

It follows from the preceding discussion that if the right to self-defence is made contingent on attribution, states will be left in most cases legally paralysed in the face of non-state armed attacks.

This is because the attribution criteria in the law of state responsibility are narrow, reflecting the binary distinction between the public (state) and private (non-state) sphere, which derives from a classical, Westphalian understanding of the state as a centralised, hierarchical and bureaucratic entity.¹³ According to this view, state responsibility for the acts of private actors attaches only when there is a demonstrable link between the NSA and the state apparatus, or their conduct is closely tied to the will and direction of the state. However, this restrictive approach is ill-suited for addressing the contemporary blurring of the public–private divide in the relations between states and NSAs, and the fact that NSAs can autonomously pursue their agendas while their action can affect states as well as the international order.

Methodologically, relying on attribution for applying the right of self-defence effectively subsumes the jus ad bellum under the framework of the law of state responsibility despite the fact that these two bodies of law serve fundamentally different purposes and operate according to different types of logic. Whereas the law of state responsibility is concerned primarily with the legal consequences that arise from internationally wrongful acts and is remedial in nature, the jus ad bellum concerns the threshold conditions – such as armed attack, necessity and proportionality – under which states may lawfully resort to force in the first place. It is prescriptive or proscriptive, not remedial in nature as is the law of state responsibility.

Consequently, determining the circumstances under which a state is responsible for the acts of NSAs is a different question from whether a state has engaged in an armed attack, or what state involvement in a non-state armed attack makes it a state armed attack for the jus ad bellum regime. Also, establishing state responsibility for the acts of NSAs does not in itself justify the use of force by way of self-defence.¹⁴

2.2. ‘Substantial involvement’ and constructive authorship of an armed attack

The other basis of the interstate paradigm draws on Article 3(g) of the UN General Assembly’s Definition of Aggression.¹⁵ This provision has been relied upon by the ICJ in interpreting the notion of armed attack and is widely regarded as reflecting customary international law.¹⁶

Article 3(g) first defines direct authorship of an armed attack when the non-state actor committing the armed attack operates as de facto organ of that state. This is the case where it is sent by that state to commit an armed attack or acts on its behalf.¹⁷ The Article also defines constructive authorship of an armed attack when a state is substantially involved therein.¹⁸

It is therefore important to explain what amounts to ‘substantial involvement’. Prima facie, it denotes assistance that enables the non-state armed attack, thereby making the assisting state an engaged participant in the armed attack though not its direct author, which would fall under the first prong of Article 3(g).

In the Nicaragua case, the United States argued that Nicaragua had committed an armed attack by being substantially involved in the guerilla armed attacks carried out in neighbouring states through the provision of arms, munitions, finance, logistics, training, safe havens, planning, command and control facilities.¹⁹ The ICJ, however, excluded ‘the provision of weapons, logistical support or other support’²⁰ from the concept of armed attack and consequently rejected the US claim. This led to criticisms by Judge Jennings and Judge Schwebel. In Judge Schwebel’s opinion, Nicaragua’s support for the El Salvadorian insurgents – which involved providing arms, munitions, other supplies, training, command and control facilities, sanctuary and lesser forms of assistance – constituted substantial involvement rendering Nicaragua the perpetrator of an armed attack.²¹ For Judge Jennings:²²

Logistical support may itself be crucial. … [It] covers the ‘art of moving, lodging, and supplying troops and equipment’ … If there is added to all this ‘other support’, it becomes difficult to understand what it is, short of direct attack by a State’s own forces, that may not be done apparently without a lawful response in the form of … self-defence.

The Court did not clarify the type or degree of assistance that might amount to substantial involvement but its rejection of the US interpretation suggests that a state’s involvement should reach such a degree as to render the NSA its de facto organ or amount to effective control.²³ Methodologically, the Court focused on what constitutes direct state authorship of an armed attack, rather than assessing whether the various forms of assistance, taken individually or cumulatively, could amount to substantial involvement sufficient to render that state the author of an armed attack.

As a result, the ‘substantial involvement’ standard was rendered redundant, having been assimilated into the ‘de facto organ’ criterion, which also explains the ICJ’s consistent reluctance to apply this standard in subsequent cases even when it referred to Article 3(g). In the Armed Activities (DRC v Uganda) case, for instance, the ICJ confined its analysis of state involvement to the ‘sending’ and ‘on behalf’ categories of this provision.²⁴

The Court’s interpretation, however, is contrary to the wording and negotiating history of Article 3(g).²⁵ The ‘substantial involvement’ provision was conceived as an independent basis for determining a state armed attack designed to capture various forms of state complicity, encompassing not only material support but also moral support or omissions such as toleration.²⁶ Furthermore, the term ‘substantial involvement’ should be compared with the phrase ‘open and active participation’²⁷ which it replaced and which was narrower in scope, excluding passive forms of support or omissions. The Court’s narrow interpretation also stands in sharp contrast with its more expansive interpretation of state involvement that amounts to a prohibited use of force.²⁸

From a practical perspective, the ICJ interpretation of ‘substantial involvement’ effectively excludes all non-state armed attacks from the scope of self-defence, unless the NSA qualifies as a de facto organ of a state. This leads to the same gaps inherent in the attribution approach discussed earlier. For example, Israel would have no right to invoke self-defence in response to the October 7 attacks, as no state can be said to have been so substantially involved as to render Hamas a de facto organ of that state.

From a methodological standpoint, it seems that the ICJ has interpreted Article 3(g) through the lens of state responsibility, even though neither attribution nor the law of state responsibility played any role in the General Assembly’s Definition of Aggression. In particular, earlier iterations of Article 8 attributed to a state ‘the conduct of private persons in fact performing public functions, or in fact acting on behalf of the State as de facto officials’, with such de facto organs including persons dispatched by a state to carry out specific missions, such as abduction or sabotage.²⁹ This has generated confusion over the nature of Article 3(g). For instance, some argue that the terms ‘sending’ and ‘on behalf of’ operate as standards of attribution, reflecting the criteria set out in Article 8 ARSIWA, while ‘substantial involvement’ resembles the ‘overall control’ test developed in international jurisprudence and scholarship.³⁰ Others regard these terms as special rules of attribution,³¹ or even as primary rules of attribution embedded within Article 3(g) itself.³² This, however, is not correct: Article 3(g) does not lay down criteria of attribution.³³ Rather, it constitutes a primary rule that defines what qualifies as a state armed attack for the purposes of the right of self-defence.

More recently, state assistance to non-state armed attacks has been addressed in the so-called Bethlehem ‘Principles Relevant to the Scope of a State’s Right of Self-Defense Against an Imminent or Actual Armed Attack by Nonstate Actors’,³⁴ which seek to clarify the circumstances under which support for such actors may trigger a lawful defensive response. Although their precise legal status remains unsettled,³⁵ these principles are described as being ‘intended to work with the grain of the UN Charter as well as customary international law’.³⁶ It is therefore important to consider how these principles relate to the ‘substantial involvement’ standard in Article 3(g) of the 1974 UN Definition of Aggression, which, as was noted, reflects customary international law.

According to Principle 7:³⁷

[a]rmed action in self-defense … may also be directed against those in respect of whom there is a strong, reasonable, and objective basis for concluding that they are taking a direct part in those attacks through the provision of material support essential to the attacks.

Principle 7 appears to suggest that only material assistance provided by a state to a non-state actor that is likely to directly cause the armed attack can justify self-defence action against that state. It thus requires a high degree of involvement akin to the ‘de facto organ’ or ‘effective control’ approach adopted by the ICJ in contrast to the lower threshold envisaged by the ‘substantial involvement’ standard.

More critically, from a normative and methodological perspective, Principle 7 does not clarify whether an assisting state that participates directly in the non-state armed attack is to be regarded in legal terms as the author of the attack in place of the non-state actor or along with the non-state actor. Rather, it appears to justify self-defence action against the state and/or the non-state actor as a matter of policy, raising questions of legal justification.

3. The ‘unable or unwilling’ doctrine and non-state armed attacks

According to the ‘unable or unwilling’ doctrine,³⁸ a state that suffered an armed attack by a NSA may take direct self-defence action against that NSA, even when the NSA operates from within the territory of another state, provided that the territorial state is unable or unwilling to prevent or suppress these attacks.³⁹ This doctrine has been invoked, explicitly or implicitly, by many states to justify their use of defensive force against non-state actors, as in the case of Daesh (ISIS) in Syria.⁴⁰

However, it faces normative and methodological challenges. To begin with, its legal status remains unsettled;⁴¹ consequently, it cannot be regarded as a reliable basis of action. Furthermore, the fact that the meaning of ‘unable’ or ‘unwilling’ remains undefined, along with the absence of criteria and evidentiary standards for making such determinations and the uncertainty surrounding the relationship between the two, risks undermining the doctrine’s normativity by making it personalised and open to instrumentalisation. This erodes its methodological integrity by compromising its coherence and consistency.

Normative and methodological questions also arise because of the overlap between ‘unwillingness’ and ‘substantial involvement’. Unwillingness may encompass various forms of support – such as toleration, funding, logistical support or training of NSAs – which coincide with the forms of conduct associated with the ‘substantial involvement’ standard discussed in Section 2.2. Questions thus arise as to which forms of assistance fall within the notion of unwillingness, which constitute substantial involvement, and where the threshold separating the two is located. Distinguishing between them is important as the legal implications differ: ‘substantial involvement’ renders the state itself the author of the armed attack whereas under the ‘unable or unwilling’ doctrine, the non-state actor becomes the lawful target of self-defence.

For instance, Iran claimed that Iraq provided Kurdish groups operating from its territory with military training, financial and logistical support, intelligence services and sanctuary, and had been unable to control them despite repeated warnings against the use of its territory by these groups for launching attacks on Iran.⁴² Iran subsequently undertook self-defence action against these groups rather than against Iraq even though this would have been permissible under the substantial involvement standard. Similarly, Iran relied on similar grounds to justify its 2018 strikes against the al-Ahvaziya group in Syria, which it held responsible for terrorist attacks, rather than against Saudi Arabia, which according to Iran supported this group.⁴³

According to the formulation of this doctrine in the Bethlehem principles, both the non-state actor and the assisting state may become the target of self-defence action depending on the extent of the state’s responsibility in aiding or assisting the non-state actor without clarifying the threshold at which such responsibility arises.⁴⁴ The difficulty, however, with the Bethlehem principles is that, as was noted, they seek to offer pragmatic solutions for the problem of non-state armed attacks but they do not always offer legal justifications.

A further challenge concerns the function of the ‘unable or unwilling’ doctrine. The ‘unable’ prong, for example, has been construed as part of the definition of a non-state armed attack in the context of cyber-attacks without, however, clarifying whether the self-defence action will be directed against the NSA or against the unable state.⁴⁵ The ‘unwilling’ prong, particularly in the form of harbouring or toleration, has instead been interpreted as a special attribution criterion notably in the context of the US self-defence action against Afghanistan following the ‘9/11’ attacks.⁴⁶ Yet, whether such lex specialis rule of attribution has been established is not settled⁴⁷ and, moreover, it is not clear whether the US relied instead on the ‘substantial involvement’ standard as discussed in Section 2.2. Certain commentators have also argued that ‘unwilling’ is an attribution criterion equating it with the ‘substantial involvement’ standard, which they also treat as an attribution criterion.⁴⁸ This brings to the fore the methodological problems discussed above – namely, how to determine what amounts to ‘unwilling’ and what amounts to ‘substantial involvement’. Methodologically, it also conflates primary rules on the use of force with secondary rules of the law of state responsibility by transforming the notion of ‘substantial involvement’ into one of attribution. Such an interpretation would imply that it is the state rather than the non-state actor that is deemed the author of the armed attack and is therefore the lawful target of self-defence, a conclusion that is not supported by state practice. In most cases, states such as Iran in the cases discussed above⁴⁹ and the coalition that acted against Daesh in Syria⁵⁰ explicitly framed their actions as directed against the non-state actor rather than against the territorial state.

Another question concerning the normative status and methodological structure of the doctrine is whether it departs from the traditional interstate paradigm of self-defence by recognising in legal terms NSAs as the authors of armed attacks and thus as lawful objects of self-defence. The answer is in the negative. The doctrine addresses only the circumstances in which a state can act in self-defence when confronted with armed attacks by non-state actors. In other words, it concerns the practical conditions of necessity in the exercise of self-defence⁵¹ rather than the normative question of whether non-state actors can, as a matter of law, be the authors of an armed attack.

Consequently, the doctrine preserves the state-to-state relational logic of self-defence by making self-defence dependent on whether the territorial state has fulfilled its international law obligations not to allow its territory to be used by non-state actors to commit armed attacks or not to assist non-state actors.⁵² In doing so, it also implicitly transposes principles of the law of state responsibility in the law regulating the use of force.⁵³

A major shortcoming of the doctrine is that it does not explain how the use of defensive force in the territory of another state that is not itself the author of the armed attack can be justified in the absence of that state’s consent. Territorial states, as well as other states, have condemned such operations as violations of sovereignty. For example, Syria denounced actions on its territory against Daesh⁵⁴ or Hezbollah⁵⁵ as violations of its sovereignty. Similarly, Lebanon condemned Israel’s strikes on its territory against Hezbollah or Hamas as a violation of its sovereignty,⁵⁶ whereas Iran condemned US and UK attacks on the Houthis in Yemen as a blatant violation of Yemen’s sovereignty and territorial integrity.⁵⁷

Responses to this question vary. For example, according to the Bethlehem principles, the requirement of consent does not operate when the territorial state is unable or unwilling without, however, offering a legal basis for such assertion; whereas if the territorial state refuses to agree on a reasonable and effective plan of action, it becomes unwilling, justifying self-defence action.⁵⁸ Others have invoked the law of neutrality, under which a belligerent may take self-defence measures on the territory of a neutral state that has failed to discharge its neutrality obligations.⁵⁹ However, extending this reasoning to an entirely different legal regime and to situations involving non-state actors represents a considerable conceptual and normative leap. One could also say that the self-defence action is corrective of the state’s failure to prevent non-state armed attacks, thereby operating as a quasi-sanction by putting pressure on the territorial state to change its behaviour and signalling that such behaviour is intolerable. That having been said, no legally convincing justification has been offered.

4. The morphology of the right of self-defence and its application to non-state actors

In the light of the normative, methodological and practical challenges associated with these approaches, this section sets out to explain the morphology of the right of self-defence in order to demonstrate the legality of defensive action against non-state armed attacks, even when such action is carried out on the territory of another state.⁶⁰ It is only by understanding the history, nature, features and scope of the right of self-defence, its relationship with other norms and legal regimes, and the distinct legal relations and consequences it entails – matters not always consistently appreciated – that its application to non-state armed attacks can be properly explained and justified.

4.1. The origins of self-defence

Recalling the history of the right of self-defence is important because it reveals key features of this right and its continuity within the modern jus ad bellum framework. Self-defence has deep roots in international law predating its codification in the UN Charter. It was part and parcel of the right of self-preservation, which gave rise to more specific rights and justified the use of force to protect a range of state interests from diverse dangers.⁶¹ Among these interests, the human and physical existence of the state were regarded as paramount and attacks on them were considered to warrant recourse to force in ‘legitimate’ or ‘necessary’ defence even within the territory of another state.⁶²

Self-defence thus historically addressed broader security concerns and predated the development of the prohibition of the use of force, which in legal terms began to emerge from the late 1920s.⁶³ Even in that context, self-defence was viewed as an inalienable right, an idea echoed in Article 51 of the UN Charter, which recognises the ‘inherent right of self-defence’.⁶⁴ While this phrase is now understood to refer to the pre-existing customary law of self-defence to avoid the natural law connotations that the term ‘inalienable’ may invoke, it nonetheless acknowledges a pre-existing right and affirms the long historical pedigree of self-defence in contrast to Article 2(4) of the UN Charter on the prohibition of the use of force, which is a Charter-based creation. The deep-rooted historical foundation of self-defence also reveals another feature of this right, discussed below – namely, that it constitutes a primary norm of the jus ad bellum regime and is not an exception to Article 2(4) but rather a coexisting and complementary norm within the jus ad bellum framework.

4.2. Self-defence is a primary norm of the jus ad bellum regime

According to Ago’s definition, primary rules establish rights and obligations whereas secondary rules set out the consequences of their breach.⁶⁵ Regarding self-defence, Article 51 of the UN Charter stipulates self-defence as a right and, indeed, in absolute terms, ‘nothing in the Charter shall impair the inherent right of self-defence’. Article 51 and the corresponding customary rule of self-defence thus confer upon states an entitlement to use force in the event of an armed attack. Such use of force is lawful per se and ab initio: it is not tainted by an initial presumption of wrongfulness – and does not give rise to consequent questions of responsibility provided it is exercised according to the conditions attached to self-defence.⁶⁶

4.3. Self-defence is not an exception to Article 2(4)

Self-defence as a primary norm is not an exception to the prohibition of the use of force in Article 2(4) of the UN Charter but is an independent rule standing on equal footing with Article 2(4).⁶⁷ There is therefore no normative hierarchy or dependency of Article 51 on Article 2(4) which characterises a rule–exception relationship.⁶⁸ Rather, the two rules are normatively and structurally autonomous from one another, with asymmetrical application based on their distinct context-specific triggers. This means that the interstate character of Article 2(4) does not operate to limit the scope of the right to self-defence, particularly whether it applies in response to non-state armed attacks.

The view that self-defence cannot be regarded as an exception to Article 2(4) is further supported by the fact that, as noted earlier, it predates that provision, and the Charter acknowledged an already established right rather than creating a new right or carving out an exception to a new rule. During the drafting of the UN Charter, it was expressly recognised that the right to self-defence should not be impaired or diminished by the inclusion of Article 2(4).⁶⁹

If self-defence were intended as an exception to Article 2(4), it would have been normatively and methodologically more consistent, and much simpler, to recognise a new treaty-based right of self-defence or integrate it directly into Article 2(4) to avoid any conflicts that may arise from the application of seemingly antithetical rules. This, however, is not how these rules were formulated in the UN Charter, each standing for a separate norm. That said, it has been argued that in the absence of such integration, conflicts between rules and their exceptions can be avoided by constructing ‘derived rules’ that merge the content of the two rules into a single normative framework.⁷⁰ Indeed, some commentators have advanced the view that Articles 2(4) and 51 form a normative ensemble.⁷¹ Yet, this is a meta-construction to justify the assumption that self-defence forms an exception to Article 2(4), which, however, is not supported by the manner in which the two rules have been formulated in the UN Charter.

In fact, the UN Charter established a legal order composed of smaller rules that are, in Raz’s words, ‘self-contained (or self-explanatory)’, guiding state behaviour by imposing duties or conferring rights.⁷² Self-defence is one such ‘self-contained and self-explanatory’ rule, equal with and coexisting alongside Article 2(4). Whereas Article 51 confers a right to use force when its context-specific trigger – an armed attack – occurs, Article 2(4) imposes a duty not to use force. Both rules derive from values embedded in the Charter system, such as peace, non-aggression and sovereignty,⁷³ and neither rule renders the other redundant, as each addresses a distinct act-situation and creates distinct legal relations.⁷⁴

The normative as well as structural separation between Articles 2(4) and 51 are also supported by their position within the UN legal order. Article 51 is not located in Chapter I of the UN Charter where Article 2(4) is found but in Chapter VII entitled ‘Action with Respect to Threats to the Peace, Breaches of the Peace, and Acts of Aggression’. Chapter VII provides for both collective and unilateral action to maintain or restore international peace and security, while limiting unilateral action in the form of self-defence until the Security Council takes the necessary measures. This placement demonstrates the integration of self-defence into the UN collective security’s legal framework. In fact, during the drafting of the Charter, the inclusion of self-defence within Chapter VII was intended to emphasise that the use of force in self-defence prior to Security Council action did not constitute a beach of the peace rather than a non-breach of Article 2(4).⁷⁵

The ICJ jurisprudence clearly reflects the view that Articles 2(4) and 51 are normatively, methodologically and structurally autonomous. The Court has applied the rule on self-defence directly to the facts under consideration by examining whether its conditions – armed attack, necessity and proportionality – were satisfied without assessing whether the alleged use of force prima facie violated Article 2(4).⁷⁶ By contrast, if self-defence were an exception to Article 2(4), methodologically the Court would first have had to decide whether the conduct in question fell within the prohibition of Article 2(4), before assessing whether it fell within the exception. However, the ICJ does the reverse, as demonstrated in the Nicaragua and Armed Activities cases. It is only after rejecting a self-defence claim that it examines other related rules.⁷⁷ As asserted by Judge Tomka, ‘a lawful exercise of the right to self-defence cannot constitute a breach of any relevant article of the United Nations Charter (in concreto, Article 2(4)), and there would be no point in analysing the latter. Only once the Court concludes that “the legal and factual circumstances for the exercise of a right of self-defence … were not present” …, is it incumbent upon it to consider, and to make findings on, the prohibition of the use of force’.⁷⁸

A final point against construing self-defence as an exception to Article 2(4) is that, if it were an exception, it would be subject to restrictive interpretation;⁷⁹ this is not the case with self-defence, which is interpreted as a primary and autonomous rule according to its own purpose. Moreover, the current expansive interpretation of self-defence would risk entirely undermining the content of the prohibition of the use of force. In effect, self-defence would become the source of its content, contrary to the principle that exceptions should not undermine the general rule.⁸⁰

4.4. Self-defence is not a legal sanction

Self-defence as a primary rule is not a legal sanction to a prior breach of an international legal obligation, namely the prohibition of the use of force.⁸¹ By decoupling self-defence from Article 2(4) and not treating it as a sanction, self-defence overcomes the hurdle posed by the fact that NSAs are not recognised subjects of international law and therefore not bound by the prohibition of the use of force.⁸² Consequently, self-defence may be exercised in response to armed attacks by non-state actors, even though such actors cannot themselves violate Article 2(4).

That having been said, there are approaches which conceived of self-defence as a sanction. For instance, Ago regarded self-defence as a sanction against wrongfulness and therefore as a secondary rule, although this view was not universally shared within the ILC.⁸³ As he explained, ‘[a]cting in self-defence means responding by force to forcible wrongful action carried out by another; and the only reason why such a response is not itself wrongful is that the action which provoked it was wrongful’.⁸⁴ Ago’s conception rested on a vision of the international legal system as one of strict correspondence between obligations and responsibility, thereby giving primacy to the obligation not to use force. From this perspective, the use of force in self-defence could only be explained as a sanction for breach of that obligation. This approach also resonates with Bowett’s view that self-defence constitutes a response to the delictual conduct of the perpetrator state.⁸⁵ Its understanding as a sanction, and indeed a forcible one, also calls to mind Kelsen’s theory of war as a sanction for illegality, a theory that has long been rejected.⁸⁶

Construing self-defence as a sanction is not supported by the history, codification, nature or aims of the right of self-defence. The UN Charter does not characterise self-defence as a sanction⁸⁷ or codify the law of state responsibility; rather, as noted, it sets out the primary rights and duties of states. Interpreting self-defence as a sanction will thus be inconsistent with its nature as a primary rule, as sanctions are secondary rules that address the legal consequences of violations of primary rules.

Furthermore, the purpose of self-defence reinforced by the requirements of necessity and proportionality is to enable states to repel and deter armed attacks,⁸⁸ not to punish the aggressor, enforce international law, or hold the author of the attack legally accountable – functions that belong to sanctions. Construing self-defence as a sanction would therefore distort its normative and methodological framework. More specifically, the requirement of necessity would be displaced, as a sanction functions as an automatic legal reaction to a wrongful act, whereas proportionality would assume a punitive dimension extending beyond the defensive purpose of self-defence.

Construing self-defence as a sanction will also normatively and methodologically collapse self-defence into countermeasures, which belong to secondary rules. As is well known, countermeasures are reactions to a prior wrongful act intended to induce compliance with international obligations and to secure cessation and reparation.⁸⁹ The similarities between the conception of self-defence as a sanction and countermeasures was acknowledged by the ILC itself but it sought to distinguish them by arguing that self-defence responds to a particular violation of international law – namely, the prohibition of the use of force – and aims to prevent the wrongful act from achieving its purpose, whereas countermeasures are ex post reactions to any breach of international law.⁹⁰ This distinction, however, is far from convincing.

If self-defence were to be seen as a special sanction in response to violations of the prohibition of the use of force while countermeasures were treated as reactions to other breaches of international law obligations, this would be inconsistent with the methodology of the law of state responsibility, which establishes general and uniform rules governing the legal consequences of all internationally wrongful acts.

Furthermore, one would expect that the situation triggering the sanction (in this case self-defence) would coincide with the situation constituting the violation. There is no normative, methodological or logical explanation thus for limiting self-defence as a sanction to only ‘grave’ uses of force rather than to any use of force, or for gradating violations of the same rule (the prohibition of the use of force) and attaching different sanctions. Contrary, however, to the ILC’s muddled reasoning, the ICJ has consistently maintained a clear distinction between self-defence and countermeasures, holding that violations of the prohibition of the use of force give rise to peaceful countermeasures, not to self-defence.⁹¹ The Court has also consistently affirmed that an armed attack triggers the right of self-defence, not countermeasures.

A further difficulty with treating self-defence as a sanction would be that collective self-defence would then be equated with countermeasures taken by third states for violations of erga omnes obligations, in this case the prohibition of the use of force.⁹² That said, it remains unsettled whether third state countermeasures are available and, even if they are, forcible countermeasures are clearly precluded.⁹³ Moreover, collective self-defence concerns assistance rendered to the victim of an armed attack to enable it to defend itself, irrespective of whether the assisting state’s own rights have been injured,⁹⁴ which is a critical condition of countermeasures. In addition, collective self-defence requires a request from the victim state,⁹⁵ whereas countermeasures by third states do not require such request.

Finally, if self-defence were conceived as a sanction for breach of the obligation not to use of force, it could not extend to imminent attacks, as in such cases no actual breach has yet occurred. Yet existing international law recognises the possibility of self-defence against imminent attacks, notwithstanding ongoing debates about the scope of imminence. The situation might have been different if self-defence were also regarded as a sanction for the threat of force under Article 2(4). However, even the proponents of the sanctions approach have not advanced this view, and Ago himself adopted a restrictive view on both the scope of the right of self-defence and the prohibition of the threat or use of force.⁹⁶

4.5. Self-defence is a reaction to an armed attack as a factual occurrence

Self-defence is triggered by a factual occurrence, an armed attack. An armed attack is a use of force but in factual not legal terms. When the ICJ, for instance, characterised an armed attack as a ‘grave’ use of force with reference to its scale and effects, it was merely describing the factual attributes of an armed attack, but not defining an armed attack in legal terms within the framework of Article 2(4) of the UN Charter.⁹⁷ This reading of ‘armed attack’ is reinforced by the wording of Article 51, which is descriptive of its fact-based trigger and not attributive, as it does not qualify an armed attack by reference to its perpetrator. It also accords with the right of self-preservation, the historical source of the right of self-defence, which entitled states to use force to safeguard their existence and essential interests against any danger, regardless of whether the threat came from another state or from private actors.

It follows from this that, in factual terms, both states and NSAs are capable of committing an armed attack,⁹⁸ contrary to the ICJ view that only states can commit an armed attack for the purposes of Article 51.⁹⁹ As India observed during the Arria Formula debates, ‘the source of the attack, whether a state or a non-state actor, is irrelevant to the existence of the right of self-defense’,¹⁰⁰ and in the same vein the African Union’s Non-Aggression and Common Defence Pact (2005) includes in its definition of acts of aggression armed force by non-state actors.¹⁰¹

This interpretation of ‘armed attack’ challenges the interstate construction of the right of self-defence and affirms the normative separation between Articles 2(4) and 51 of the UN Charter. At the same time, it maintains the separation between the jus ad bellum and state responsibility by decoupling the existence of an armed attack from attribution. It also supports the earlier point that self-defence is not a legal sanction triggered by a prior breach of international law.

4.6. Self-defence targets the author of the armed attack

It goes without saying that self-defence action must be directed against the author of the armed attack, whether that author is a state or a non-state actor. This prevents normative gaps by ensuring that states are not left without lawful protection from armed attack because they originate from non-state actors.

However, where the state in the territory of which the self-defence action occurs is not the author of the armed attack, the use of defensive force against non-state actors may nevertheless infringe that state’s sovereignty. Indeed, as explained previously, territorial states, as well as third states, have condemned such actions as a violation of sovereignty.¹⁰²

In such cases, self-defence as a circumstance precluding wrongfulness and a secondary rule comes to the fore. It is therefore important to clarify the relationship between self-defence as a primary rule and its operation within the framework of secondary rules, and to explain the distinct legal implications that arise.

4.7. Self-defence as circumstance precluding wrongfulness

Self-defence as a circumstance precluding wrongfulness operates within the framework of the law of state responsibility. According to Article 21 ARSIWA: ‘The wrongfulness of an act of a State is precluded if the act constitutes a lawful measure of self-defence taken in conformity with the Charter of the United Nations’. The function of self-defence as a circumstance precluding wrongfulness is thus different from self-defence as a primary rule, although it is dependent on the primary rule. More specifically, its function is to excuse the responsibility of the state acting in self-defence for collateral violations of international law obligations committed in the course of exercising its primary right of self-defence and occasioned by its exercise.¹⁰³ Applied now to the case at hand, it will preclude the wrongfulness of the violation of the territorial state’s sovereignty committed in the course of exercising its right of self-defence against NSAs.¹⁰⁴

There are, however, two issues that require further consideration. The first issue is whether the use of defensive force against non-state actors in another state’s territory may also constitute a violation of the prohibition of the use of force vis-à-vis the territorial state and, if so, whether Article 21 can excuse its wrongfulness.

However, this question remains largely academic in that, as noted earlier, affected states have generally characterised such actions as infringements of their sovereignty rather than as violations of the prohibition of the use of force. This attitude seems to echo the US approach to the British incursions on its territory in the Caroline case, which were condemned as violations of its sovereignty.¹⁰⁵

Another consideration which may explain states’ reluctance to characterise such uses of force on their territory as breaches of Article 2(4) is the absence of hostile intent on the part of the defending state. As noted earlier, states often stress the fact that their actions are directed exclusively against NSAs and not against the territorial state.¹⁰⁶ Earlier writers also distinguished between war and self-defence in situations where a state used force against individuals or groups located within another state, arguing that in contrast to war, the self-defence action lacked hostile intention and was not directed against the state or its people and objects.¹⁰⁷

Although there is no consensus currently as to whether an unlawful use of force under Article 2(4) requires hostile intent, support for such a requirement can be drawn from the text of Article 2(4), which prohibits the use of force ‘against the territorial integrity or political independence’ of a state.¹⁰⁸ On this view, if there is no intention to use force or no intention to coerce the other state’s will when using force, such use of force would fall outside the scope of Article 2(4), though it may still form an unlawful breach of sovereignty.¹⁰⁹

If the defensive action is nevertheless considered a use of force against the territorial state and a breach of Article 2(4), the legal question is whether Article 21 ARSIWA can excuse its wrongfulness. This raises the question of the exculpatory scope of self-defence as a circumstance precluding wrongfulness. Article 21 does not excuse breaches of obligations of ‘total restraint’¹¹⁰ but these obligations do not cover the prohibition of the use of force; they refer to ‘intransgressible’ humanitarian law principles or to non-derogable human rights norms.¹¹¹ What may be relevant in this case is Article 26 ARSIWA, according to which the wrongfulness of violations of jus cogens norms is not precluded. However, it is not established yet that the prohibition of the use of force possesses this status in contrast to the prohibition of aggression.¹¹²

A related issue is whether such a defensive action might be characterised as an armed attack against the territorial state, triggering its right of self-defence. There are very few examples where states made such a claim, but they concerned instances of long presence of foreign troops on their territory or the occupation of parts of their territory.¹¹³ That said, these states have not taken any defensive action. A further consideration militating against such a construction is the absence of animus aggressionis on the part of the defending state, as illustrated by the Oil Platforms case in which the ICJ held that there was no armed attack in the absence of Iran’s intent to attack the specific vessels.¹¹⁴

The second issue concerns the question of whether self-defence as a circumstance precluding wrongfulness has erga omnes application because Article 21 appears to regulate only the legal relationship between the attacking and defending state, whereas in the present case the territorial state is a third party to this relationship.

Although the ILC left open the question of the effects of self-defence as circumstance precluding wrongfulness on third states,¹¹⁵ it may be argued that Article 21, as a secondary rule, is of general application, consistent with the ILC’s overall approach to the law of state responsibility. It therefore applies to any incidental self-defence breach of obligations owed by the defending state to any other state. It should also be recalled that Article 21 is dependent on, and parasitic upon, the primary rule of self-defence. It does not substitute for, expand or narrow the substantive scope of the right of self-defence, which is determined exclusively at the level of the primary rule. Accordingly, it may be argued that Article 21 precludes the wrongfulness of incidental breaches of obligations owed to other states in so far as they are linked to the lawful exercise of the right of self-defence, however this right is defined by the primary rule.

5. Conclusion

This article has addressed the question of how states can justify the use of defensive force against non-state actors on the territory of other states. It first mapped out the current legal landscape and concluded that it leaves many normative, methodological and practical questions unsettled. It then explained the morphology of the right of self-defence. It argued that self-defence is a primary rule of international law permitting the use of force against an armed attack as a factual occurrence regardless of its perpetrator. It also argued that self-defence is not an exception to Article 2(4) or a legal sanction. This led to the conclusion that self-defence can be exercised against NSAs. If, however, the self-defence action takes place on the territory of another state that is not the author of the armed attack but is where the NSA is located, self-defence as a secondary rule and, more specifically, as a circumstance precluding wrongfulness as stipulated in Article 21 ARSIWA will excuse the responsibility of the defending state for the incidental breach of that state’s sovereignty in the course of its self-defence action. The article thus offers a deeper understanding of the nature, features, role and place of self-defence within the jus ad bellum regime, which provides a more coherent legal justification for using defensive force against NSAs.