Antonios Tzanakopoulos is a DPhil Candidate at St Anne’s College, Oxford. He has an LLM from New York University Law School. During the 57th session of the International Law Commission (2005), he was research assistant to Professor Giorgio Gaja, Special Rapporteur on the Responsibility of International Organizations. His Oxford thesis is on the responsibility of United Nations for wrongful non-forcible measures by the Security Council.
A recent article by White and MacLeod in the EJIL (EU Operations and Private Military Contractors: Issues of Corporate and Institutional Responsibility) discusses, in part, the attribution of conduct of Private Military and Security Companies (PMSCs) to an International Organization (IO) in the context of a peacekeeping operation (PKO). The authors take issue with Article 5 of the International Law Commission’s (ILC) Draft Articles on the Responsibility of International Organizations (DARIO) and the high threshold of “effective control” that this provision requires for attribution of conduct to an IO. However, Article 5 DARIO is specifically adopted to deal with the attribution to an IO of the conduct of a military contingent belonging to a State, and does not apply in the case of attribution of PMSC conduct. It is Article 4 DARIO that applies in such a case. Paragraph 1 of that provision states that:
The conduct of an organ or agent of an international organization in the performance of functions of that organ or agent shall be considered as an act of that organization under international law whatever position the organ or agent holds in respect of the organization.
This being the case, attribution of conduct by a PMSC hired by an IO to the IO is, ostensibly, automatic and thus much easier than attribution of PMSC conduct to a State. In the latter case one would have to argue basically either that the PMSC exercises elements of governmental authority or that it is directed or (effectively) controlled by that State (see the discussion here, here, here, and here). Could it in fact be so, and how can this difference be explained?
Both States and international organizations are abstract (fictional) entities, which cannot of course act in the physical world. Rather, conduct always originates in individuals, i.e. natural persons. The ‘normative’ operation of attribution is thus required to bridge the gap between the physical actor and the subject of international law. In the law of State Responsibility, as codified in the ILC’s Articles on the Responsibility of States for Internationally Wrongful Acts (ARSIWA), the conduct of acting individuals is attributed to the State automatically when these individuals are connected to the State through an institutional or organic link (4 ARSIWA, cf 7 ARSIWA). Absent such an institutional or organic link, acts of individuals are only exceptionally attributed to the State when a control link can be established (8 ARSIWA).
This simplified description of the basic avenues for attribution may disregard certain specific cases of attribution (such as those under 6 or 9 and 11 ARSIWA for example), but it sets out the fundamental distinction, which shall prove to be conceptually useful: if the acting entity is a State organ, the act is automatically attributable. Control over the conduct is not just presumed; it is irrelevant. If the acting entity is not an organ, then the only way in which the conduct can be attributed is on the basis of a showing that the specific conduct in question took place under the control of the State.
Similar considerations apply to attribution of conduct to international organizations. If an IO organ is acting, attribution is automatic; if not, a control link must be established. While international organizations rarely act in the physical world (in most cases they will be rhetorical organizations, merely passing recommendations, or they will act on the normative level, imposing obligations on States to act), when they do in fact act, that is to say when they do undertake certain operations, they will rarely act through their ‘constitutional’ organs, i.e. the organs identified in the constitutive instrument. The Security Council cannot for example undertake peacekeeping itself; neither can the General Assembly. They both lack any operational capacity of the sort. Peacekeeping must be undertaken by armed forces, and the UN does not possess any.
It is mainly for this reason, namely because of the relative scarcity of IO organs with any operational capacity (or because of the rudimentary organic makeup of IOs), that the general rule of attribution of conduct to IOs does not solely refer to ‘organs’ of the IO, as is the case with the general rule of attribution to States. Rather, it also refers to ‘agents’ of the IO, a term given a special meaning by the ICJ and the ILC. According to the ILC (see here at 104 ), the term is to be understood-like the ICJ understands it-’in the most liberal sense’ (citing the Reparation for Injuries at 177). An agent thus is not necessarily an ‘official’ but ‘any person through whom [the organization] acts’ (see the ILC here at 106 , and also Reparation for Injuries at 177), i.e. there is no requirement of any official link in this case for automatic attribution to take place. Any conferral of power upon someone to act ‘on behalf’ of the organization would suffice to establish the requisite organic link.
If this were all, it should be very easy to attribute the acts of a national armed forces contingent participating in a UN-commanded PKO to the UN: if anything, the organization is ‘acting through’ the national contingent, thereby making it its agent. Power to act ‘on behalf’ of the UN will have been granted through the relevant UN decision to establish the PKO, the Status of Forces Agreement, and even on a semantic level by the wearing of a UN ‘blue helmet’. This would mean then that all acts of the contingent, an agent of the UN, in the context of the PKO, are directly and automatically attributable to the UN.
However, there is an element that causes substantial complication here, or at least the ILC thought so. The national contingent remains an organ of the lending (or home) State, and even more importantly that State retains some jurisdiction (thus: control) over the contingent (most prominently criminal jurisdiction and disciplinary powers). What blocks automatic attribution of its conduct to the IO in such a case is the operation of the general rule of (automatic) attribution in Article 4 ARSIWA: acts of the organ of a State are attributable to that State.
To respond to this predicament, the ILC adopted Article 5 DARIO. In accordance with that provision, the conduct of an organ of a State that is placed at the disposal of an international organization-but is not fully seconded to it-is only attributable to the organization if the latter exercises ‘effective control’ over that conduct. Otherwise (one would logically assume) it is attributable to the lending State (automatically, as the conduct of its organ). It is to overcome this attribution to the State that effective control over the State organ must be shown to exist. This distinction seems to imply that one has to choose the entity to which the conduct is attributable, i.e. that it can either be the organization or the State, but not both. The European Court of Human Rights (ECtHR) obviously adopted such an approach in Behrami (though implicitly: at  and , distinguishing Bosphorus) and was much criticized for it.
However, the ILC itself has not excluded the possibility of dual (or multiple) attribution of conduct to more than one State or international organization at a time (see its 2004 Report at 101 ). Why this shouldn’t be the case for a national contingent constituting an ‘agent’ of an international organization and an organ of the lending State at the same time is not entirely clear. But the more troublesome aspect of this provision is that it introduces the perennial debate over the threshold of factual control over an entity for purposes of attribution. It seems that the debate over that threshold will not die down however many times the ICJ confirms it as being that of ‘effective’ control (see Nicaragua at 65 [115 in fine] and Bosnian Genocide at -).
The near impossibility of ever proving such a high threshold of control is notorious, whether it is with respect to control of States over private individuals under 8 ARSIWA, or with respect to control of international organizations over ‘not fully seconded State organs’. It has been demonstrated countless times: in the jurisprudence of the ICJ (see Nicaragua at 62-65 - and Bosnian Genocide at -), in Tadić before the ICTY Appeals Chamber, which famously proclaimed the threshold to lie at ‘overall’ control (see here at  and ), in academic commentary, e.g. in attempts to establish attribution of PMSC conduct to States (see Lehnardt in idem & Chesterman [eds] From Mercenaries to Market  139; and Hoppe  19 EJIL 989), and finally in the unfortunate (in this respect at least) Behrami decision of the ECtHR, which did not at all manage to show the requisite level of control but rather re-positioned the threshold to that of ‘ultimate authority and control’ (see for comment Larsen  19 EJIL 509).
In their discussion of attribution of PMSC conduct to an international organization in the context of a peacekeeping operation, White and MacLeod struggled with the need of proving that the organization exercises effective control over the PMSC under Article 5 DARIO and advocated a lowering of the threshold of control as well, in-line with UN practice and along the lines of Tadić.
However, when an international organization employs a PMSC in the context of a peacekeeping operation, Article 5 DARIO is not called into operation. Since the PMSC is not an organ of any State, its actions are in the first instance attributable to no one. As such, there is no need to show any sort of control over the conduct to overcome some prima facie attribution, as is the case with the organ of a State employed by an IO.
Indeed, given that an ‘agent’ is ‘someone through whom the organization acts’, the mere existence of some sort of agreement between the PMSC and the IO should suffice for the establishment of the agency link. It is significant, if only at a semantic level, that recent academic commentary on the possibility of the UN employing PMSCs for PKOs has used the terms ‘agency agreement’ to describe the link between the IO and PMSCs (Patterson  13 JCSL 215 at 216, 221, 230 [emphasis added]). If this is in fact the case, the PMSC becomes an agent of the international organization without the added complication of anyone else retaining any control over it. Article 5 DARIO does not come into play, and attribution of PMSC conduct within the PKO context to the international organization is rather automatic under Article 4 DARIO. There is no requirement to show that the organization does in fact control effectively each and every instance of conduct.
This latter statement is important indeed. Attribution of conduct on the basis of an institutional or organic link encompasses all conduct of the organ or agent acting in official capacity (even if in excess of authority or contravention of instructions, unless in purely private capacity). This means that there is almost no need to look into the specific conduct-any conduct is automatically attributable (with the very rare exception of purely private conduct). Conversely, attribution of conduct on the basis of a control link, because exceptional, may take place only on a case-by-case basis, enquiring into the level of control over each and every specific instance.
In order to attribute the conduct of PMSCs to IOs employing them in the context of PKOs, the mere showing of a contractual link will suffice for the qualification of the PMSC as an IO agent. The attribution of the conduct of the agent (the PMSC) to the IO will then be automatic. There will be no need to show factual control of the IO over PMSC conduct, which in any case would be well-nigh impossible if the threshold is effective control. In this instance, at least, it appears that PMSC conduct is easier (than usual) to attribute to an entity clearly bound by international law.
There is a caveat here, however. In its 2004 Report (at 109 ) the ILC notes that private conduct under the direction or control of the IO will be attributable to the IO under 4 DARIO as the conduct of an agent; if a contractual link suffices to make a person or entity an IO agent, a fortiori effective control should turn them into an agent’ as well. In this way the Commission explains why there is no provision in DARIO corresponding to 8 ARSIWA: because it is simply not needed. What this does, however, is to spoil the neat scheme presented above that goes organic link-automatic attribution of all conduct; control link-ad hoc attribution of specific conduct. Under 4 DARIO, a person or entity may become an ‘agent’ also when the IO exercises effective control over them, in the absence of any contractual link. But then one would have to enquire into the specific conduct to determine whether it was directed or controlled. And it will only be that specific directed or controlled conduct that is attributable. How is thus one to know-in the case of an IO agent-whether all conduct is attributable on the basis of the organic link or only specific conduct may be attributable on the basis of a control link?
Presumably the ILC offers guidance in the following passage: ‘in exceptional cases, a person or entity would be considered, for the purpose of attribution of conduct, as entrusted with functions of the organization, even if this was not in accordance with the rules of the organization‘ (see here at 109  [emphasis added]). When the person or entity is someone to whom the organization has entrusted functions in accordance with its rules, the link is organic and attribution automatic. If, on the other hand, a person or entity seems to be someone through whom the organization is acting but there has been no overt delegation of functions (even through a contractual arrangement), then specific private conduct may be attributable if the organization directs or controls it. This is considered to be a de facto delegation of functions which may be presumed to be in violation of the rules of the organization. This distinction that must be drawn when applying 4 DARIO opens the door to a whole new set of issues that are not susceptible to abstract answers: what is in accordance with the IO rules and what is not? Would the employment of a PMSC constitute a delegation of functions allowed by the IO rules? If yes, then we’re back to automatic attribution, but if not, we’re exactly in the same situation as when trying to attribute under 8 ARSIWA or 5 DARIO.
Still it is argued that in most cases, the employment of a PMSC in the context of a PKO, if done formally through a decision of the competent organs of the IO and the conclusion of an agreement between the IO and the PMSC, will establish an organic link and lead to automatic attribution. The question thus remains why attribution of PMSC conduct to IOs should be easier than it is to States. The answer could lie in the fact that organizations are much less comprehensively structured than States. States have a multitude of organs through which they act, and in principle become automatically responsible only for the conduct of these organs belonging to their official structure. They do so not because they control these organs, but because they explicitly selected them in order to act through them with a prospect of permanence. There are very few such organs of IOs, not to mention that they have not been selected by the organization but rather by the States adopting the constituent instrument. The organization will thus need to act through other entities as well (if not mainly), particularly given that few constituent IO organs (if any) have any operational capacity. As such, it makes sense to extend automatic attribution to those entities the organization entrusts with functions and thus explicitly selects to act through, even without the prospect of permanence. Where the organization could not have used any of its organs, it used the PMSC. Where the State could have used its army, it used the PMSC. If it seems that this does nothing but offer States a much larger possibility for evasion of attribution, it is because it is true. But this is arguably a problem and an unfortunate evolution of the law of attribution to States and of the parallel policy preference for privatization. One should not be hasty to extend such a problematic situation to the law of attribution to IOs, which seems to be able to avoid the pitfall in principle.
One final note; something often forgotten is that attribution of conduct does not necessarily mean engagement of responsibility. However trite, this fact-when ignored-can cause significant confusion. Whether the conduct will constitute a breach of international law depends on the entity to which it will be attributed. E.g. certain conduct attributable to the UN may not be in breach of UN obligations under the Charter or general international law; but if the same conduct is attributable to a State it may be in breach of the State’s international obligations under a specific treaty. Even more importantly, even if conduct is attributable to two or more subjects of international law simultaneously (e.g. if the conduct of a contingent contributed to a PKO is simultaneously attributable to both the IO [4 DARIO] and the lending State [4 ARSIWA]), this does not necessarily mean that the responsibility of both subjects will be engaged so as to lead to joint and several responsibility: what may constitute a breach for the State may not constitute a breach for the IO and vice versa. This will depend on the primary obligations incumbent on the specific subject.
Further, when there is an international obligation on the State or the international organization to act to prevent certain (private) conduct, it is the omission to act to prevent the conduct that is attributable to the State or the organization (as a State or organization organ’s failure to act) and in breach of the international obligation, rather than the conduct in question itself. In such a case attribution is often self-evident, but a required step nonetheless.