Containment Instead of Refoulement: Shifting State Responsibility in the Age of Cooperative Migration Control?


 Containment policies whereby destination States provide funding, equipment and training to transit States that intercept refugees on their behalf suggest that destination States try to circumvent the prohibition of refoulement and raise the question to what extent destination States can avoid responsibility for violations of the rights of migrants and refugees by cooperating with transit States. Answering this question requires broadening the analysis beyond the principle of non-refoulement, including not only international human rights law, especially the right to leave and the concept of jurisdiction, but also the law of State responsibility, notably the prohibition of complicity. This article argues that, although it remains debatable whether the principle of non-refoulement applies when transit States intercept migrants and refugees on behalf of sponsoring destination States, the wider network of international law rules constrains the latter’s ability to avoid responsibility when implementing cooperative migration control policies.


INTRODUCTION
Various contemporary migration control practices across the globe reveal that destination States such as the United States of America (USA), Australia and European Union (EU) Member States cooperate with transit countries to reduce migrant arrivals. Such policies raise the suspicion that they are designed, at least in part, to circumvent the prohibition of refoulement, which is widely seen as the cornerstone of international refugee law. 1 They also raise the question as to what extent destination States can avoid responsibility by cooperating with transit States?
Indeed, while the externalisation of migration control is not a new phenomenon, developments in recent years are characterised by a shift in migration control implementation from destination States to transit States, resulting in containment policies, whereby migrants and refugees are 'contained' in transit States rather than returned from a destination State to a transit State. Containment policies raise two specific legal challenges. The first concerns the territorial scope of the principle of non-refoulement: does it extend to containing people in a State where they are at risk, including of onward refoulement? The second issue concerns the question of allocating conduct: if a transit State prevents the onward movement of migrants to a destination State on behalf of the latter, does it trigger the responsibility of the destination State? Or only the responsibility of the transit State, or both?
To the best of this author's knowledge, these questions have not yet been addressed by courts, although a case against Italy is currently pending before the European Court of Human Rights (ECtHR) regarding its involvement in pullback operations by the Libyan Coast Guard. 2 Although there is some case-law on unilateral extraterritorial interceptions by destination States, either by 'pushing back' boats on the high seas 3 or by preventing people from boarding a plane at foreign airports, 4 the applicability of the principle of non-refoulement to situations where interceptions are carried out by transit States rather than destination States has not yet been adjudicated. It is therefore worth exploring the legal issues raised by current containment policies.
The aim of this article is to show that the issue of destination States' responsibility for asylum seekers beyond their borders requires broadening the analysis beyond the principle of non-refoulement. This includes not only various norms of international human rights law but also the law of State responsibility. In that sense, rather than determining whether, in a specific case, a destination State incurs responsibility under international law, the article discusses more generally the various rules that must be taken into account when determining destination States' responsibility.
Accordingly, the main argument of this article is that, although sponsoring destination States may argue that the principle of non-refoulement does not apply when transit States intercept refugees on their behalf, this does not mean that such practices automatically enable them to avoid responsibility under international law. Indeed, not only the principle of non-refoulement but also rules of international human rights law and the law of State responsibility constrain State actions, which becomes clear if one examines them from a broader perspective. The analysis refers to three specific instances of containment policies: Italy-Libya, Australia-Indonesia and the USA-Mexico. However, this overview is not meant to be exhaustive. Moreover, containment policies are but one example of cooperative migration control policies, which also include other practices whereby destination States cooperate with States of transit and/or origin in order to stop, or at least reduce, migration flows. Other practices include third-country processing for instance (as in the case of Australia's cooperation with Nauru and, until recently, Papua New Guinea). The article uses the terms (sponsoring) destination State and (cooperating) transit State because they are often used in policy documents and scholarship, although it is acknowledged that these definitions are relative. In particular, transit States may in fact become destination States as a result of containment policies. Last, the analysis focuses on migrants who are affected by cooperative migration control policies, which includes, but is not limited to, asylum seekers and refugees.
The article is structured as follows. Section 2 traces the rise of cooperative migration control policies and discusses three examples of bilateral cooperation between a destination and a transit State: Italy-Libya, Australia-Indonesia and the USA-Mexico, including the risks of human rights violations suffered by those intercepted. Section 3 analyses which violations of human rights and refugee law the policies discussed in Section 2 may cause. It discusses to what extent the principle of non-refoulement applies, but also examines other relevant norms, with a special focus on the right to leave. Section 4 examines when sponsoring destination States exercise extraterritorial jurisdiction, and hence may incur responsibility, in the context of migration control. Section 5, in turn, shows that, even if destination States do not exercise jurisdiction, they may nevertheless incur responsibility for supporting transit States who intercept migrants and refugees on their behalf. Last, Section 6 concludes that, although destination States may argue that they can circumvent the principle of non-refoulement by cooperating with transit States, the wider network of international law rules to which non-refoulement belongs constrains their ability to avoid responsibility when implementing cooperative migration control policies.

COOPERATIVE MIGRATION CONTROL PRACTICES
This section briefly discusses the rise of cooperative migration control policies, before describing three specific instances of containment practices implemented by transit States in various parts of the world-Libya, Indonesia and Mexico-on behalf of neighbouring destination States: Italy, Australia and the USA. It also addresses the risks such policies pose to the rights of intercepted migrants, asylum seekers and refugees.

A. The Rise of Cooperative Migration Control Policies
Destination States today implement a wide range of deterrence measures, reflecting a dominant paradigm for international refugee policy. 5 Since the 1980s, destination States have thus implemented policies 'that seek to keep most refugees from accessing their jurisdiction, and thus being in a position to assert their entitlement to the benefits of refugee law' . 6 More specifically, the nature of such policies has changed, moving from 'traditional' measures to prevent refugees from reaching destination States, such as visa controls and carrier sanctions, and high seas interdiction to cooperation-based measures, such as the offering of financial incentives and the provision of equipment, machinery or training for instance. 7 These developments in the nature of migration control policies can be explained by changes in migratory patterns and technologies, policy transfers and developments in international refugee and human rights law. 8 Indeed, various legal challenges have been brought against unilateral migration control policies. Practices whereby destination States intercept migrants, including refugees, beyond their borders have been challenged in court, sometimes successfully. The majority of the US Supreme Court held in Sale that Article 33 of the Refugee Convention 9 does not apply to aliens interdicted on the high seas. 10 However, the Inter-American Commission on Human Rights, ruling on the same facts, agreed with UNHCR and the dissenting opinion of Justice Blackmun in Sale that Article 33 had no geographical limitations. 11 In the European context, the Prague Airport case concerned British immigration officers temporarily stationed at the Prague airport who refused leave to enter the United Kingdom (UK) to Czech nationals of Roma origin. The then House of Lords found that the procedure did not breach the principle of non-refoulement but that it did unlawfully discriminate against Roma on racial grounds. 12 Particularly relevant in the context of this article is the ECtHR's finding in the Hirsi case that Italy had breached various provisions of the European Convention on Human Rights 13 (ECHR) because it had intercepted asylum seekers on the high seas and returned them to Libya. 14 It has been argued that 'precisely when they try the hardest to protect rights beyond territorial borders, courts acquire the most significant role in providing the conditions for the rights' further violation' . 15 Indeed, although decisions such as Hirsi can be seen as evidence that international refugee and human rights law constrains migration control policies, they can also be said to indirectly drive and enable policy developments in a more regrettable way. 16 Thus, recent developments in migration control suggest that 'destination states are actively learning from past judicial interventions' and intention- ally design their policies so as to avoid triggering obligations. 17 Destination States seem to reason that while the principle of non-refoulement prohibits them from returning migrants, especially refugees, to a place of risk, they can act to prevent them from coming in the first place, thereby realising the desired effect of limiting the number of migrants and refugees in destination States but without breaching the principle of non-refoulement. Current cooperative migration policies involving Italy, Australia and the USA illustrate this point.

(i) Italy and Libya: 'pullbacks' by the Libyan Coast Guard
Italy possibly presents the clearest example of destination States' belief that they can circumvent the principle of non-refoulement by cooperating with transit States. Indeed, it can be argued that Italy's post-2016 cooperation with Libya developed in response to the ECtHR's Hirsi ruling. A key element in Hirsi is that the ECtHR established that Italy had exercised jurisdiction over the applicants. 18 More specifically, it held that Italy exercised both de jure and de facto control over the applicants because the Italian authorities had transferred them onto Italian vessels and disembarked them in Tripoli. 19 However, by emphasising the control requirement, the ECtHR can be said to have implicitly acknowledged that without such control, there would be no jurisdictionand hence no obligation to respect human rights obligations, including the principle of non-refoulement. Mann thus notes that Hirsi 'contributed to understandings of how to evade judicial review in future cases. By saying that a state must not turn back asylum seekers with boats under their de jure or de facto control a court is also inviting such policies, as long as they can be conducted with no such control' . 20 Indeed, the more recent migration control policies implemented by Italy seem designed not only to save lives at sea but also to prevent migrants from reaching Italian shores. On 2 February 2017, Italy and Libya signed a new Memorandum of Understanding on Cooperation on Development, Combating Illegal Immigration, Human Trafficking and Smuggling and on Strengthening Border Security. 21 They thereby agreed to implement cooperation initiatives as foreseen in previous bilateral agreements, including efforts to prevent clandestine immigration. Under the current agreements, the migrants who are returned to Libya are intercepted by the Libyan Coast Guard rather than by Italian ships. 23 Italy supports such practices by donating boats and maintaining them and coordinating and instructing rescue operations and through financial aid. 24 More recently, Italy has also closed its ports to ships carrying rescued migrants and handed over the coordination of rescue operations to the Libyan Coast Guard, while Libya has officially declared its search and rescue region. 25 Italy thus seems keen to prevent migrant boats from reaching international waters-or even leaving Libya altogether-and from being rescued by European vessels, which disembark rescued migrants in Europe rather than Libya. 26 Italy is not alone in supporting the Libyan Coast Guard's pullback practices. Indeed, the EU and its Member States also sponsor the Libyan Coast Guard.

(ii) Australia and Indonesia: preventing departures
Australia has likewise implemented a variety of migration control policies in cooperation with several States in the region. 31 More specifically, in the same way that Hirsi can be seen to have contributed to Italy's current policies, it has been argued that Australia's increased cooperation with Indonesia is a reaction to the Australian High Court's decision in M70/2011, which found Australia's offshore processing framework to be illegal. 32 Mann thus observes that 'after M70/2011, significant evidence shows that incentives for preventive interceptions have increased' . 33 Unlike Italy, however, Australia's efforts to prevent refugees reaching its shores also include unilateral interceptions, whereby Australian vessels intercept migrant boats and either return them to Indonesia or transfer their passengers to an offshore processing centre. 34 However, '[i] n an attempt to reduce the number of refugees arriving in Australia via Indonesia, Australia entered into a number of agreements and provided substantial funding to Indonesia to increase its border patrol policies' . 35 In 2000, Australia thus concluded the Regional Cooperation Arrangement together with Indonesia and the International Organization for Migration (IOM). Under this agreement, Indonesian authorities intercept migrants and refugees suspected of travelling irregularly to Australia, while the IOM provides 'repatriation assistance' for migrants wishing to return to their country of origin. 36 Furthermore, Australian police have worked closely together with their Indonesian counterpart in order to survey, disrupt and intercept asylum seekers attempting to travel to Australia through Indonesia. 37 Moreover, like Italy with Libya, Australia provides Indonesia with funding, training and equipment. 38 Australia also funds the accommodation of asylum seekers in Indonesia through the IOM, which also provides 'repatriation assistance' for migrants wishing to return to their country of origin. 39  Like Italy and Australia, past efforts by the USA to control migration focused on intercepting boat migrants from Haiti. However, in recent years the focus has shifted to its land border with Mexico following an increase in the number of unaccompanied minors from the Northern Triangle countries (El Salvador, Guatemala and Honduras) apprehended by authorities of the USA at the USA-Mexico border. 41 Thus, the USA provides assistance to Mexico as part of security cooperation within the framework of the Mérida Initiative, a bilateral partnership launched in 2007. 42 The Initiative is built on four pillars, the third of which is creating a 'twenty-first-century border' . 43 It focuses not only on Mexico's northern border with the USA but also on its southern border with Guatemala and Belize. 44 Since 2014, Mexico has thus implemented the 'Southern Borders Programme' with support of the USA. The latter includes equipment for biometric data sharing and communications; helicopters and patrol boats; technical assistance; and training of migration, police and judicial officials. 45 It was reported in 2017 that the USA had provided $US24 million in equipment and training assistance to Mexico's National Migration Institute (INM), including non-intrusive inspection equipment, mobile kiosks, canine teams and training for INM officials in the southern border region, and had committed more than $US75 million more in that area. 46 During the first two years of the Southern Borders Programme, the number of migrant apprehensions in Mexico almost doubled, although numbers have levelled more recently. 47 Thus, between October 2014 and May 2015, the INM detained over 110,000 Central American migrants, which was more than the US border patrol during the same period. 48 The number of deportations of Central American migrants from Mexico has likewise increased after the implementation of the Southern Borders Programme. 49 More recently, in early 2019 the USA started implementing 'Migration Protection Protocols' , commonly known as the 'Remain in Mexico' policy. Asylum seekers are thus returned to Mexico where they are expected to await the outcome of their immigration proceedings. 50 Furthermore, so-called 'metering' practices at official ports of entry restrict the number of asylum seekers allowed to enter the USA, obliging asylum seekers to wait in line in Mexico. 51 Pressure from the USA is also alleged to have put an end to the granting of humanitarian visas by Mexico and increased migrant detention in Mexico. 52

C. The Plight of Intercepted Migrants and Refugees
While the above-mentioned policies are similar insofar as they concern situations where a destination State provides training, funding and equipment to a transit State which intercepts migrants and refugees on its behalf, the consequences for those intercepted vary. Numerous reports suggest that the current human rights situation of migrants in Libya is extremely concerning, as it includes killings, torture, arbitrary detention, rape, forced labour and slavery. 53 This suggests that the situation in Libya today is so dire that nobody can be returned to Libya, despite attempts by Italy and the EU to improve detention conditions in Libya. 54 Likewise, the situation of asylum seekers and refugees who remain in limbo in Indonesia may amount to inhuman and degrading treatment. They 'are not allowed to work, have only minimal healthcare or social support, and may face arbitrary detention and destitution' . 55 Migrants face a lack of physical safety, severe material deprivation and social isolation 56 while detention conditions are very harsh. 57 There are also increasing reports of mental health problems due to the hopelessness people feel at living in what is perceived as an 'open prison' . 58 Furthermore, Indonesia is not a State party to the 1951 Refugee Convention and has no national asylum system. 59 It has also been argued that IOM's facilitation of voluntary return risks amounting to refoulement from Indonesia to their countries of origin insofar as asylum seekers and refugees have few alternatives. 60 Migrants travelling through Mexico, in turn, risk suffering human rights violations committed both by State agents and private actors. Reported abuses include kidnappings, human trafficking, summary executions, enforced disappearances, sexual violence and robbery, assault, and extortion, accompanied by physical and psychological abuses. 61 The high levels of violence and insecurity in Mexico suggest that migrants in Mexico are at risk of harm. Furthermore, reports suggest that the asylum procedures in Mexico do not comply with international human rights standards, and that Mexico violates the principle of non-refoulement by returning asylum seekers to the Northern Triangle countries, which, because of gang warfare and violence, are some of the most dangerous places on earth. 62 Thus, despite reports that the USA and Mexico are discussing the possibility of establishing a bilateral safe third-country agreement, the designation of Mexico as a 'safe third country' for Central American asylum seekers has been contested. 63 In sum, it can be argued that Libya, Indonesia and Mexico fail to offer asylum seekers and refugees the protection they need. More generally, intercepted migrants are at risk of suffering severe human rights abuses in these countries. While the seriousness of the risk and the gravity of the violations may vary, the plight of intercepted migrants in transit suggests that returning someone to those countries-and by extension containing them there-may breach the principle of non-refoulement and other human rights norms.

THE RISK OF HUMAN RIGHTS VIOLATIONS CAUSED BY CONTAINMENT
POLICIES While the previous section discussed various examples of containment policies and their effects on intercepted migrants and refugees, this section examines which violations of human rights and refugee law are at stake in the context of cooperative migration control policies. The principle of non-refoulement in refugee and human rights law, the right to leave and other relevant human rights are discussed in turn.

A. Does the Principle of Non-refoulement Apply to Containment Policies?
The principle of non-refoulement is embodied in Article 33(1) Refugee Convention as well as human rights instruments at the international and regional levels. Some provisions, like Article 3 of the Convention against Torture 64 (CAT), explicitly prohibit refoulement, whereas the principle has been 'read into' other provisions, such as Article 7 of the International Covenant on Civil and Political Rights 65 (ICCPR). Moreover, it is generally accepted that the principle of non-refoulement in international refugee and human rights law has acquired the status of international customary law, at least as regards the prohibition of refoulement to a risk of torture, 66 while some commentators have argued that it also qualifies as jus cogens. 67 Therefore, although Libya and Indonesia are not parties to the 1951 Refugee Convention and/or the 1967 Protocol, 68 they are nevertheless bound by the norm of non-refoulement, both through custom and through human rights treaties to which they are parties. In particular, all States examined here have ratified the CAT and ICCPR. 69 Accordingly, the analysis focuses on the Refugee Convention, CAT and ICCPR, although similar arguments can be made under other refugee and human rights instruments, notably Article 3 of the ECHR in the case of Italy.
The scope of the principle of non-refoulement in international human rights law differs from that in international refugee law in several ways. Crucially, unlike Article 33(1) of the Refugee Convention, the principle of non-refoulement in international human rights law applies to all human beings, regardless whether they are refugees or not, including when they are in their country of origin. Thus, although it may be unclear whether intercepted migrants are refugees (and hence entitled to protection under the Refugee Convention), human rights law protects them in any case. Furthermore, the principle of non-refoulement in international human rights law is absolute, and although the precise scope depends on the exact wording and interpretation of each provision, overall it prohibits return to a risk of torture and cruel, inhuman or degrading treatment or punishment rather than persecution on one of the grounds in the Refugee Convention. 70 The first question that arises in the context of containment policies is whether the principle of non-refoulement applies to persons who have not yet reached the State's territory. As discussed in Section 2, Hirsi confirms that, at least in the European context, it applies extraterritorially when a State exercises jurisdiction. The IAComHR's decision in Haitian Centre for Human Rights observed that Article 33 of the Refugee Convention 'had no geographical limitations' . 71 However, the Prague Airport and Sale rulings as well as statements by the USA contesting the extraterritorial applicability of the principle of non-refoulement 72 suggest this position is not unequivocal. Some Australian case-law, which could be read as supporting Sale, 73 further muddies the waters. Nevertheless, according to the UNHCR, Article 33(1) applies 'wherever a State exercises jurisdiction, including at the frontier, on the high seas or on the territory of another State' . 74 Furthermore, many commentators seem to agree that the principle of non-refoulement applies at least when a State exercises jurisdiction, including extraterritorially. 75 It thus remains unclear whether the principle of non-refoulement applies to containment policies and may in fact depend on the specific treaty provision in question. In particular, a distinction can be made between explicit non-refoulement provisions such as Article 33(1) of the Refugee Convention and Article 3 CAT on the one hand and implicit provisions such as Article 7 of the ICCPR and Article 3 of the ECHR on the other hand.  that he would be in danger of being subjected to torture' (emphasis added). Battjes argues that the wording of these provisions 'seems to exclude their application on the territory of the state where a person fears persecution or is in danger of being tortured' . 76 Likewise, the Court of Appeal in the Prague Airport case observed: Article 33 forbids 'refoulement' to 'frontiers' and, whatever precise meaning is given to the former term, it cannot comprehend action which causes someone to remain on the same side of the frontier as they began; nor indeed could such a person be said to have been returned to any frontier. 77 The then House of Lords did not specify whether it agreed that refoulement cannot take place 'on the same side of the frontier' . The situation in the Prague Airport case is ambiguous because the applicants were also in their country of nationality and therefore outside the scope of the Refugee Convention for that reason. 78 Containment practices, however, concern situations where refugees are intercepted on the territory of a cooperating transit State rather than in their country of origin. Accordingly, the requirement that refugees have left their country of origin does not pose an obstacle to the applicability of the Refugee Convention, unlike in the Prague Airport case, and in any event is irrelevant under Article 3 of CAT. On the other hand, the fact that the interceptions are carried out by agents of the transit State, often on the territory of the transit State itself, makes it more difficult to argue that those intercepted are refouled from the destination State 'to' the transit State. However, interceptions of refugees before they reach the territory of destination States raise questions about States' good faith compliance with Article 33 of the Refugee Convention and similar provisions, as required by Articles 26 and 31 of the VCLT. 79 The majority in Sale thus admitted that gathering and returning refugees 'to the one country they had desperately sought to escape . . . may even violate the spirit of Article 33' . 80 The United Kingdom (UK) House of Lords in the Prague Airport case found that the principle of good faith did not prevent a State from taking steps to control the movements of people outside its borders who wish to travel to it but have not yet reached its frontier. 81 Goodwin-Gill and McAdam nevertheless argue that 'for States to seek to avoid their obligations by contracting them out to other States frustrates the goals of the multilateral treaty regime and is incompatible with the 1951 Convention's object and purpose' . 82 It can therefore be argued that containment policies are incompatible with good faith compliance with the principle of non-refoulement. 76  Thus, in situations where refugees are intercepted on the territory of a transit State by agents of that State, for instance in Indonesia or Mexico, the applicability of the principle of non-refoulement is not straightforward. A literal reading of Article 33(1) of the Refugee Convention and Article 3 of CAT would hold that these provisions do not apply. Likewise, in the case of the Libyan 'pullbacks' , only if the interception were to take place either by the Italian authorities (as was the case in Hirsi) or by the Libyan Coast Guard but within Italy's territorial waters (resulting in return 'to' Libya) would the principle of non-refoulement apply. However, as noted above, such an interpretation raises questions in terms of the principle of good faith.

(ii) Implicit non-refoulement provisions
The ICCPR, like Article 3 of the ECHR, does not contain an explicit non-refoulement provision. Rather, the principle of non-refoulement is conceived as a component part of the broader prohibition on torture and cruel, inhuman or degrading treatment or punishment. The Human Rights Committee thus held that Article 7 of the ICCPR prohibits States from exposing individuals 'to the danger of torture or cruel, inhuman or degrading treatment or punishment upon return to another country by way of their extradition, expulsion or refoulement' . 83 In the European context, the ECtHR famously held in Soering v United Kingdom that extradition is prohibited 'where substantial grounds have been shown for believing that the person concerned, if extradited, faces a real risk of being subjected to torture or to inhuman or degrading treatment or punishment in the requesting country' . 84 It has applied similar reasoning to situations of expulsion and return of asylum seekers. 85 Moreover, the principle of non-refoulement has also been 'read into' other human rights norms, including the rights to life, liberty and a fair trial. 86 As the text of Article 7 of the ICCPR (and Article 3 of the ECHR) itself does not prohibit refoulement, it can be seen to offer more flexibility in terms of its scope and content than explicit non-refoulement provisions. Thus, the absence of the expression 'to the frontiers of territories' or 'to another State' in Article 7 of the ICCPR precludes arguments to the effect that this provision only applies if a border crossing takes place. In fact, the scope of protection under Article 7 of the ICCPR is broader than only non-refoulement, requiring States more generally not to ill-treat individuals within their jurisdiction and to protect them from other sources of harm. The Human Rights Committee has thus held that States parties to the ICCPR must protect against acts of ill-treatment 'whether inflicted by people acting in their official capacity, outside their official capacity or in a private capacity' . 87 The scope of Article 7 of the ICCPR includes issues such as domestic violence and female genital mutilation for instance. 88 Likewise, the ECtHR has repeatedly held that a State may incur responsibility under Article 3 of the ECHR 'where the authorities fail to take reasonable steps to avoid a risk of ill-treatment about which they knew or ought to have known' . 89 Den Heijer thus argues that the prohibition of refoulement under the ICCPR and ECHR can be construed as 'an obligation to shield a person from harm . . . regardless of territorial considerations, provided that a person is within the jurisdiction of a Contracting State' . 90 This harm may result from the conduct of private actors or foreign State agents, either on the State's territory or abroad. The ECtHR has, for instance, held that the transfer of detainees from British to Iraqi custody breached the UK's obligations under Article 3 of the ECHR, although no physical border crossing had taken place, because it exposed the applicants to 'a real risk of being sentenced to death and executed' . 91 This suggests that the prohibition of refoulement applies when a transfer of jurisdiction takes place, regardless whether a physical border is crossed or not. 92 Arguably, if both returning a person to a country where she risks suffering ill-treatment and handing her over to the authorities of that country without leaving its territory can be interpreted as a violation of Article 7 of the ICCPR, preventing her from leaving that country can also be considered to violate that provision, provided the State doing so exercises jurisdiction. Thus, provided it exercises jurisdiction, a destination State like Italy, Australia or the USA could be found responsible for breaching Article 7 of the ICCPR on account of the fact that it exposed migrants or refugees to a risk of harm by preventing them from leaving an unsafe transit country. If this interpretation of Article 7 as a prohibition of exposure to a risk of harm is correct, it also raises the question to what extent it informs the interpretation of explicit non-refoulement provisions such as Article 33(1) of the Refugee Convention and Article 3 of CAT.
In any event, to what extent a State incurs responsibility under this understanding of Article 7 of the ICCPR depends on the factual circumstances of the case. In particular, as will be discussed in Section 4, it depends on which State exercises jurisdiction over the intercepted migrants. Moreover, although it remains debatable whether the principle of non-refoulement itself (unlike the broader prohibition of exposure to a risk of harm in international human rights law) applies to policies of containment by transit States, other human rights norms, notably the right to leave, are also relevant.

B. The Right to Leave: Non-refoulement's Counterpart?
A comprehensive answer to the question whether destination States can circumvent responsibility by 'outsourcing' interceptions to cooperating transit States requires broadening the analysis beyond the principle of non-refoulement. This is not to undermine its importance but rather to look for complementary protection norms which deserve attention in the context of current migration control policies.
As the practices examined in this article have the effect of preventing individuals from leaving transit States, it is worth asking to what extent the right to leave may restrict States' ability to prevent asylum seekers from reaching their territories. Indeed, the right to leave is 'a necessary prerequisite to the enjoyment of a number of other human rights, most notably the right to international protection from torture, inhuman or degrading treatment or punishment' . 93 Insofar as current policies are shifting from returns (from destination States to transit States) to containment (in transit States), the legal analysis can likewise be seen as shifting (or at least extending) from the principle of non-refoulement to (include) the right to leave.
The right to leave is enshrined in various human rights provisions, including Article 12(2) of the ICCPR and Article 2(2) of Protocol No 4 to the ECHR. It is beyond the scope of this article to carry out a detailed analysis of the exact scope of the right to leave. 94 For the purpose of this article, it suffices to note that, although the right to leave is not absolute, the Human Rights Committee has noted that 'restrictions must be provided by law, must be necessary in a democratic society for the protection of these purposes and must be consistent with all other rights recognised in the Covenant' . 95 Furthermore, 'the restrictions must not impair the essence of the right . . . the relation between right and restriction, between norm and exception, must not be reversed' . 96 Commentators have questioned whether it is permissible to restrict the right to leave in order to prevent breaches of other states' immigration laws. 97 More generally, '[m]easures of general nature that limit leaving on a massive scale, cannot be compatible with the right to leave since no assessment has been made as to their proportionality in relation to the specific individuals affected' . 98 Thus, policies designed to prevent migrants, notably refugees, from leaving Libya, Indonesia and Mexico may breach the right to leave.

C. Other Violations of Human Rights and Refugee Law
Last, other human rights, in addition to the principle of non-refoulement and the right to leave, are also relevant in the context of the cooperative migration control policies dis-cussed in this article. Indeed, as noted in Section 2, migrants contained in unsafe transit States may suffer a number of human rights violations. To what extent this is the case varies between countries and depends on the specific circumstances of each individual. However, the list of potential human rights violations in Libya, Indonesia and Mexico includes arbitrary detention, torture and other forms of ill-treatment, slavery, forced labour, summary executions, enforced disappearances and human trafficking. 99 These may amount to violations of inter alia Articles 6, 7, 8, 9, 10 and 12 of the ICCPR.
Whereas the situation in Libya is so dire that arguably any migrant risks suffering severe human rights violations on Libyan territory, the situation in Indonesia and Mexico may be more nuanced. As Indonesia is not a State party to the Refugee Convention and has no national asylum system, refugees' right to protection is not fulfilled in Indonesia. Yet living and detention conditions in Indonesia, particularly in combination with the lack of prospects for improvement, may amount to inhuman and degrading treatment. 100 Likewise, refugees' rights are not fulfilled in Mexico insofar as asylum procedures do not comply with international human rights standards, and Mexico violates the principle of non-refoulement. More generally, it can be argued that preventing migrants in Mexico from reaching the USA exposes them to a risk of abuses such as dire living and/or detention conditions, sexual violence, ill-treatment and killings.

DOES THE DESTINATION STATE EXERCISE JURISDICTION?
The previous section revealed that containment policies may violate the principle of non-refoulement, the right to leave and other human rights. In particular, it showed that destination States have the obligation not to expose individuals within their jurisdiction to the risk of harm. Indeed, the question whether it is the destination State and/or the transit State which exercises jurisdiction is crucial to determining the former's responsibility. Accordingly, this section examines to what extent sponsoring destination States incur responsibility for the violations discussed in Section 3 on the ground that they exercise jurisdiction. The key question is therefore whether the sponsoring State exercises jurisdiction over the migrants and refugees who are intercepted before they reach its territory.
A State is generally considered to exercise jurisdiction over its territory. 101 Thus, to the extent that the interceptions take place on the territory or in the territorial waters of transit States, they will in principle exercise jurisdiction. However, it is commonly accepted that States can also exercise jurisdiction outside their territories. The Human Rights Committee has thus famously held that 'it would be unconscionable to so 99 See also Dastyari and Hirsch, supra n 35. 100 By way of comparison, the ECtHR has found that a situation of extreme material poverty for asylum seekers could amount to a violation of Article interpret the responsibility under Article 2 of the Covenant as to permit a State party to perpetrate violations of the Covenant on the territory of another State, which violations it could not perpetrate on its own territory' . 102 Indeed, both the transit and the destination States can exercise jurisdiction simultaneously. Therefore, although the territorial jurisdiction of the transit State is relevant for determining whether the destination State exercises extraterritorial jurisdiction, 103 destination States cannot argue that the fact that another State exercises jurisdiction per se means that they do not. 104 Extraterritorial jurisdiction under human rights law is commonly understood as being triggered either by the exercise of effective control over territory or authority or control over a person abroad. 105 However, the current migration control policies implemented by Italy, the USA and Australia often do not involve direct contact between their authorities and the intercepted individuals. The fact that migrants and refugees are intercepted by transit States on behalf of destination States rather than by destination States themselves complicates the jurisdictional assessment. As noted by Hathaway, 'it is far from clear that a state can be said to exercise jurisdiction by the simple issuance of policies intended to apply extraterritorially, but which are wholly implemented by third parties operating inside the sovereign territory of another state' . 106 In fact, it has been argued that containment policies are not only specifically intended to circumvent the principle of non-refoulement but also jurisdiction (and hence obligations) under human rights law more generally. 107 However, recent developments in human rights case-law and scholarship suggest that the absence of physical control over intercepted migrants does not necessarily preclude sponsoring destination States from exercising jurisdiction. The following analysis, although by no means exhaustive, discusses various understandings of jurisdiction which suggest that sponsoring States may be less successful at avoiding jurisdiction than might seem at first sight. Indeed, it has been argued that the ECtHR's jurisprudence reveals a trend towards a more expansive approach to the Convention's extraterritorial application, 108 while UN bodies and other regional bodies such as the Inter-American Commission on Human Rights have arguably always included a more expansive understanding of extraterritorial jurisdiction. 109 The following analysis relies heavily on the case-law of the ECtHR because it has the most extensive and detailed body of caselaw on jurisdiction and is therefore particularly relevant for Italy. 110 However, to the extent that the ECtHR's case-law informs the understanding of jurisdiction beyond the European context, similar arguments can be made as regards the support provided by Australia and the USA to Indonesia and Mexico, respectively.
In the specific context of interceptions at sea, in Hirsi the ECtHR found that Italy exercised jurisdiction over the intercepted applicants because they were 'under the continuous and exclusive de jure and de facto control of the Italian authorities' . 111 Other case-law of the ECtHR and Committee against Torture confirms that control over a ship's passengers triggers jurisdiction. 112 However, other ECtHR cases suggest that jurisdiction can be triggered without making physical contact with intercepted passengers. 113 Likewise, the Human Rights Committee and Committee against Torture found that Australia exercised jurisdiction over offshore regional processing centres despite the fact that Australia did not exercise physical control over their inmates. Rather, they based their finding on the fact that Australia established the centres, financed them, transferred asylum seekers to them and chose which private contractors ran the centres. 114 A case that was recently brought to the ECtHR regarding Italy's involvement in Libyan 'pullback' operations will hopefully help clarify whether Italy's support for Libyan 'pullback' operations triggers its jurisdiction or not. 115 In any event, various pronouncements by human rights treaty monitoring bodies suggest that a destination State can exercise jurisdiction over intercepted migrants and refugees even when it does not exercise physical control over them.
Moreover, existing case-law recognises that States can also breach the human rights of individuals outside their territory in more immaterial ways than through the exercise of physical control over a person. This is the case for 'acts that a state controls and carries out on its own territory, leading to a human rights violation on the territory of another state' . 116 At the European level, the ECtHR has repeatedly held that 'acts of the Contracting States performed, or producing effects, outside their territories can constitute an exercise of jurisdiction by them within the meaning of Article 1 of the Convention' 117 This is the case if executive or legislative measures have 'direct and immediate' effects beyond their territory. 118 This strand of case-law arguably includes cases in which a person is killed abroad without being arrested and detained first, 119 as well as cases in which a person comes within the jurisdiction of another State than the one on whose territory she finds herself because of a legal rather than a physical act on the part of the acting State. 120 Den Heijer thus notes that in various cases, 'the ECtHR accepted that the Convention applied to executive or adjudicative measures which were specifically directed at persons resident abroad' . 121 Various decisions of the Human Rights Committee and Inter-American Commission on Human Rights can also be seen to fall in this category. 122 Furthermore, in various cases a State was found to have exercised jurisdiction on account of the acts of its diplomatic and consular agents abroad. 123 Altwicker thus suggests extending the jurisdiction test to include control over situations with extraterritorial effects on the enjoyment of human rights, provided there is a sufficient jurisdictional link between the situation controlled by a contracting state and the affected individual. 124 Following a similar reasoning, sponsoring destination States could be found to exercise jurisdiction on account of the effects that their policies have on the rights of intercepted migrants. Indeed, it has been argued that a State ordering or encouraging another State's authorities to block or pull back migrants and that the provision of financing, training and/or equipment for migration control by third States can trigger the sponsoring State's jurisdiction, provided there is a sufficiently direct link between the funding and the breach of the obligation. 125 Furthermore, Gammeltoft-Hansen and Hathaway argue that the ECtHR's judgment in Al-Skeini suggests that 'in addition to the territorial and personal control bases for establishing jurisdiction, states may also be found to have jurisdiction where they exercise public powers abroad' . 126 They identify three requirements that must be fulfilled and argue that these are often met in cases where States engage in extraterritorial migration control: the legal authority of the extraterritorial State must be established in accordance with custom, treaty or other agreement; the State activity must qualify as a public power that is normally exercised by the government of the territorial State; and the human rights violation must be attributable to the extraterritorially acting state, rather than to the territorial state. 127 In other words, in their view the exercise of migration control functions beyond a State's territory is often sufficient to establish that the extraterritorially acting State exercises jurisdiction under international human rights law. 128 However, referring to Articles 4 to 8 of the Draft Articles on the Responsibility of States for Internationally Wrongful Acts and the 'effective control' test, Guild and Stoyanova remark that it is questionable whether the provision of equipment, training, money or intelligence meets the third (attribution) requirement, particularly in light of the fact that attribution to the transit State must be excluded. 129 To the extent that this approach also applies to the USA and Australia, it thus remains unclear whether containment policies trigger the jurisdiction of destination States on the ground that they exercise public powers abroad.
In the context of Italian-Libyan cooperation, various authors have also referred to the ECtHR's Transnistria case-law. Guild and Stoyanova thus note that 'the Libyan Coast Guard would not operate "if it were not for the support that Libya receives" and that this support has 'decisive influence' . 130 Indeed, in Ilaşcu and Catan the ECtHR held that Russia exercised jurisdiction over the Moldovan Republic of Transdniestria because the latter remained 'under the effective authority, or at the very least under the decisive influence, of the Russian Federation, and in any event that it survives by virtue of the military, economic, financial and political support given to it by the Russian Federation' . 131 Moreno-Lax and Giuffré likewise argue that funding, training and equipping, which is 'explicitly conditioned on [Libya] "managing" migratory flows and impeding exit for transit towards Europe, can be said to constitute a form of "decisive influence"' . 132 They further highlight destination States' duty to prevent human rights violations of which they knew or ought to have known and the fact that these States are 'in a position to avoid the possibility of ill-treatment from materialising' . 133 Applying a similar reasoning to the conduct of Australia and the USA, one could likewise argue that they exercise jurisdiction and incur responsibility because the support they provide to Indonesia and Mexico, respectively, in the form of funding, training and equipment amounts to 'decisive influence' .
Last, Jackson argues for an expansive understanding whereby jurisdiction may arise due to the complicity of an ECHR Member State in breaches committed by a third State. 134 Under Jackson's proposed interpretation, Soering prohibits States from being complicit in torture by providing the principal State with the person of the potential victim, yet 'there is no good reason to confine its application to one very specific form of complicity' . 135 Likewise, the Human Rights Committee observed in Munaf that a State party may be responsible 'if it is a link in the causal chain that would make possible violations in another jurisdiction' . 136 Davitti and Fries argue that under this understanding of jurisdiction, Italy could be found to exercise jurisdiction over (hypothetical) offshore processing of asylum claims in African States. 137 Likewise, Jackson's interpretation of jurisdiction could trigger destination States' jurisdiction when they support transit States in intercepting refugees on their behalf through the provision of equipment, training and funding.
In sum, the foregoing discussion sought to illustrate that it remains unclear whether destination States exercise jurisdiction over migrants and refugees intercepted by transit States on their behalf. If jurisdiction is understood as physical control over an individual, this is unlikely to be the case, but if one accepts that States can affect human rights beyond their borders in other ways, containment policies may trigger the jurisdiction of sponsoring States. Crucially, whether a destination State like Italy, Australia or the USA exercises jurisdiction depends on one's understanding of jurisdiction but also on the specific facts of the case. If it does, it could incur responsibility for breaching the principle of non-refoulement or, more generally, for exposing migrants to a risk of harm in the transit State, for breaching the right to leave and possibly also for other violations suffered in the transit State. Nevertheless, the absence of jurisdiction does not preclude a finding that sponsoring States incur responsibility for the plight of migrants and refugees intercepted in unsafe transit States on their behalf, as discussed in the next section.

DOES THE DESTINATION STATE INCUR DERIVED RESPONSIBILITY BECAUSE IT IS COMPLICIT?
The previous sections suggested that if sponsoring States such as Italy, the USA and Australia exercise jurisdiction, they may also breach the principle of non-refoulement and/or other human rights such as the right to leave. Does this mean, however, that, as long as they do not exercise jurisdiction over intercepted migrants, they do not incur responsibility? The answer to that question lies in the broader field of public international law and more specifically the law on State responsibility. Wrongful Acts 138 ('the Articles on State Responsibility') are commonly accepted as 'an expression of the customary law of state responsibility' 139 and 'an authoritative formulation of international law relating to international responsibility' . 140 International courts and tribunals, including the ECtHR, refer extensively to the Articles on State Responsibility as an authoritative statement of the law on State responsibility. 141 They can therefore be used as reference point for the following analysis. Under the international law, a State is prohibited from aiding or assisting another State in committing an internationally wrongful act. This complicity rule is embodied in Article 16 of the Articles on State Responsibility: A State which aids or assists another State in the commission of an internationally wrongful act by the latter is internationally responsible for doing so if: In other words, if Article 16 of the Articles on State Responsibility applies, the transit State incurs international responsibility for human rights violations, while the sponsoring State incurs responsibility for the aid or assistance provided. Although there seems to be general acceptance that this rule is part of customary international law, its exact contours remain unclear. 142 There is both a lack of jurisprudence and considerable discussion among commentators regarding its interpretation. 143 It is worth examining the requirements of the complicity rule in the context of containment policies and discussing to what extent it limits sponsoring States' ability to circumvent the principle of non-refoulement. The first question that arises is whether the conduct of a transit State like Libya, Indonesia or Mexico constitutes an internationally wrongful act. This is the case if it is responsible for any of the violations examined in Section 3 as regards migrants and refugees within its jurisdiction. As discussed above, it remains debatable whether the principle of non-refoulement applies when transit States intercept migrants on behalf of destination States. However, it may be easier to establish that such interceptions breach the right to leave and the broader prohibition of exposure to harm embodied in Article 7 of the ICCPR, as well as specific human rights norms such as the right to life and the right to freedom of movement. Whether this is the case depends on the specific circumstances of each individual.
Assuming that the cooperating transit State is responsible for a violation of refugee or human rights law, the next question is whether the requirements of Article 16 of the Articles on State Responsibility are met. The forms of support provided by Italy, Australia and the USA, which include training, financing and equipping, probably qualify as aid or assistance. 144 Furthermore, Article 16 of the Articles on State Responsibility requires a sufficiently close causal link between the support provided and the violation committed by the transit State. Insofar as destination States provide support with the explicit aim of preventing migrants and refugees from leaving transit States, this requirement is easily fulfilled as regards the right to leave. Thus, for instance, the fact that the vessels provided by Italy to Libya have been used to return refugees to Libya probably qualifies as a sufficiently close nexus. 145 However, it is more difficult to show that there is a sufficiently close link between the support provided by destination States and other human rights abuses suffered by migrants in transit States. Does the provision of equipment and training assistance by the USA to the Mexican National Migration Institute cause human rights violations in Mexico? Does Australian support under the Regional Cooperation Arrangement cause violations of refugee rights in Indonesia? Whether this is the case depends on the assistance provided, the violation at stake and the required closeness of the causal link. 146 In situations where destination States provide assistance not only for the interception of migrants and refugees but also for their accommodation and protection in transit States, as is the case in Indonesia and Libya for instance, it will be easier to argue that there is a sufficiently close causal link between the destination State's assistance and the human rights violations in the transit State.
The requirement in Article 16(a) that the sponsoring destination State has 'knowledge of the circumstances of the internationally wrongful act' also deserves close attention. Indeed, the Commentary specifies that 'the aid or assistance must be given with a view to facilitating the commission of that act' 147 , which raises the question whether Article 16 requires intent and/or knowledge on the part of the complicit State. Although the matter is not settled in terms of law, it has been suggested that the adequate standard is 'knowledge or virtual certainty that the recipient State will use the assistance unlawfully' . 148 Indeed, Crawford argues that intent may be imputed if aid is given 'with certain or near-certain knowledge as to the outcome' , 149 while Jackson argues that the relevant standard is that of 'awareness with something approaching practical certainty as to the circumstances of the principal wrongful act' . 150  'knowledge of the circumstances' standard defended by Lanovoy. 151 Aust, in turn, holds that the complicit State does not need to wish for the outcome of the internationally wrongful act; rather, it must intend to contribute to its commission. 152 He further suggests that in international human rights law, the intent standard could be modified into a due diligence standard. 153 Depending on the specific facts of the case, this standard is probably met in the context of containment policies, notwithstanding the fact that policies are often framed as being designed to prevent arrivals and save lives. 154 Since transit States intercept asylum seekers on behalf of destination States, it is difficult for the latter to claim that they did not know that their assistance would be used to prevent migrants and refugees from reaching their territories and thereby expose them to a risk of harm. Thus one can argue that Italy, Australia and the USA provide support to Libya, Indonesia and Mexico, respectively, with virtual certainty that this support would be used in breach of migrants' and refugees' right to leave.
Furthermore, Gammeltoft-Hansen and Hathaway argue that the relevant standard is that of constructive knowledge: under this interpretation the requirement of Article 16(a) is met also as regards other human rights violations because sponsoring States 'knew or should have known' that their support would lead to such human rights violations. Arguably, even under a strict interpretation of Article 16(a) of the Articles on State Responsibility, the intent requirement is met as regards Article 7 of the ICCPR if one understands this provision as proscribing exposure to a risk of harm. Indeed, in that case it is sufficient to show that migrants are knowingly exposed to a risk of severe harm for the intent requirement to be met. 155 However, if intent rather than (constructive) knowledge is required, beyond the right to leave and a broad understanding of nonrefoulement as exposure to a risk of harm, it would be more difficult to demonstrate that Italy, Australia and the USA intended intercepted migrants to suffer human rights violations. 156 Last, Article 16(b) of the Articles on State Responsibility requires that the violation would be internationally wrongful if committed by the sponsoring State. This raises the question whether it is sufficient for both States to be bound by the norm in question or whether this obligation needs to stem from the same provision for both States. 157 As noted earlier, all States are parties to the ICCPR and bound by the customary principle of non-refoulement. Therefore, this requirement is met at least under these two sources.
In sum, even if they do not exercise jurisdiction over intercepted migrants and refugees, destination States like Italy, Australia and the USA may incur derived respon-sibility under Article 16 of the Articles on State Responsibility for the human rights violations they suffer. Indeed, various authors have argued that Italy's support for Libya meets the requirements of Article 16, including the requirement of intent. 158 Dastyari and Hirsch conclude that Australia is also responsible under Article 16 of the Articles on State Responsibility for the support it provides to Indonesia. 159 Moreover, this provision may be of particular importance with regard to the USA, in the light of its statements to the effect that the ICCPR and Refugee Conventions do not apply extraterritorially and that non-refoulement is not a norm of customary international law.

CONCLUSION
This article discussed the rise of cooperative migration control policies and highlighted three specific examples of containment practices whereby a cooperating transit State intercepts migrants, including refugees, on behalf of a sponsoring destination State: Italy-Libya, Australia-Indonesia and the USA-Mexico. The analysis showed that it remains unclear whether the principle of non-refoulement in international refugee law applies to situations of containment. It also revealed, however, that human rights law may offer more protection to intercepted migrants and refugees. Indeed, the principle of non-refoulement in human rights law applies to all intercepted migrants, regardless whether they are refugees or not, and arguably its scope is wider insofar as it prohibits exposing a person to a risk of harm in a broader sense. Moreover, human rights law further protects migrants through the right to leave and other relevant norms such as the prohibition of torture and other forms of ill-treatment and arbitrary detention. The analysis further suggested that sponsoring destination States could incur responsibility for violations of the principle of non-refoulement and other human rights norms either because they exercise jurisdiction over the intercepted migrants or because they provide aid and assistance to transit States. To what extent this is the case depends on the specific facts of the case, including what support is provided and what violations are alleged, and on the interpretation of key concepts, notably jurisdiction and complicity.
The analysis thus suggests that, although destination States may argue that the principle of non-refoulement does not apply when transit States intercept migrants on their behalf, this does not mean that such practices enable them to circumvent responsibility under international law altogether. Indeed, the responsibility of Italy, Australia and the USA must be conceived not only in light of developing understandings of the principle of non-refoulement but also with reference to international human rights law and the law of State responsibility. Arguably, contemporary cooperative migration control policies, which focus on containment rather than return, require us to revisit the principle of non-refoulement, in terms of both its scope and its place in the refugee protection regime. This article has thus argued that, while it remains a key principle of refugee and human rights law, containment policies ask for a broader focus of analysis, where non-refoulement is only one rule of international law that constrains States trying to prevent refugees and migrants from reaching their territories. Furthermore, as the 'centre of gravity' of migration control moves from destination States to transit and