Lieber Studies Non-State Actor Attribution Volume Series – Shadow Warfare, State Attribution and Substantial Involvement

Editors’ note: This post appears as an entry in a series dedicated to Dr Jennifer Maddocks’ monograph, Attribution of Non-State Actor Conduct during Armed Conflict, appearing in the Lieber Studies book series with Oxford University Press.

When is a State “the origin or cause of” the actions carried out by a Non-State Actor (NSA)? NSA attribution is a crucial but challenging aspect of hostilities that occur in shadow warfare. The term “shadow warfare” is commonly applied to a wide range of clandestine and covert operations that take place across the whole spectrum of armed conflict. With a shadow defined as “an area of darkness, caused by light being blocked by something,” these operations are conducted with the intention of being hidden from public scrutiny. As Sean McFate notes “[s]hadow wars are armed conflicts in which plausible deniability, not firepower, forms the center of gravity” (p. 199). Operations in the shadows are an inherent part of hybrid warfare or gray zone conflict, and other irregular operations that are adjunct to more conventional hostilities.

Many States expend significant effort developing and employing specialized military, intelligence agencies and police forces, paramilitary units, contract personnel, and even ordinary civilians to operate in the shadows to influence and shape world events, as well as conduct hostilities (see here). Some States rely not only on proxy groups, but even criminal gangs to carry out espionage, sabotage, and assassination in the shadows of more traditional conventional conflict. For example, Russian intelligence services are alleged to have made extensive use of NSA groups and individuals (see here, here, and here) to conduct such surreptitious activity in Europe.

In Jenny Maddocks’s new book, Attribution of Non-State Actor Conduct During Armed Conflict, the complexity of the relationship between States and various NSAs is highlighted with case studies that have occurred in Syria, Ukraine, and the Democratic Republic of the Congo. The Syrian example includes reference to Hezbollah, which “became involved in the Syrian conflict at Iran’s instigation” (p. 38). As I outlined in a 2022 article, Hezbollah is just one of the Iranian proxies (e.g. Hamas, Palestinian Islamic Jihad, the Houthis, Iraqi Shi’ite groups) involved in its hostilities against Israel and/or the United States whose acts may be attributed to Iran.

This post explores how the uncertainty surrounding international legal issues such as NSA attribution facilitates the masking of covert and clandestine operations by State proxies. Maddocks’s critique of the law attributing the actions of NSAs to States and considers her suggested approaches for addressing gaps in accountability. It also addresses the issue of holding States accountable for their substantial involvement in the actions of NSAs.

The Legal Shadows

The ability to hold States accountable for NSA conducted clandestine and covert operations is impacted by a lack of consensus as to what law applies. Even where there is consensus there is often disagreement concerning its application. Many of these operations are obscured by the shadows created because of legal uncertainty regarding the scope and substance of international law. States rely on this legal ambiguity to mask and even facilitate the conduct of hostilities. While international law governs the recourse to war by a State and regulates hostilities during armed conflict there remain numerous ambiguous areas where that law has unequal or unclear application. Regarding the applicable law, there is a significantly more detailed treaty regime applicable to inter-State warfare than in its non-international counterpart. There have been attempts to specifically address indirect warfare involving NSAs, such as through Additional Protocols I and II to the 1949 Geneva Conventions, but this treaty law has not been universally adopted. Further, the content and scope of customary international law that might fill any gaps is itself often the subject of considerable debate within the international legal community.

International consensus on the law has been particularly difficult to obtain regarding operations below the level of conventional war. This includes questions regarding the thresholds for the commencement and termination of armed conflicts, and the gravity standard to be applied when assessing if an armed attack has occurred such that a State may respond in self-defence. An all-too-common feature of international law is that legal issues are subject to conflicting binary interpretations. This can be seen in the “restrictionist” and “expansionist” interpretations of State self-defence; the different non-international armed conflict criteria of protracted armed violence and organized armed groups, or the totality of the circumstances approach (Fighting at the Legal Boundaries, pp. 363-78); and the overarching, or limited interpretations of the continuing application of the principles of self-defence during an armed conflict (p. 58-69).

The assessment of when Russian activities in Europe may raise to the level of an armed attack under Article 5 of the North Atlantic Treaty are shrouded in considerable legal and policy uncertainty. It has been observed that Article 5 “is consistent with Article 51 of the United Nations Charter, which recognises that a state that is the victim of an armed attack has the inherent right to individual or collective self-defence, and may request others to come to its assistance.” However, Russia continues to take advantage of the legal uncertainty about when an armed attack occurs, or an armed conflict is established to carry out low level aggression using NSAs against European NATO States. A significant challenge in linking Russia to such action to enable a measured response is one of attribution. Without such attribution the actions of the NSAs are largely left to be dealt with as law enforcement matters.

The Law of NSA Attribution

The attribution of NSA activity to a State is another legal issue contributing to the darkening of the legal shadows. Questions concerning NSA attribution in its widest sense (linkage to a State) can arise in the context of breaches of humanitarian law, the categorization of conflict, and in a less explored context regarding armed attacks under the law of State self-defence. Maddocks provides a timely and compelling analysis of NSA attribution in focusing particularly on two widely recognized interpretations of the legal tests. The “effective control” test set out in the 1986 International Court of Justice Nicaragua Case regarding breaches of international humanitarian law, and the “overall control” standard established by the International Tribunal for the Former Yugoslavia (ICTY) in the 1999 Tadic Appeal Case (para. 131) dealing with conflict categorization.

State attribution, which is viewed as rule of customary international law, is best articulated in Article 8 of the 2001 Draft articles on Responsibility of States for Internationally Wrongful Acts prepared by the International Law Commission (ILC). That Article states, “The conduct of a person or group of persons shall be considered an act of a State under international law if the person or group of persons is in fact acting on the instructions of, or under the direction or control of, that State in carrying out the conduct” (emphasis added). As Maddocks notes, States party to an armed conflict are bound by “primary obligations” (e.g. the prohibition against indiscriminate attacks) and it is a “secondary law” of State responsibility that determines whether when the violation of the law engages State responsibility, as well as the legal consequences of the violation (p. 57).

The degree of control required to attribute the acts of an NSA to a State continues to be the subject of considerable debate. It is an issue that remains unresolved even though the effective control standard has been suggested to be the lex lata (“the law as it is”) (p. 180-81). Significantly, the effective control test referred to in the Nicaragua Case has been interpreted very narrowly. An approach reinforced in the 2007 ICJ Bosnian Genocide Case which required “that the State’s instructions were given, in respect of each operation in which the alleged violations occurred, not generally in respect of the overall actions taken by the persons or groups of persons having committed the violations (emphasis added)” (para. 400). As Maddocks notes, “the conduct is attributable to a State only if its officials exert tactical control over the specific acts at issue at the time of their commission” (p. 187). Her reference to the tactical level of war highlights a particular strength she brings to her analysis of NSA attribution, the perspective of a practitioner.

The problem resulting from a stringent effective control test is that a “normative safe zone” (p. 194-95) is created in which States appear to be free to rely on NSAs to do things they themselves do not want to be held accountable for. Another author calls it a “responsibility vacuum” (de Wet, p. 109). Problematically, it is likely to prove impossible to collect the detailed evidence necessary to establish the requisite level of control for each operation (Maddocks, p. 194). Perhaps unwittingly, the narrowness of the effective control standard contributes significantly to the creation of legal shadows within which States rely on NSAs to carry out clandestine and covert operations without being accountable for their actions.

The “overall control” test established in the 1999 Tadic Appeal Case specifically rejected the effective control test in a case dealing with conflict categorization. The Tribunal stated the effective control test was “unconvincing … based on the very logic of the entire system of international law on State responsibility” (para. 116) and “is at variance with international judicial and State practice” (para. 124). It is noteworthy that the tribunal favoured practice over theory. Rather than require specific direction for each operation the overall control test is more general in nature. It is established when a State “has a role in organising, coordinating or planning the military actions of the military group, in addition to financing, training and equipping or providing operational support to that group” (para. 137).

Confronted with the two different attribution standards, the ILC commentary in the 2001 Draft Articles on State Responsibility clearly preferred the effective control test. The two tests were distinguished on the basis that the ICTY was dealing with individual criminal responsibility rather than the issue of State responsibility (p. 48). However, the commentary did not discount there could be two different tests for attribution, one for establishing State responsibility and the other for conflict categorization. This was also acknowledged in the Bosnia Genocide Case (para. 405). Such siloing of attribution creates complexity and potentially increases uncertainty.

If the ICJ approach of effective control is seen as “the law as it is” it is one of very narrow application. A particular weakness of the effective control test is that demanding a tactical level of control requires a high degree of detailed knowledge while sacrificing an ability to thwart efforts by States to operate under the radar of legal accountability. Importantly, this test does not reflect how command is often exercised on the contemporary battlefield. Command, even for regular military formations, is likely to be carried out in a decentralized fashion (e.g. mission command) with limited top-down direction. Further NSAs, such as those operating in occupied territory, often adopt a devolved cellular organization for security reasons. It is an approach that purposely limits communication between cells.

How much weight should be put on the ICJ or the ILC approach favouring effective control? One of the challenges in interpreting international law is the lack of a court system producing binding precedent in the same manner as domestic courts. The void is often filled by experts sitting on international commissions or committees that provide opinions on various aspects of international law. However, as Marco Milanovic noted “The ILC’s work is certainly not gospel and its authority, as well as that of the ICJ for that matter, does not place it or the ICJ beyond criticism” (p. 560). As a Canadian judge commented regarding UN Committees, “While they clearly reflect the views of knowledgeable individuals, they do not reflect the current state of international law, but more the direction that those groups believe the law should take in the future” (para. 239). The rulings of the ICJ and the opinions of committee members such as those serving on ILC are clearly influential, but they do not necessarily end the debate.

As an alternative approach, the overall control test is intended to be a standard of more general application. One that focuses on practicality in seeking to enhance NSA accountability more broadly, and by doing so reign in the acts of States. It is evident from the various International Committee of the Red Cross (ICRC) Commentaries dealing with Common Article 2 to the 1949 Geneva Conventions published between 2016 and 2025 (see here, here, here, and here) that other experts have not fallen into line with the ICJ and ILC position. The ICRC clearly opted for the broader “overall control” test. As Maddocks notes they apply the same test to “both conflict classification and attribution under the law of State responsibility” (p. 178-79). The 2025 ICRC Commentary (para. 343) indicates that test might prevent a State from “hiding behind the proxy’s veil in order to avoid its obligations and international responsibilities under international humanitarian law” and make it possible to challenge a claim by a State the NSA was not acting on its behalf. It is clear the ICRC is concerned about a legal shadow being cast over the actions of State proxies.

However, the overall control test presents its own challenges. Maddocks points out “the use of the overall control test as a rule of attribution potentially stretches the links that must exist between the State and the NSA too far” (p. 196). A State may meet the criteria of the overall control test but have no knowledge of certain activity such as the recruitment of child soldiers. In other circumstances a State may employ an NSA on an ad hoc basis thereby avoiding the more general indicia of control (e.g. equipping, financing, planning military activity) reflected in the overall control standard (pp. 196-97). She notes that the ICRC approach suffers from these same flaws (p. 180).

So, forty years after the Nicaragua Case, and a quarter century after the Draft Articles on State Responsibility were developed, much remains unresolved concerning how attribution for wrongful acts by NSAs is to be assessed. This is where Maddocks’s analysis in Attribution in Non-State Actor Conduct is particularly useful. She suggests a new test based on the concept of “command responsibility.” In doing so Maddocks provides a means of functionally bringing the effective control and overall control tests closer together in terms of their application. Command is intimately linked to the concepts of “control” and “responsibility,” with the former being defined as “control over someone or something and responsibility for him, her, or it.” As Jenny Maddocks notes, “if a State is in a position of authority over an NSA such that it can control the constituent acts performed in the course of an operation, or to order their cessation, this should be sufficient to attribute the NSA’s conduct to the State” (p. 198-99).

The suggested test “is analogous to the control exercised by commanders over their troops” in introducing the idea that attribution of NSA conduct is established where “the State had the capacity to exert its control and the State’s organs knew or ought to have known of the potential legal violations” (p. 199). An analogous situation is the superior responsibility test of knowing or consciously disregarding information clearly indicating wrongdoing would extend accountability to the participation of civilian officials (see Ronen).

Maddocks’ test falls in the middle ground between effective and overall control (p. 204), although it is ultimately one of more general application. A key strength is that it focuses on the “human link” between the conduct in potential violation of international law and the artificial entity of the State” (p. 58). She does acknowledge that a more relaxed interpretation of the attribution rules “would not close the accountability gap entirely” as “States would still be able to provide support to NSAs, such as intelligence, funding, arms, and ammunition without bearing responsibility for their contributions to the harm that may result” (p. 214). She suggests applying the Article 16 Draft Articles on State Responsibility prohibition against providing aid and assistance to States carrying out an internationally wrongful act (p. 219-20) by analogy to States supporting NSAs. However, she also notes there remain interpretive challenges in doing so (p. 223-25).

The “Substantial Involvement” Test

Jenny Maddocks’s analysis of NSA conduct is not limited to Article 8. She also addresses the applicability of primary international legal rules such as the prohibition on the use of force by States through “forcible acts committed by NSAs.” She suggests “it is the State’s own organs’ conduct in assisting the NSA that amounts to a use of force that is attributable to the State based on the rule reflected in Article 4” of the Draft Articles on State Responsibility. Here, attribution is dealt with in a narrow classic interpretation of the term (presumably under Article 8) rather than in the more general context of a State being the origin or cause of the use of force. The example provided is Iranian support provided to Hamas. Maddocks suggests this involvement would be insufficient to lead to attribution, but “might nevertheless have violated the prohibition on the use of force through its provision of armes and other assistance to Hamas.” She also notes that the ICJ has been more willing to find a violation of a primary international law norm than “than attribute private conduct to a State” (p. 226). Interestingly both Articles 4 and 8 rely on a similar “human link” between the State and the activity carried out by the NSA to establish attribution.

There is another primary norm-based accountability issue that has arisen in contemporary conflict. In Israel’s outline of the key factual and legal aspects of the Iran-Israeli hostilities of June 13-14, 2025, it is stated regarding Iranian proxies that, “Iran has consistently been substantially involved in acts of armed force carried out by these groups in their conflicts against Israel… .” In terms of international law rules governing the use of force, Iran’s substantial involvement in acts of armed force carried out by these groups legally constitutes an armed attack by Iran against Israel. (p. 11-12). Like the effective control test, the “substantial involvement” standard is established in the Nicaragua Case where an armed attack is understood to not only include the sending of armed bands, etc., to carry out attacks of sufficient gravity but also the State being substantially involved therein (para. 195). The Israeli substantiation references Iranian military commanders being located within Hezbollah command and control centers, public acknowledgment of Iran’s role by Hezbollah, Hamas and Palestinian Jihad, and Iran’s involvement in the planning of widescale invasions (p. 12-13).

To date the substantive involvement test has attracted limited interest from the international law community (e.g. see here, here, here, and here). Attributing attacks by NSAs to a State has historically been tied to whether a threatened State could respond under the right of self-defence. A highly restrictionist view is it cannot, although the events of 9/11 led to most international lawyers abandoning that interpretation (Kretzmer, p. 273). Interestingly, proponents of the more restrictive approach (A Plea Against Abusive Invocation of Self-defense) argue self-defence can only be exercised where “acts of war perpetrated by a terrorist group can be attributed to the State, or by virtue of a substantial involvement of that State in the actions of such groups.” What then constitutes substantial support?

In a 2025 article Jenny Maddocks suggested that considerable ambiguity surrounds the precise meaning of substantial involvement. In her view, regarding the attacks of October 7, 2023, Iranian support to Hamas violated the prohibition of the use of force but there was some doubt it constituted Iranian involvement in an armed attack (p. 743). However, elsewhere it has been suggested (see Corten, p. 470) “there would be nothing to prevent interpreting the criterion of ‘substantial involvement’ as covering massive support for armed bands operating from the territory of the State in question, without the State actually participating in the armed actions as such.” Another author notes that State practice implies “state support in the form of weapons, finances, training, logistical support—or any combination of these—to non-state actors attacking another state could amount to the type of ‘substantial involvement’ in an armed attack” (de Wet, p. 103). It has also been suggested that under this standard the State should be seen as a facilitator rather than necessarily a dominate power behind the NSA (Tsagourias, p. 816). The substantial involvement standard is relied on to effectively broaden a narrow interpretation of attribution in the Nicaragua Case (para. 115).

The definition of “substantial” as “large in size, value or importance” suggests a much broader test than the narrow effective control standard originating in the same ICJ opinion. Significantly, Israeli reliance on criteria such as the locating of Iranian authorities within the command and planning structures of its proxies suggests interpretation much closer to the overall control standard. A link between substantial involvement and overall control has been indicated elsewhere (Ruys, p. 418). The focus appears to be on command, planning, and support involving the strategic and operational levels of war rather than requiring tactical level direction associated with effective control.

Conclusion

Jenny Maddocks’s alternative command (or superior) responsibility-based approach offers a very useful and importantly practical framework within which to help interpret what substantial involvement means. Applying this approach to all three tests: effective control, overall control, and substantial involvement could significantly limit legal ambiguity. This common interpretive approach would help reduce the normative safe zone within which States rely on NSAs to carry out operations they do not want to be publicly associated with. As such it shines a necessary light into the legal shadows that mask many covert and clandestine operations that States want hidden from public scrutiny.

***

Ken Watkin served for 33 years in the Canadian Forces, including four years (2006-2010) as the Judge Advocate General.

The views expressed are those of the author, and do not necessarily reflect the official position of the United States Military Academy, Department of the Army, or Department of Defense. 

Articles of War is a forum for professionals to share opinions and cultivate ideas. Articles of War does not screen articles to fit a particular editorial agenda, nor endorse or advocate material that is published. Authorship does not indicate affiliation with Articles of War, the Lieber Institute, or the United States Military Academy West Point.

Photo credit: IDF