Operationalizing Obligations to Prevent Mass Atrocities: Proposing Atrocity Impact Assessments as Due Diligence Best Practice

Abstract

Although there is wide agreement that there are jus cogens prohibitions against the commission of war crimes, crimes against humanity and genocide, there is significantly less clarity regarding obligations to prevent such atrocities. This paper explores the existing prevention framework and is intended to speak to those States that seek to take preventive steps, understand their legal obligations and, most importantly, to operationalize those obligations by taking concrete action. It outlines what is known regarding the duty to prevent each international crime under both treaty and customary international law, while elucidating the gaps that make these obligations so difficult to enforce. This paper argues that States’ existing obligations require them to examine their capacity to prevent, and that one form of examination, which should be adopted as a matter of best practice, is to undertake an Atrocity Impact Assessment, evaluating both a State’s current impact in a country or region, and the potential measures it could take to assist in averting mass atrocities.

1. Introduction

Given developments in the system of ‘early warnings’ of the last two decades, it is extremely unlikely that modern atrocities will occur without prior forecasting. To the contrary, each situation, today, where mass atrocities unfold tends to be preceded by multiple alarm bells. Well before the recent violence against the Rohingya minority in Myanmar, for example, the U.S. Holocaust Memorial Museum issued a 2015 report stating that the Rohingyas were at ‘grave risk of additional mass atrocities and even genocide’ (USHMM 2015). The same year, the report ‘Countdown to Annihilation’ warned that, of six possible stages,¹ the Rohingya appeared to be facing ‘the final two stages of genocide’ (Green et al. 2015: 16). And in August 2017—just four months before the Myanmar military begun its violent campaign in Rakhine State—an internal United Nations (UN) memorandum warned that the specific trigger of mass violence in Myanmar would be, in the immediate future, an attack by the Arakan Rohingya Salvation Army (ARSA). Precisely as predicted, shortly thereafter, an ARSA attack triggered a ‘bone-crushing massacre’ by the Tatmadaw (Lynch 2018). Yet, and despite such an abundance of precise warnings, the international community abjectly failed to take appropriate measures to prevent the violence.

Debates on the failure to prevent atrocities are often framed in moral and political rather than legal terms: this is not surprising given courts’ hesitation to enforce existing preventive obligations (The State of the Netherlands v. Stichting Mothers of Srebrenica et al.: para. 3.7.3), and what States themselves have characterized as a ‘lack of clarity’ as to what, precisely, international law requires of them when faced with warnings of atrocities on the horizon (UN General Assembly 2016: para. 3). Uncertainty surrounding the scope of States’ preventive obligations is compounded by the complex legal framework governing international crimes, and the fact that—as we further articulate below—any duties to prevent genocide, crimes against humanity and war crimes are derived from distinct legal sources.

Against this background, this article examines the legal nature of States’ obligations to prevent each ‘atrocity crime’,² and seeks to provide operational guidance for States seeking to implement their preventive responsibilities under international law. It considers the scope of preventive obligations under both treaty and customary international law, including existing points of disagreement, and provides an assessment of the most problematic ambiguities in this body of law. This article then proposes the use of a specific tool—Atrocity Impact Assessments—to assist States in operationalizing their preventive obligations. We submit that, even if not required as a matter of law, this policy tool can assist States to more readily translate early warnings of genocide, crimes against humanity and war crimes into meaningful preventive action.

2. The existing prevention framework

Following the UN General Assembly’s decision to endorse the Responsibility to Protect (R2P), genocide, crimes against humanity and war crimes have been grouped together (with ethnic cleansing) into the single conceptual category of ‘mass atrocities’ (UN General Assembly 2005), the commission of which ‘threaten[s] the peace, security and well-being of the world’ (Rome Statute: pmbl. para. 3). While politically and diplomatically useful, this conceptual category is fictitious from a legal perspective: genocide, crimes against humanity and war crimes occupy their own respective loci in international law, with each crime holding a unique legal history, status and trajectory. Indeed, the obligation to prevent such atrocities cannot be understood as a homogenous obligation that applies equally to all three international crimes, but is rather a complex interplay of legal responsibilities premised on a fragmented legal framework. In this section, we survey treaty and customary international law, and examine existing authority on relevant preventive obligations with a view to elucidate the current state of the prevention framework.

2.1 Prevention under treaty law

2.1.1 Prevention under the genocide convention

Of all the atrocity crimes, the legal obligations pertaining to genocide are most clearly articulated and established in international law. This is in no small part due to the widespread adoption and ratification of the 1948 Convention on the Prevention and Punishment of the Crime of Genocide (hereinafter Genocide Convention), and case law providing authoritative judicial interpretations of the obligations enshrined therein. As the UN International Law Commission (ILC) has clarified, the Convention does not contain a singular duty to prevent genocide, but actually ‘several elements relating to prevention’ (UN General Assembly 2019b: 54), including: a general preventive obligation (art. I), the obligation to enact national measures to give effect to the provisions of the Convention (art. V), and a provision to call upon competent organs of the UN to take appropriate preventive action (art. VIII). Below, we examine some key features of these preventive obligations as they relate to our argument.

2.1.1.1 The extraterritorial nature of the obligation

Among the most salient features of the obligation to prevent genocide is its extraterritorial nature. Although some States, most notably the United States, continue to object to the obligation being characterized as such (van Schaack: 2014; Nahlawi 2020: 52), the International Court of Justice’s judgment in the landmark Bosnian Genocide case (Bosnia and Herzegovina v. Serbia and Montenegro) seems to leave little room for interpretation on the matter. In its historic decision, the International Court of Justice (ICJ) held that, pursuant to Article I, Parties to the Genocide Convention have a ‘direct obligation to prevent genocide’ (Bosnia and Herzegovina v. Serbia and Montenegro: para. 165), which requires States not to themselves commit genocide, and places them ‘under positive obligations, to do their best to ensure that such acts do not occur’ (ibid: para. 432). Critically, the Court held that the obligation is not limited to a State’s territorial jurisdiction, but applies ‘wherever [a State] may be acting or may be able to act in ways appropriate to meeting the obligations in question’ (ibid: para. 183).

In setting out the extraterritorial nature of the obligation, the Court held that the duty to prevent genocide is governed by a due diligence standard rooted in the subjective ‘capacity’ of each States Party ‘to influence effectively the action of persons likely to commit, or already committing, genocide’ based on, among other things, ‘the geographical distance of the State concerned’ and ‘the strength of the political links, as well as links of all other kinds, between the authorities of that State and the main actors in the events’ (Bosnia and Herzegovina v. Serbia and Montenegro: para. 430). In addition, the Court held that ‘the State’s capacity to influence must also be assessed by legal criteria’, given that each State’s capacity to act will vary ‘depending on its particular legal position vis-à-vis the situations and persons facing the danger, or the reality, of genocide’ (ibid: para. 430).

By describing the obligation as extraterritorial in nature, but basing its application on a set of legal and factual criteria, the ICJ adopted a novel approach to expounding preventive obligations in relation to genocide (Milanović 2007: 685), requiring, in short, that States Parties ‘do everything within their individual and collective power to prevent genocide provided they have the requisite capacity to effectively influence the relevant genocidal actors and the knowledge of the existence of a serious risk that genocide might occur’ (Heieck 2018: 48). In other words: ‘the more a state can do, the more it must do’ (Milanović 2007: 687).

2.1.1.2 An obligation of conduct, not of result

Second, as with other due diligence obligations, this obligation is one of conduct and not one of result: States are not under an obligation to succeed in preventing genocide but must rather employ the means reasonably available to them to try (UN General Assembly 2001: 153–54; Coco and de Souza Dias 2020). The Court has emphasized in this regard that a State will not automatically incur responsibility for the failure to prevent merely because a genocide occurs, but rather if the State ‘manifestly failed to take all measures to prevent genocide which were within its power, and which might have contributed to preventing the genocide’ (Bosnia and Herzegovina v. Serbia and Montenegro: para. 430).

2.1.1.3 An obligation erga omnes, ex partes

Third, and importantly, the duty to prevent genocide under the Convention is considered both an obligation erga omnes and erga omnes partes, i.e. a duty owed by each State Party to the community of other States Parties (Longobardo 2015: 1202; Mettraux 2015: 71). As recalled by the ICJ in Belgium v. Senegal, this means that contracting States ‘do not have any interests of their own; they merely have, one and all, a common interest, namely, the accomplishment of those high purposes which are the raison d’être of the Convention’ (Belgium v. Senegal: para. 68).

Notably, in its recent provisional measures decision in the Myanmar genocide case, the ICJ held that the erga omnes partes nature of the obligation was enough, without more, to permit standing for The Gambia’s claim (The Gambia v. Myanmar: paras. 41–42). Among other things, the procedural consequence of the erga omnes partes nature of the obligations is that States Parties to the Genocide Convention may each bring claims regarding the application and interpretation of the Convention before the ICJ, regardless of whether the claimant State was ‘specially affected’ by an act of genocide, in conformity with their obligations under Article VIII.

2.1.1.4 The knowledge standard

Fourth, the ICJ held that the obligation to prevent genocide is not premised on a State’s actual knowledge of the risk of genocide, but that pursuant to a constructive knowledge standard: ‘[a] State’s obligation to prevent, and the corresponding duty to act, in fact arise at the instant that the State learns of, or should normally have learned of, the existence of a serious risk that genocide will be committed’ (Bosnia and Herzegovina v. Serbia and Montenegro: para. 431, emphasis added). Indeed, this constructive knowledge standard is part of what differentiates the failure to prevent genocide from complicity in genocide, which requires, among other things, actual knowledge that a genocide is imminent or ongoing (Milanović 2007: 687).

2.1.1.5 Ambiguities regarding breaches of the duty

Despite such guidance, a number of areas of uncertainty remain, the most consequential of which may concern the timing and duration of a potential breach of the duty to prevent genocide. In addressing the issue, the Court referenced the ILC Articles on the Responsibility of States for Internationally Wrongful Acts (ARSWA art. 14, para. 3) and held that a State can only be held responsible for breaching its obligation to prevent a genocide only if genocide is actually committed, with the breach of the preventive duty beginning at the moment the genocide begins and lasting for the period the genocide is ongoing if no action is taken in accordance with the State’s capacity to stop it (Bosnia and Herzegovina v. Serbia and Montenegro: para. 431). At the same time, the Court also acknowledged that this:

[…] does not mean that the obligation to prevent genocide only comes into being when perpetration of genocide commences; that would be absurd, since the whole point of the obligation is to prevent, or attempt to prevent, the occurrence of the act. (Bosnia and Herzegovina v. Serbia and Montenegro: para. 431, emphasis added)

Rather, the Court held that although a breach of a State’s obligation to prevent genocide may not occur until the genocide takes place, its duty to act arises at the instant that the State learns or should have learned of the existence of a serious risk of genocide.

Although it is conceptually intuitive to find that the commission of an atrocity is a prerequisite to challenging the failure to prevent it, a closer examination of the practical implications of this approach illuminates its fallacy. First, a number of commentators have pointed out that the idea that we cannot evaluate a State’s compliance with its duty to prevent genocide until a genocide has occurred is at odds with the notion of an obligation of conduct (Longobardo 2019; Gattini 2007; Forlati 2011). As Gattini has observed:

[O]ne could ask why it should not be possible to hold responsible a state which manifestly breached its obligation to prevent a violation of a peremptory norm of international law, even if the event was averted at the very brink owing to the intervention of third parties. (Gattini 2007: 702)

In the same vein, Forlati has argued that it is ‘unreasonable to exclude that responsibility may autonomously arise at least in the case of flagrant omissions’ (Forlati 2011: 200). Further underscoring the implications of this approach in the specific context of the Bosnian genocide, Milanović has equally noted:

It could certainly be argued that there was a serious risk of genocide in several other municipalities in Bosnia other than Srebrenica, most notably in the area around Prijedor, yet Serbia could still not be held responsible as ultimately genocide was not committed in Prijedor. (Milanović 2007: 687–88)

Further, conditioning the duty on the occurrence of a single event is at odds with the very nature of mass atrocities, the commission of which are widely understood to be predictable processes rather than spontaneous, unforeseeable events (UN Office on Genocide Prevention and the Responsibility to Protect 2014: iii). An objective state of facts can be proven as to the existence of a serious risk of genocide being committed—which should trigger preventive action—regardless of whether the genocide occurs.³ In other words, requiring that a genocide occurs before a breach of preventive obligations can be found forecloses the possibility of meaningfully challenging any failure to take appropriate preventive action until the moment the bloodshed begins. This seems incongruous with the very nature of an obligation to prevent.

2.1.2 The treaty-based duty to prevent war crimes.

Turning to war crimes: although they are not defined in a single multilateral treaty, the core legal provision concerned with preventive duties is Common Article 1 of the Geneva Conventions, pursuant to which High Contracting Parties undertake an obligation to both ‘respect and ensure respect’ for the Conventions in ‘all circumstances’ (Geneva Conventions, art. 1, emphasis added). According to the International Committee of the Red Cross (ICRC), Common Article 1’s obligation to ‘respect and ensure respect’ can be understood as reflecting both internal and external dimensions (ICRC 2016a). Internally, each High Contracting Party is required to both respect and ensure respect for the Conventions in relation to its own conduct, as well as conduct by its armed forces, other persons and groups acting on their behalf, and their populations as a whole (ibid: para. 118). This is understood to constitute a general duty to prevent breaches of the Conventions by those over whom a State has authority, based on the State’s means, the foreseeability of the violations, and the degree of influence the State exercises over those persons (ibid: para. 150).

Yet it is Common Article 1's external dimension which is most salient to our argument, for it raises the possibility of an extraterritorial nature of the duty to prevent war crimes. As stated by the ICRC, the prevailing view today is that Common Article 1 requires that States ensure respect for the Geneva Conventions by third States, including States that have not acceded to the Geneva Conventions (ibid: para. 120; Focarelli 2010: 128; Hakimi 2014: 254). This view accords with the notion that:

The interests protected by the Conventions are of such fundamental importance to the human person that every High Contracting Party has a legal interest in their observance, wherever a conflict may take place and whoever its victims may be. (ICRC 2016a: para. 119)

Such interpretation finds support in the ICJ’s Construction of a Wall case, where the Court held that, under Common Article 1, ‘every State party to that Convention, whether or not it is a party to a specific conflict, is under an obligation to ensure that the requirements of the instruments in question are complied with’ (Construction of a Wall: paras 158–59). The ruling has been interpreted by legal scholars as recognizing a treaty-based duty to prevent war crimes, premised on the concept of ‘ensuring respect’ for the provisions of the Geneva Conventions (Hathaway et al. 2018; Trahan 2018a; Nahlawi 2020: 61).

According to the ICRC’s 2016 Commentary to Common Article 1, the external dimension of Common Article 1 is considered to embody both negative and positive obligations: the negative obligation prohibits High Contracting Parties from encouraging, aiding or assisting in violations of the Conventions, while the positive obligation requires them to ‘do everything reasonably in their power to prevent and bring such violations to an end’ (ICRC 2016a: paras 188 and 154; Heieck 2018: 2). Again, the obligation is understood to be one of due diligence, requiring States Parties ‘to take all possible steps, as well as any lawful means at their disposal, to safeguard respect for IHL rules by all other States’ (Dörmann and Serralvo 2014: 724; ICRC 2016a: para. 165).

Notably, States’ due diligence obligations are only triggered if the event in question is ‘actually foreseeable’ (ibid: 730) and, in the view of the ICRC, the obligation ‘is not limited to stopping ongoing violations but includes an obligation to prevent violations when there is a foreseeable risk that they will be committed’ (ICRC 2016a: para. 164, emphasis added). While the specific preventive measures to be taken are at the discretion of the State, they may include measures aimed at exerting diplomatic pressure, coercive measures, and measures in cooperation with an international organization (Dörmann and Serralvo 2014: 725–26).

Some States have pushed back against this ‘expansive’ interpretation of Common Article 1, advocating instead for a significantly narrower interpretation of their preventive obligations.⁴ According to this ‘restrictive’ view, Common Article 1 merely obligates contracting States to adopt the measures necessary to ensure respect for the Conventions by the population over which it exercises authority, and other persons or groups whose conduct is attributable to the State (Robson 2020: 103; Focarelli 2010: 128). Importantly, this view asserts that Common Article 1 does not create an obligation to prevent breaches by other States or parties to a conflict (Robson 2020: 103). Recent scholarship also contends that the Geneva Conventions’ travaux préparatoires suggests that the scope of the duty to ensure respect does not, in fact, extend to the prevention of breaches by other States (ibid: 112). For example, Robson notes that, at the 1949 Geneva Conference, only the ICRC delegate spoke in favour of expanding the scope of the duty to ensure respect (ibid: 113), adding that a broad interpretation of the duty to prevent would be ‘impractical’, as efforts to allocate the burden between States not directly related to the breach would be fraught (ibid: 114). A related concern is articulated by Focarelli:

If all 194 contracting states have an obligation to take all positive measures in their power to respond to the one of them which has happened to breach the Geneva Conventions, then all those which do not react turn out to be in breach of common Article 1 and, since contracting states generally do not react, one should conclude that there are 193 or so breaches of Article 1 for any breach of the Geneva Conventions, a very extreme construction which is far from being supported by state practice. (Focarelli 2010: 171)

While it is correct to say that there is an absence of State practice supporting an extraterritorial duty to prevent war crimes, and difficulties concerning the allocation of the preventive burden amongst States, the same is true—on both counts—of the extraterritorial duty to prevent genocide. Regrettably, such atrocities often occur without a single preventive measure being attempted despite the erga omnes partes nature of the obligation. Furthermore, in evaluating whether an expansive or restrictive interpretation of Common Article 1 is more persuasive, one should not conflate the failure to enforce States’ preventive obligations with the content of those obligations.

A third interpretation, proposed by Hakimi—grounded in jurisprudence by both the ICJ and the European Court of Human Rights—would be that a positive duty to prevent breaches of the Convention applies extraterritorially if a State either exercises governmental authority over or substantially supports the Party in breach (Hakimi 2014: 270–74). According to this view, a State would ultimately be responsible for failing to prevent breaches of the Geneva Conventions if it ‘failed to exercise due diligence to restrain the perpetrators’ over whom it exercises government authority in a territory which it occupies or whom it substantially supports (ibid: 270–71).

2.1.3 The treaty-based duty to prevent crimes against humanity.

If a single gaping hole can be identified at the core of the codified legal framework governing atrocity prevention, it is in relation to crimes against humanity. To date, there is no multilateral treaty governing the prevention and punishment of crimes against humanity.⁵ The process of drafting such a treaty, however, has been part of the ILC’s program of work since 2014, culminating in its 2019 adoption of a set of Draft Articles on the Prevention and Punishment of Crimes Against Humanity (‘Draft Articles’, ILC 2019). The Draft Articles, aiming to provide the basis for a new convention, set forth detailed obligations on prevention. While they do not yet constitute treaty law, their drafting was the product of extensive submissions, discussions and negotiations among States and, as the basis of a possible future convention (currently under consideration by the UN General Assembly’s Sixth Committee), they reveal the general direction and potential scope of a future treaty-based duty to prevent crimes against humanity.

The notion of prevention appears multiple times in the Draft Articles,⁶ with core preventive obligations arising explicitly in Draft Articles 3 and 4. Draft Article 3(2) provides that ‘[e]ach State undertakes to prevent and to punish crimes against humanity’ (Draft Articles: art. 3(2), emphasis added); as noted in the ILC Commentary to the Draft Articles (ILC 2019), Article 3(2) is premised on the analogous article in the Genocide Convention stating that Contracting Parties ‘undertake to prevent and to punish’ genocide (Genocide Convention: art. I). Despite (or perhaps because of) its parallels in the Genocide Convention, several States have expressed concern that their preventive obligations under this Article are unclear, requesting further information on its scope (UN General Assembly 2019c).

In its Commentary, the ILC notes that Draft Article 3 can be regarded as setting forth a ‘general obligation’, with further content ‘addressed in various ways through the more specific obligations set forth in the draft articles that follow, beginning with draft article 4’ (UN General Assembly 2019b: 49). In his 2019 Report, the ILC Special Rapporteur explained that this provision should not be understood as repetitive of Draft Article 4, but rather regarded as ‘fixing an objective’ in light of which the other provisions of the text are to be interpreted and applied (UN General Assembly 2019b: para. 51).⁷ Understood in that context, Draft Article 4 sets forth greater detail as to States’ obligation to prevent. The ILC Commentary explains that, with respect to subparagraph (a), Draft Article 4 requires States to ‘pursue actively and in advance measures designed to help prevent the offence from occurring’ (ibid: 57), making explicit reference to ‘effective legislative, administrative, judicial or other appropriate preventive measures’ (ibid: 54a), as well as ‘cooperation’ (ibid: 54 b). According to the Commentary, the term ‘effective’ implies that each State is ‘expected to keep the measures that it has taken under review and, if they are deficient, to improve them through more effective measures’ (ibid: 58).

Importantly, it is not immediately clear whether the jurisdictional language contained in Draft Article 4(a) implies an extraterritorial dimension. Although some commentators have read Draft Article 3 to be extraterritorial in nature, and Draft Article 4 to be aimed internally (UN General Assembly 2019d: 11), the ILC Special Rapporteur has clarified that ‘by referring to acts occurring in any territory under its jurisdiction, the language is broader than a reference solely to conduct occurring in the state’s territory, but narrower than language that could suggest an obligation upon the State to develop […] measures to prevent any conduct worldwide’ (UN General Assembly 2015 para. 115, emphasis added).

In essence, while such a formulation unequivocally establishes an obligation to take measures domestically to prevent conduct within a State’s own territorial jurisdiction, the fact that the language used in Draft Article 4 is broader than a mere reference to the State’s territory begs the question of whether, in some cases, States should take preventive measures domestically that would have extraterritorial impact. This reading might be further supported by the ILC Commentary setting out, among the steps a State should take, the adoption of ‘national laws and policies as necessary to establish awareness of the criminality of the act and to promote early detection of any risk of its commission’ (UN General Assembly 2019b: 59–60, emphasis added).

The ILC Commentary further adds that international responsibility arises if the State fails to use its best efforts to organize the government and administrative apparatus to prevent crimes against humanity, as far as possible (ibid: 60). In addition to the fact that this language is suggestive of a due diligence standard, the ILC’s decision to premise the preventive obligations on the Genocide Convention indicates that such a standard would apply. As argued by van den Herik and Irving, these are ‘twin provisions’ (e.g. Article I Genocide Convention and Draft Article 3) which can ‘be regarded as being the same in contents, scope and nature’ (van den Herik and Irving 2018: 4; see also Kamminga 2018; UN General Assembly 2019d: 11).⁸

2.2 Prevention under customary international law

Although it is relatively uncontroversial to assert that the prohibitions against the commission of genocide, crimes against humanity, and war crimes constitute peremptory norms (Tladi 2018: para. 114; International Law Commission 2001: 85 para. 5), it is admittedly more difficult to ground the duty to prevent those crimes in customary international law. And although some commentators have argued that the respective duties to prevent each of the three atrocity crimes are not only grounded in customary international law but have also attained jus cogens status, significant controversy remains in this area (van den Herik and Irving 2018: 5; Heieck 2013). Nevertheless, and without asserting that such duties constitute peremptory norms, we set out below what we believe is evidence of a customary law basis for, at least, the duties to prevent genocide and war crimes, acknowledging that greater controversy exists as to crimes against humanity.

2.2.1 Customary international law and the duty to prevent war crimes and genocide

Customary international law is established on the basis of its two constitutive elements: a general practice among States, and its acceptance as law, or opinio juris (International Law Commission 2018: 124; Nicaragua v. United States of America: para. 183). With respect to the latter, the conventions governing the prevention of genocide and war crimes have been found to not only set out treaty-based obligations of prevention, but to themselves reflect and codify existing customary international law.

As to war crimes, for example, the ICJ made clear in its Nicaragua case that the duty to ‘respect’ and ‘ensure respect’ set out in Common Article 1 (as discussed above) ‘does not derive only from the Conventions themselves, but from the general principles of humanitarian law to which the Conventions merely give specific expression’ (Nicaragua v. United States of America: para. 220). This is further evidenced by UN General Assembly and Security Council resolutions calling upon High Contracting Parties to the Geneva Conventions to ‘ensure respect’ for its provisions (UN General Assembly 1988: para. 5; UN Security Council 1990: para. 5). As argued by Nahlawi in relation to the conflict between Israel and Palestine, the General Assembly has repeatedly passed resolutions calling upon not only particular States (i.e. those with a nexus to the conflict) but for all High Contracting Parties to ensure respect for the Fourth Geneva Convention (Nahlawi 2020: 60):

Although none of these Resolutions are legally binding, their recurring nature, combined with States’ widespread acceptance of their provisions as evidence through the overwhelming voting pattern in their favour allows for the extraction of […] opinio juris regarding the authoritativeness of this interpretation of Common Article 1. (ibid)

Similarly, there is increasing recognition that the duty to prevent genocide has grounding in customary international law (ibid: 53; Arbour 2008: 449-450; Mettraux 2015: 67). Heieck, in particular, has pointed to 60 years of uncontroverted opinio juris in General Assembly resolutions, Security Council decisions, statements of the Secretary-General, declarations of other UN bodies, and the responses of States to atrocity crimes in support of the customary status of the norm, despite a lack of consistent state practice (Heieck 2018: 100–110). He also asserts that, based on the ICJ’s jurisprudence in the North Sea Continental Shelf and Nicaragua cases, the absence of state practice is not prejudicial to this argument, and the traditional two-part test for establishing customary international law is ‘not appropriate in cases of fundamental concern to the international community, such as situations calling for the prevention of genocide.’ Indeed, in such cases, the ICJ has ‘stressed opinio juris at the expense of state practice’ (ibid: 89–118).

Even if one is unpersuaded by this customary law analysis, as Trahan has noted, the issue may be immaterial given the firm and extensive grounding genocide has in treaty-based international law (Trahan 2018b). The need to find solid customary international law footing for the duty to prevent may, instead, be of greater importance in relation to crimes against humanity, at least until a multilateral convention on their prevention and punishment is adopted.

2.2.2 Customary international law and the duty to prevent crimes against humanity

The existence of a customary law basis for the duty to prevent crimes against humanity is, without a doubt, contentious. On the one end, the ILC Special Rapporteur on the issue has noted that the General Assembly has referred to an obligation to prevent crimes against humanity in a number of instances, including in its 1973 Principles of International Co-Operation in the Detection, Arrest, Extradition and Punishment of Persons Guilty of War Crimes and Crimes Against Humanity (UN General Assembly 2015: 46). Similarly, in the comments and observations submitted by States to the ILC Draft Articles, a number of governments submitted statements which could be interpreted as opinio juris supporting a duty to prevent.

The Netherlands, for example, quoted its Advisory Committee on Issues of Public International Law as stating the ‘draft articles aim to codify existing rules under customary international law, such as the obligation to prevent’ (UN General Assembly 2019d: 2 and 12, emphasis added). Similarly, Chile stated that the ILC work ‘articulates […] obligations arising from the customary prohibition of crimes against humanity, namely, the duty of states to prevent them and to punish them’ (UN General Assembly 2019c: 9, emphasis added). The United Kingdom, by contrast, asserted that the ILC draft did not codify existing obligations under customary international law, but rather considered it to constitute a ‘proposal for the progressive development of the law (lex ferenda)’, noting its ‘concerns about creating such a broad, and potentially ever expanding, set of obligations for States in relation to crimes against humanity’ (ibid: 60–61).

A customary international law duty to prevent crimes against humanity may also find footing in the widespread acceptance of Articles 40 and 41 of the ILC Articles on the Responsibility of States for Internationally Wrongful Acts, which set forth a positive obligation for States to ‘cooperate to bring to an end through lawful means’ any serious breaches of an obligation arising under a peremptory norm of international law (International Law Commission 2001: arts. 40–41). Although the status of the duty to prevent crimes against humanity is uncertain, in light of the jus cogens prohibitions against the commission of crimes against humanity, it would appear that Articles 40 and 41 set forth at least a duty to cooperate to end the commission of crimes against humanity.⁹

Nevertheless, bringing the commission of atrocity crimes to an end must be distinguished from finding a proactive duty to prevent their commission. In addition, as Monica Hakimi points out, the ILC Articles ‘do not even try to identify the conduct that would satisfy a duty to cooperate, or when a state might incur responsibility for failing to cooperate’ (Hakimi 2014: 255). Finally, it must be acknowledged that the Articles have not been formally adopted in a multilateral treaty, and do not represent ‘hard law’, though they carry significant weight and authority (Nahlawi 2020: 63–64; Barnridge 2006: 89).¹⁰

3. Operationalizing states’ preventive obligations

What, then, can at this stage be derived from this evolving and incomplete body of law? In light of the above, we outline the principal issues with the current legal framework, before providing an assessment of how States can best operationalize their legal obligations where these may arise. Ultimately, we argue that despite the ambiguities in the law, there are critical points at which duties to prevent atrocity crimes can be invoked—and there are courses of action which could assist in meeting States’ preventive obligations. More specifically, we argue that once a State should know that there is a serious risk that mass atrocities will occur, they have an obligation to proactively assess their preventive capabilities, and that they can do so by undertaking what we are terming an ‘Atrocity Impact Assessment.’

3.1 The state of the prevention framework

In our evaluation, there are three principal issues with the current legal framework in relation to the duties to prevent each atrocity crime. The first is a lack of alignment between the treaty-based obligations imposed for each crime, presenting an untenable situation for States seeking to operationalize their obligations at the moment that risk appears. While the treaty-based duty is firmly established in relation to both genocide and war crimes, as mentioned, it does not yet exist for crimes against humanity, creating a number of practical issues. Specifically, as Sadat has argued, the absence of such a treaty has led, among other things, to (1) an impunity gap, under which it is more difficult to prosecute individuals for crimes against humanity; (2) a State responsibility gap, as no compromissory clause exists to permit litigation before the ICJ over States’ obligations over crimes against humanity; and (3) a situation of ‘definitional uncertainty’ over whether a particular atrocity was a crime against humanity (Sadat 2011).

Further, even existing treaties and multilateral conventions use different language vis-à-vis States’ preventive obligations for each atrocity crime, generating additional incongruities as to the content of treaty-based duties to prevent. The duty to prevent genocide under Article I of the Genocide Convention bears little resemblance at first glance to the Common Article 1 obligation to ‘respect and ensure respect’ for the Geneva Conventions, and jurisprudence interpreting the Genocide Convention—elaborating on the scope of the duty to prevent—has carefully stipulated that it should not be extended beyond the confines of that treaty (Bosnia and Herzegovina v. Serbia and Montenegro). Yet, for those assessing the existence of a risk, it is often difficult to tell what form of atrocities may ultimately be committed (or proven) in a given context, which is dependent upon the perpetrators’ intent, the existence and characterization of an armed conflict, and the ability to fit harm to a population into particular legal categories. At the moment that risk appears—the moment States urgently need to act—it may not be evident precisely which crimes will occur. For this reason, to the extent that legal obligations differ between crimes, it may be most practicable to adopt guidance relating to the most well-elaborated legal standard, which today is genocide.

The second problem with the current legal framework is the failure to view prevention as an obligation which should trigger action before bloodshed begins. Too often, the failure to prevent is simply equated with the commission of atrocities themselves, when in reality: (1) States need to adopt a preventive course of conduct when risk appears, without the benefit of knowing whether that risk will materialize into actual atrocities; (2) atrocity crimes may still occur even if preventive action has been taken; and (3) there can be an abject failure to prevent, yet atrocities may still be averted, through a stroke of luck or change in circumstance. In short, the existence of atrocity crimes cannot be regarded as a proxy for the failure to undertake preventive action. While the occurrence of atrocities is often—in reality—the impetus for considering whether States have failed to take preventive action, it should be possible to evaluate whether States have taken preventive action whether or not mass atrocities take place. Preventive action should indeed be taken as soon as the State learns of the risk of atrocities: otherwise, the notion of prevention becomes illusory and States would, for all intents and purposes, engage in mitigation instead.

The third problem with the current legal framework is a lack of clarity as to the content of States’ preventive obligations. Our review of the fragmented legal framework leaves us with a constellation of procedural and substantive points of guidance regarding the various duties to prevent, but significant gaps as to the content between them.¹¹ What we do know is that: (1) the duty arises at the instant a State learns or should have learned of the serious risk that atrocities would be committed, but—in the case of genocide—may not be invoked unless and until a genocide occurs; (2) the obligation is one of conduct, requiring States to use the means available to them to try to avert mass atrocities; and perhaps most importantly, (3) States are held to a due diligence standard, with the extent of each State’s responsibility evaluated against its capacity to influence perpetrators.

However, as many States have themselves raised (UN General Assembly 2016: para. 3), a lack of clarity exists concerning the forms of preventive actions which must be taken in fulfilment of preventive duties, even if States wanted to act responsibly and proactively. In addition, it is often impossible for States to know which preventive measures have been employed by other States, particularly if they include non-public measures such as backdoor diplomacy or the unpublicized rescission of military or economic aid. In fact, even within a given State’s apparatus, a lack of inter-agency coordination might be of serious hinderance to coherent assessments as to the State’s own preventive capacity (D’Alessandra, Whidden: 2022). Consistent enforcement of preventive obligations is therefore impossible to compel.

Nevertheless, we take the position that compliance with preventive obligations should, necessarily, start with an assessment of a State’s preventive capabilities. This would indeed be a logical first step prior to any subsequent action being taken in pursuance of preventive obligations a State may have. Furthermore, in the event that a State’s preventive record might be challenged, the existence of such an assessment would go a long way in supporting States seeking to defend their chosen course of action. For this reason, we believe such capacity assessments—which we discuss below as Atrocity Impact Assessments—to be helpful tools in operationalizing States’ preventive obligations, where these arise.

3.2 Atrocity impact assessments as a tool for prevention

States’ chosen tools for their preventive capability assessments may vary, and can take many forms. One available option, which we term an ‘Atrocity Impact Assessment’ (AIA), would entail undertaking a targeted assessment of a State’s activities, policies, and relationships with respect to potential atrocity crimes, as well as the specific means that each State has within its capacity to prevent them. While we are not asserting that there is a legal obligation to conduct an AIA, we believe States should consider adopting this particular form of assessment as a ‘best practice’ tool to ensure, as mentioned above, that they stay on the right side of the law and that they are doing what they can to avert atrocity crimes. An overview of our proposal and the general contours of such an assessment are set out below.

3.2.1 Due diligence and precedent for atrocity impact assessments.

As we have noted, under the duty to prevent each of the atrocity crimes, States are held to a due diligence standard. Impact assessments are regularly used in connection with this standard in other contexts: human rights impact assessments (HRIAs), for example, are understood to be a key component of human rights due diligence (BSR 2013: 6), and are used by governments around the world (as well as private sector companies) to provide clarity about human rights impacts and opportunities for mitigation. In fact, the Third Revised Draft of the Binding Treaty to Regulate, In International Human Rights Law, the Activities of Transnational Corporations and Other Business Enterprises (UN Human Rights Council 2021) appears to push future legal developments in this same direction. HRIAs may be conducted to assess a government or company’s impact in a particular country, or to assess the impact of a particular policy, activity, or product. Such assessments are generally regarded as helping to shape a proactive, rather than reactive, approach to human rights risks and opportunities.

Another example requiring States to conduct due diligence risk assessments in relation to atrocity crimes specifically is the Arms Trade Treaty (ATT). Pursuant to Article 7 of the ATT, States considering exporting conventional arms must, prior to authorizing an export, conduct a risk assessment regarding, among other things, the potential that the conventional arms could be used to commit or facilitate serious violations of international humanitarian or human rights law (ATT: art. 7(1)). The State Party must also consider whether there are measures that could be undertaken to mitigate the risks identified in the risk assessment (ibid: art. 7(2)). Though there is significant discretion afforded as to how to implement the ATT’s risk assessment obligation (Zwijsen, Kanetake, and Ryngaert 2020: 159–60), the ATT Secretariat has published a list of possible sources for States Parties to consider in conducting risk assessments (ATT Secretariat n.d.). The ICRC has also proposed a set of indicators that States should evaluate as part of their risk assessments, as well as guidance on possible risk mitigation measures (ICRC 2016b: 37–39). Importantly, the risk assessment obligation aims to regulate State conduct without legally relying on the actual commission of wrongful acts, instead constituting a procedural step aimed at preventing the occurrence of those acts in the first place (Zwijsen, Kanetake, and Ryngaert 2020: 158). In the context of mass atrocities, where prevention ought to be operationalized before atrocities occur, a similar procedural step could assist in shaping proactive State conduct.

Though the conduct required to discharge an obligation to act with due diligence cannot be readily transposed from one area of international law to another (McDonald 2019: 68; Bosnia and Herzegovina v. Serbia and Montenegro: para. 429), the general concept of due diligence requires a process or procedure of informed decision-making, which can be aided by conducting a risk or impact assessment. As explained by de Souza Dias and Coco, due diligence is, in short, a ‘standard of good governance’, meant to assess whether a State has employed its best efforts to address risks, threats, or harms (Coco and de Souza Dias 2020). It is meant to be a flexible standard, designed to take into account the circumstances at hand, the State’s capacity and the nature and foreseeability of the risk (Kulesza 2016: 264).

In the context of atrocity prevention, Irving and van den Herik have described the due diligence standard as serving to ‘transpose’ difficult and situation-specific questions regarding the substantive content of prevention into questions on informed decision-making and process (van den Herik and Irving 2018: 9). In other words, the due diligence standard in the context of atrocity prevention does not mandate a particular form of preventive action, but, at its core, requires a State to ‘make informed decisions about the most appropriate course of conduct’, and to ‘explain those decisions and subsequent conduct’ (ibid: 8).

This understanding of due diligence also accords with the knowledge standard applicable to the duty to prevent genocide, which as we have noted, is premised on not only what a State knew regarding risk, but also what it should have known (Bosnia and Herzegovina v. Serbia and Montenegro: para. 432). The constructive knowledge element can, indeed, be understood to ‘imply a risk-assessment that, at the very least, has close synergies with procedural due diligence obligations’. (van den Herik and Irving 2018: 10–11). It is also worth noting that, in Bosnia v. Serbia, the ICJ has made explicit reference to the notion that, ‘[i]n this area the notion of ‘due diligence’, which calls for an assessment in concreto, is of critical importance’, while recalling that liability does not attach merely because the desired result (i.e. the aversion of genocide) is not achieved (Bosnia and Herzegovina v. Serbia and Montenegro: para. 430, emphasis added).

In short, fulfilment of legal obligations to prevent atrocities, and associated due diligence standards, appear to contemplate a procedural step: a targeted assessment aimed at surfacing risks that States should be aware of to avoid liability, and to assist them in formulating plans for preventive action. Under an AIA, a State would assess its particular impact in a given country or region, and the specific measures that it could take to prevent the commission of atrocity crimes in that context. As with impact assessments in a human rights or environmental law context, an AIA can best be understood as a tool that can support governments in identifying and mitigating their adverse impacts (Danish Institute for Human Rights 2017: 3–4).

There is, in fact, important precedent for conducting risk or impact assessments in the context of atrocity crimes.¹² During the Obama Administration, for example, the US State Department and USAID developed a joint ‘Atrocity Assessment Framework’ as supplemental guidance to the State/USAID Conflict Assessment Frameworks, aimed at assisting decision-makers to ‘understand the dynamics that underpin a situation where there are indications of atrocity risk or where atrocities are underway’ and to identify and prioritize government options to reduce the risk of atrocity crimes (USAID and US Department of State 2015). Importantly, the assessments contemplated under this framework include the identification of both triggers for atrocity risk and windows of opportunity for prevention (ibid). The framework states that an atrocity assessment should result in concrete outputs including key atrocity dynamics, developments to monitor, and recommendations for atrocity prevention or response (Buchwald and Keith 2019).

Moreover, in connection with their preventive obligations, States often undertake to adopt legislation and policies necessary to further prevention. Committing to undertaking Atrocity Impact Assessments would be one such practice, aimed at making good on States’ commitments to prevent atrocity crimes to the best of their ability.

3.2.2 Proposed triggers for conducting an atrocity impact assessment.

As discussed above, legal duties to prevent atrocity crimes are not necessarily limited to a State’s territorial jurisdiction or control (Bosnia and Herzegovina v. Serbia and Montenegro: para. 183). Under what circumstances, then, would an AIA be triggered? In the context of genocide, States face legal exposure if they were either aware, or should normally have been aware, of the serious risk that genocide would occur. Though it is difficult to outline all circumstances in which a State should be aware of the serious risk of mass atrocities, a few bright lines can be drawn. First, in the relatively rare circumstance in which the ICJ has addressed the issue directly, ruling that a serious risk of mass atrocities exists, all States should be undertaking their own AIA to address their potential contribution to the risk of atrocity crimes, and their ability to prevent them. This occurred, for example, in January 2020, when the ICJ issued its Provisional Measures Order in the case between The Gambia and Myanmar under the Genocide Convention (The Gambia v. Myanmar).

Similarly, in situations where an international fact-finding mission has made a specific finding on the likelihood of atrocity crimes, States will have difficulty denying that they are on notice of the risks. This is particularly true given that, more and more, such inquiries are mandated by UN bodies to make findings specifically on whether atrocities have been or are being committed (D’Alessandra 2017). This, too, occurred in the context of Myanmar, as the Independent International Fact-Finding Mission submitted a report to the Human Rights Council in September 2019, finding ‘that there is a serious risk that genocidal actions may occur or recur, and that Myanmar is failing in its obligation to prevent genocide, to investigate genocide and to enact effective legislation criminalizing and punishing genocide’ (UN Human Rights Council 2019: para. 9). As with the ICJ’s Order, this type of finding should trigger States to undertake an AIA of their actual and potential impact on the relevant atrocity crimes, considering both their current role in contributing to the risk of mass atrocities and their capacity to prevent them.

Finally, the UN Office on Genocide Prevention and the Responsibility to Protect (‘UN Joint Office’) has dedicated significant resources to developing tools and frameworks for assessing the existence of a serious risk of mass atrocities. In 2014, the UN published a ‘Framework of Analysis for Atrocity Crimes’ aimed at assessing the risk of genocide, crimes against humanity, ethnic cleansing and war crimes (UN Joint Office 2014). The Framework contains two core analytical tools for assessing the risk of atrocity crimes: a list of risk factors, and a set of indicators for each. Though these resources can be deployed by States themselves, one of the main tasks of the UN Joint Office is to identify early warning signs of the risk of atrocity crimes, and, where appropriate, to bring that situation to the attention of the Secretary-General and the Security Council (UN Joint Office n.d.). When this occurs, any State that receives notification of the risk should be prompted to conduct an AIA as part of its effort to fulfil its preventive obligations.¹³

Thus, without outlining all circumstances in which an assessment is required, at a minimum, the following circumstances should trigger an AIA: (1) a judicial finding by the ICJ that there is a serious risk of atrocity crimes; (2) an analogous finding by an international fact-finding mission; and (3) a notification on the serious risk of atrocity crimes by the Special Advisers on the Prevention of Genocide and the Responsibility to Protect. In each of these circumstances, it would be difficult for a State to deny that it had knowledge—actual or constructive—of the serious risk that atrocities would be committed. States in receipt of any of these forms of warning should be undertaking an AIA to examine their ability to help avert atrocity crimes.

3.2.3 Proposed contours of an atrocity impact assessment.

Under an AIA, States should examine the specific measures within their power to assist in preventing atrocity crimes. This analysis derives from the Bosnian Genocide case, and is further supported by the Commentary to the Draft Articles on Crimes Against Humanity, which explains that Draft Article 4 requires States to ‘pursue actively and in advance measures designed to help prevent the offence from occurring’ (UN General Assembly 2019b: 57). According to the ICJ, State responsibility is ‘incurred if the State manifestly failed to take all measures to prevent genocide which were within its power, and which might have contributed to preventing the genocide’ (Bosnia and Herzegovina v. Serbia and Montenegro: para. 430). Operationalizing this standard would therefore require States to engage at the outset in an analysis of the specific measures each has the capacity to take to prevent atrocities, for failure to do so would necessarily expose States to liability in the event that a genocide unfolded.

In other contexts, impact assessments map an entity’s operations and policies against the specific rights that may be affected, to identify vectors for harm and opportunities for mitigation. In the private sector, for example, human rights impact assessments may outline the potential effects of company products, policies, and operations on human rights in a given context, and will then use a prioritization framework to guide internal action (usually considering both the likelihood and severity of harm). In the context of an AIA, we would expect that States would be engaging in a similar mapping exercise, which may be supported by consultations with both internal stakeholders and those on the ground. We would expect that States would be evaluating issues including (but certainly not limited to):

• Leverage over likely perpetrators:

  • Has a comprehensive mapping been done of likely perpetrators/conflict actors, and are any potential perpetrators susceptible to being dissuaded or incentivized to defect?

  • What are the actual or perceived grievances of likely perpetrators, and is there scope for substantive action to reduce the force of those grievances? (Atrocity Assessment Framework 2015)

  • What and where are the valued assets of likely perpetrators, and what would an effective sanctions regime look like?

  • Does the State provide aid to either the State from which potential victims or the State from which perpetrators hail? Is any of that aid tied to the security sector and, if so, is it at risk of being used in potential future atrocity crimes?

• Ability to interfere with the logistics of atrocity commission:

  • How are weapons being sourced, and how might the flow of arms and military equipment be disrupted?

  • What types of communication equipment is being used by perpetrators, and are there opportunities to disrupt communication flows if necessary?

  • What kinds of trade does the State conduct in the region, and has sufficient due diligence been completed around the impact of any business transactions?

  • Who are third party enablers, disablers, or bystanders that might be engaged to support prevention? (Atrocity Assessment Framework 2015)

• Ability to protect those at risk:

  • What means do targeted groups have to protect themselves, and how could these efforts be bolstered or supported? (Atrocity Assessment Framework 2015)

  • What humanitarian actors are located in the region, and how might the State better support those actors to ensure uninterrupted food, water, and medical supplies to potential atrocity victims? (Serafino et al. 2014)

  • In the event of mass atrocities, what might likely humanitarian corridors and evacuation routes look like from impacted areas, and what kinds of pre-emptive action could be taken to ensure those remain protected?

• Windows of diplomatic opportunity:

  • What is the State’s relationship with both the government of the affected State, and the State from which perpetrators hail? What kinds of leverage might exist for encouraging preventive action, and are there shared spaces or fora in which the State might be able to strategically apply pressure in support of potential victims?

  • Does the State have the ability to cooperate with other governments or organizations to support preventive action?

  • What backchannels may be available, particularly to those individuals with a direct line to or influence over potential perpetrators?

  • Should the State be alerting other governments or regional/international organizations, including UN bodies and mandate holders, to the risk of potential harm? What collaborative actions can be undertaken, including through regional or multi-lateral bodies?

• Assessment of the legal landscape:

  • What kinds of legal recourse does the State have available to it in relation to the atrocity risk? What jurisdiction might international, regional or domestic courts be able to exercise over potential atrocities, and how might that be leveraged to dissuade perpetrators?

  • Does the State share any treaty or convention commitments with potential perpetrators which render available legal avenues for action?

  • Could the State join in any legal proceedings or offer support to legal proceedings initiated by those at risk?

• Engagement in the information space:

  • What kinds of narratives are being used to fuel atrocity risk? Are vulnerable groups being dehumanized or ‘othered’, and how can these narratives be countered or rendered less potent?

  • What are the core channels through which dangerous narratives are circulating? Is hate speech fuelling the risk of atrocity crimes, and could the private sector be constructively engaged to reduce the virality or potency of hate speech narratives?

• Risk of sexual and gender based violence:

  • What are the likely risk vectors for sexual and gender based violence?

  • What are the patriarchal structures that might exacerbate risk in this context, and are there opportunities to undermine misogynistic policies and structures that contribute to the risk of sexual and gender based violence? (Global Justice Center 2018)

  • Are there risks—either of sexual and gender-based or other violence—specifically affecting particularly vulnerable categories of individuals, such as, among other things, children, the disabled, or sexual minorities?

• Military capabilities:

  • Does the State have a military presence in the affected region that might be leveraged to protect victims on the ground?

  • Is there scope and legal basis for military intervention for preventive or humanitarian purposes?

Situations of atrocity risk are as unique as fingerprints: there are no one-size-fits-all solutions, nor blueprints for atrocity prevention. As a result, an AIA will likely result in a unique set of preventive recommendations for each situation of atrocity risk. Nevertheless, there are ‘tools in the toolbox’ which should routinely be evaluated for potential use: these include economic sanctions, diplomacy and bargaining, humanitarian assistance and refugee support (especially with States neighbouring the territory at risk of a mass atrocity event), asset freezes, suspension of economic or military aid, arms embargoes, diplomatic missions (which might have a cooling effect on ongoing violence), international inquiries, as well as the possibility of prosecution in domestic and international fora and sustained pressure on the international stage. These tools also include referral to UN or other multilateral bodies, cooperation with other States that may exercise greater leverage in a given situation, the suspension of membership before regional or treaty-based bodies, and/or promises of future constructive engagement in response to meaningful change (D’Alessandra, Sutherland, and Wu 2021: paras. 16–17).

Importantly, available tools may also include instituting litigation before the International Court of Justice, as The Gambia did with respect to Myanmar, quite notably without any assertion of direct injury as a result of the atrocities committed. The tools that might be employed are as varied as imaginative minds can conceive of, and an AIA might incentivize such imagination at a stage where it could make a difference. Equally, this notion underscores the importance of multilateral cooperation, for even where States may lack capacity to undertake prevention activities directly, other forms of action—such as providing diplomatic support, not blocking binding resolutions by the UN Security Council or action by the UN General Assembly, or other forms of diplomatic action—can indirectly support them in fulfilling their preventive obligations.

It may be questioned, of course, whether an AIA should be assessing the risk of atrocity crimes generally, or be limited to genocide; after all, the ‘should have known’ standard formally appears only in relation to States’ duty to prevent genocide. As States do not yet have a treaty-based duty to prevent crimes against humanity, the duty has, of course, not yet had the benefit of judicial interpretation for purposes of assessing whether the knowledge standard is, as with genocide, applicable to instances in which the State is ‘aware or should normally have been aware’. Yet, given the intentional parallels between the drafting of the duty to prevent crimes against humanity and the duty to prevent genocide, it is reasonable to assume that future interpretations of the twin duties would be consistent, and that the ICJ’s dicta as to the ‘scope, nature, and contents of the obligation to prevent genocide apply mutatis mutandis to the obligation to prevent crimes against humanity…’ (van den Herik and Irving 2018: 4). Further, and on a practical level, it is often not apparent, before atrocities are underway, whether a risk of atrocity crimes will manifest as genocide as opposed to crimes against humanity or, in the context of an armed conflict, war crimes. At the moment in time that States need to operationalize the duty to prevent, an atrocity impact assessment, rather than a genocide impact assessment, is thus a more appropriate course of action.

3.2.4 Transparency and disclosure

In many contexts, risk and impact assessments are conducted purely to drive internal awareness of risks and available remediations, and to ensure comprehensive consideration of the risk landscape. As a result, and particularly in the AIA context, it will often be neither appropriate nor constructive to expect that a government will disclose the full findings of an impact assessment, which is likely to include highly sensitive and non-public information. Analogous dynamics play out with impact assessments in the corporate context, and an appropriate level of transparency often simply requires ‘the disclosure of a short summary report that includes a description of the [impact assessment] process and method used along with a summary of high-level findings’ (BSR 2013). Disclosing this type of summary report can also foster dialogue with external stakeholders about next steps. Significantly, it may also indirectly incentivize and build momentum for other governments to conduct AIAs.

3.2.5 Significance of an atrocity impact assessment

At a fundamental level, the practice of conducting an AIA would contribute to ensuring that State actors cannot retreat into the ‘twilight between knowing and not knowing’ (Power 2013). It would assist in shifting States’ focus to a point earlier in the atrocity lifecycle, away from atrocity response and towards the actual aversion of mass violence. Further, Atrocity Impact Assessments would assist in normalizing a practice of States proactively assessing their capacity to prevent, and build momentum within the international community for collaboratively assessing high-impact opportunities for preventive action. Finally, on a practical level, by engaging in robust impact assessment earlier in the atrocity lifecycle, States would be alerted to a broader range of options for effective action, leading to better and more targeted preventive strategies.

4. Concluding thoughts

Today, the Rohingyas—like many other populations at risk—face a new set of grave threats to their existence. Accompanying those threats are clear warning signs, just as there were before the last round of atrocities they endured. Faced with those warning signs, some States will seek to take preventive action, while others will wait until it is, yet again, too late. This paper intends to speak to those States that seek to take preventive steps, understand their legal obligations and, most importantly, to operationalize them with concrete action. It has outlined what is known regarding obligations to prevent each of the three atrocity crimes under both treaty and customary international law, while elucidating the gaps which make such obligations so difficult to enforce. These include a lack of clarity as to content, and a failure of States to perceive obligations as procedural in nature, which require them to examine their current impact and preventive capacity. Too often, this manifests as an absence of State action, compounded by lack of enforcement. States can and must do better.

Obligations to prevent atrocity crimes have, for too long, existed as either a moral responsibility or a political platitude. It is time to ground these obligations in law, and to insist upon their enforcement. Though there is still significant contention regarding preventive obligations, the duty to prevent genocide, at a minimum, has been elucidated under international law, and requires States to act on early warnings. We urge States looking to fulfil their legal obligations to undertake an Atrocity Impact Assessment as a matter of best practice, evaluating both their current impact in a country or region and the potential measures they could take to assist in averting mass atrocities. Such assessments would assist in pushing conversations regarding risk, preventive measures, and State capacity to the moment they matter most—when they could be effective in averting the commission of the gravest, most egregious crimes on earth.

Footnotes

Acknowledgements

The authors are grateful to Dapo Akande, Beth Van Schaack, Wes Rist, Talita DeSouza Dias, Miles Jackson, Kate Johnston, Tsvetelina van Benthem, Nicole Widdersheim, Erin Rosenberg, Hannes Jöbstl and Stephen Rapp for their comments; to Peter Grenzow and Alexander Wolffenduttel for their editorial and research assistance; and to the Public International Law Discussion Group and members of the Institute for Ethics, Law and Armed Conflict for their invaluable input. All errors remain the authors’ only.

Conflict of interest statement

None declared.

References

Treaties

Cases