Whilst we await the outcome of the Rwanda challenges in the High Court, this is an opportune moment to consider whether ‘externalisation’ and ‘safe third country’ practices are compatible with international refugee law.

1. Conceptualising ‘externalisation’ and ‘safe third country’

UNHCR has defined ‘externalisation’ as “[m]easures preventing asylum-seekers from entering safe territory and claiming international protection, or transfers of asylum-seekers and refugees to other countries without sufficient safeguards…”.

‘Safe third country’ concepts generally revolve around the one core legal issue, namely whether a State can deny protection on grounds that it is available elsewhere. Not all ‘safe third country’ measures involve ‘externalisation’.[1] Some ‘safe third country’ measures involve an agreement for a receiving State to process asylum claims i.e. an externalised form of a ‘safe third country’ measure e.g. the Australia’s Memorandum of Understanding (“MOU”) with Nauru and  Papua New Guinea. Other forms of ‘safe third country’ agreements on the other hand, include a resettlement element e.g. the Libya-Niger Emergency Transit Mechanism or the EU-Turkey Statement.

Given the plethora of policies across industrialised States, how does one begin to decipher whether such practices are compatible with international refugee law?

2. Compatibility of ‘externalisation’ and ‘safe third country’ concepts with international refugee law

There is surprisingly little consensus regarding the compatibility of ‘externalisation’ and ‘safe third country’ concepts with international refugee law.[2]  One fundamental question is whether international refugee law permits such practices at all? The 1969 Vienna Convention on the Law of Treaties and its provisions on treaty interpretation require the Refugee Convention’s text to be given its “ordinary meaning” in light of its “object and purpose”.  Unfortunately, the Convention’s text neither explicitly authorises nor prohibits such practices. 

The UNHCR Executive Committee has not condemned such practices in its Conclusion but has instead emphasised the existence of appropriate safeguards, giving the inference of potential compatibility of such practices with international refugee law.[3] Two obvious questions arise. Under what circumstances would such practices be compatible with international refugee law and secondly, assuming compatibility, what, if any constraints, operate on States in carrying out these practices?

One clear constraint is the non-refoulement principle. Secondly, there is also a compelling argument that such practices must be compatible with all the rights set out in the Refugee Convention (Articles 2-34).[4]

With these parameters in mind, this article turns to consider the different contexts in which ‘externalisation’ and ‘safe third country’ practices have arisen, firstly in border controls at land and sea and secondly, offshore processing, and whether these practices have been assessed to be compatible with international refugee law.

3. Land Border ‘Pushbacks’

One border control measure that has become increasingly widespread is land border ‘pushbacks’. Such practices involve intercepting and summarily forcing back persons arriving at the land border without assessing their claims for entry or protection. An externalised form of land border ‘pushbacks’ involve carrying out or assisting ‘pushbacks’ from outside the States’ own territories, thus preventing a border crossing in the first place.[5] According to the Report of the Special Rapporteur on Torture, both types of ‘pushbacks’ raise serious questions of compatibility with international refugee law because of a real risk of breaching the non-refoulement principle.

To illustrate, the Grand Chamber of the ECtHR in Ilias and Ahmed held that Hungary had failed to provide effective access to an asylum procedure by carrying out a ‘pushback’ of the applicants into Serbia. Hungary, in declaring Serbia a ‘safe third country’ did not take account of publicised information that those returned to Serbia would be denied effective access to an asylum procedure and at real risk of summary removal to Macedonia. The ECtHR found that the risk of chain refoulement could have been alleviated if the applicants’ return had been negotiated with Serbia as opposed to a ‘pushback’ across the border.[6]

Contrast this with N.D. and N.T., where the Grand Chamber of the ECtHR held that Spain’s decision in summarily forcing back N.D. and N.T. without assessing their claims was a consequence of the applicants’ “culpable conduct” in their unauthorised arrival in Spain.[7] Do unauthorised arrivals permit ‘pushbacks’? The ECtHR was clear that States may not disregard their non-refoulement obligations when carrying out border controls.[8] Perhaps one way of explaining the negative outcome in N.D. and N.T. is that it should be confined to its facts. Both applicants had shown no intention to apply for international protection at designated places along the border and provided no cogent reasons why they had not applied for international protection in the time that they had spent in the camp on the Moroccan side of the Spanish-Moroccan border.[9] N.T. did not claim to be in need of protection and N.D. was found not to be in need of protection after an asylum procedure.[10]

In a recent tragic ‘pushback’ case of M.H. and Others, Croatia attempted to argue the applicants, an Afghan family of fourteen, whose 6 year old child had been hit and killed by a train when they were summarily returned to Serbia, had engaged in “culpable conduct” by circumventing procedures for legal entry into Croatia. The ECtHR held that Croatia failed to provide the applicants with any genuine and effective access to procedures for legal entry into Croatia and the measures violated the Right to Life (Article 2 of the ECHR) and the prohibition on collective expulsion of aliens (Article 4 Protocol No. 4).

Following the ECtHR’s line of reasoning, it remains the case that land ‘pushbacks’ without a substantive assessment of whether the third country is ‘safe’ and without guarantees of access to asylum procedures in the ‘safe third country’ will breach international refugee law because nonrefoulement cannot be guaranteed.

4. ‘Pushbacks’ at Sea or Interdiction

‘Pushbacks’ at sea or interdiction on the high seas have shifted from direct to indirect forms over the past three decades, now increasingly performed by third countries on behalf of destination States. Interdiction on the high seas is a form of externalised border control. Pursuant to international treaties and the UN Convention on the Law of the Sea, there are limited permitted reasons for interdiction given freedom of navigation limits interception rights. One permitted reason can be found in Article 8(7) of the Smuggling Protocol but such action must still take account of the rights of individuals pursuant to the Refugee Convention (Article 19(1)). Therefore, it can be said that State practices on the high seas preventing asylum applicants from entering its territorial sea such as ordering a boat to modify its course; escorting a boat to a third country and/or handing over passengers to a third country whereby they may face a risk of refoulement are incompatible with international refugee law.[11]

Early arguments often put forward by States in favour of preventing asylum-seekers from entering the State’s territorial sea and claiming international protection is that the principle of non-refoulement only protects those already in the State’s territory, a view upheld by the US Supreme Court in Sale. In a powerful dissenting judgment, Blackmun J held that Article 33 of the Convention is clear on the parameters of non-refoulement, “It limits only where a refugee may be sent “to,” not where he may be sent from.” Although Stevens J in Sale acknowledged the “moral weight” of the argument that the Refugee Convention’s goals mean non-refoulement applies “whether or not the refugees are within that nation’s borders”, the undeniable legacy of Sale is that many State parties to the Refugee Convention have adopted measures to assign protective responsibility over refugees to other States.[12] These measures, whilst having the potential to enhance responsibility sharing, have in practice, raised serious questions regarding its compatibility with international refugee law.[13]

The proposition that non-refoulement is territorially bound has since been rejected by UNHCR’s Executive CommitteeHaitian Centre for Human Rights et al. and by the ECtHR Grand Chamber in Hirsi Jamaa et. al. v Italy, a case concerning the interdiction outside of Italy’s Search and Rescue region of responsibility, of some 200 migrants by the Italian Revenue Police and the Coastguard and returning them to Libya.

In Hirsi, the ECtHR grouped instances of extraterritorial jurisdiction into 3 classes: effective control over an area outside national territory; control and authority over an individual abroad; and “other instances” involving the activities of “diplomatic agents … on board craft and vessels registered in, or flying the flag of, that State.”[14] The ECtHR held that the question of whether a State was exercising jurisdiction extraterritorially must be determined by whether there was “full and exclusive control over a prison or a ship.” The ECtHR then considered its decision in Medvedyev where in that case, the Winner, a vessel flying the Cambodian flag on the high seas, was suspected of trafficking illicit drugs, was boarded and its crew placed under the control of French military personnel. In Medvedyev, the ECtHR had concluded there was “de facto continued and uninterrupted control” from the time of the French military’s interception.[15]

By contrast, Hirsi concerned three flagless ships.[16] The ECtHR emphasised the customary international law rule that “a vessel sailing on the high seas is subject to the exclusive jurisdiction of the State of the flag it is flying.” Therefore, given the applicants in Hirsi were transferred on board military ships flying the Italian flag, the ECtHR held “the applicants were under the continuous and exclusive de jure and de facto control of the Italian authorities.” The ECtHR added that “Italy cannot circumvent its “jurisdiction” under the Convention by re-characterising the events in issue as rescue operations on the high seas.” The ECtHR determined unanimously that the ‘pushbacks’ at sea was in breach of Italy’s ECHR obligations under Article 3, Article 13 and Article 4 Protocol No. 4.[17]

With ‘pushbacks’ at sea banned, post-Hirsi, State practices have shifted to more indirect forms of interdiction. Italy through its 2008 Treaty of Friendship and 2017 MOU with Libya has invested in a Libyan Search and Rescue capacity so Libya can assume responsibility for rescue and disembarkation.[18] The ‘pullbacks’ of vessels off the Libyan coast suspected of migrant smuggling has the legal support of the UN Security Council and is funded by Italy and the EU.

The important doctrinal issue is whether Italy’s ‘pullback’ of migrant boats, orchestrated through a proxy State (Libya) has a sufficient degree of “effective de facto control” to reach the jurisdictional threshold of Article 1 ECHR. It is an argument that will be tested in SS v Italy, currently pending before the ECtHR. The interveners’ (International Commission of Jurists et al) argument is that Italy de facto commands the Libyan Coast Guards Search and Rescue/ interdiction response thus engaging the jurisdictional threshold of Article 1 ECHR.

Is “de facto control” assumed per se in every migrant boat situation outside of Italy’s Search and Rescue zone because of Italy’s funding for the Libyan Coast Guards? For example, in the case of the “Left-to-Die-Boat” incident of March 2011, scholarly opinion is that there may well be no responsible EU Member State accountable for the loss of lives due to the lack of jurisdiction. If there is no jurisdiction, does it follow that no State bears any legal responsibility for the loss of lives? Gammeltoft-Hansen and Hathaway argues that international law will hold States responsible for aiding or assisting another State’s wrongful conduct.[19] This is most clearly set out in Article 16 of the 2001 International Law Commission Draft 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: (a) That State does so with knowledge of the circumstances of the internationally wrongful act; and (b) The act would be internationally wrongful if committed by that State.”

It is arguable that States who provide maritime patrol vessels and/or directly fund migration control efforts that assist another State in ‘pullbacks’ which lead to ill-treatment of migrants and drowning is aiding or assisting another State’s wrongful conduct.[20] These practices raise profound legal and moral concerns and if left unchallenged, will dismantle the fundamental principles of international refugee law. As Naas, Head of Advocacy at the Search and Rescue NGO Sea-Watch observes, “Imagine a boat in distress with 90 people … all European or U.S. passport holders. Can you imagine what an impressive flotilla would search day and night for the boat in distress, supported by military and helicopters…? This reality is the greatest demasking of the so-called European values.” Perhaps, rather than funding Libya without conditions attached to its Search and Rescue functions, Italy (and the EU) should instead consider the “moral weight” of reinvesting the millions of Euros earmarked for Libya, towards a functioning Search and Rescue, thereby discharging its international refugee law obligations.

5. Offshore processing

Australia’s regional processing centre agreements with Nauru and Papua New Guinea is a form of externalised ‘safe third country’ practice. Nauru and Papua New Guinea have been used on and off as offshore regional processing centres since 2001 despite the fact that neither had acceded to the Refugee Convention at the time. In 2002, the UN Working Group on Arbitrary Detention expressed concerns at the lack of an effective remedy for detainees held at these regional processing centres. In 2013, the UN Human Rights Committee found multiple violations of international law in Australia’s treatment of refugees. The case concerned nearly fifty refugees including five children held in indefinite immigration detention because of adverse security assessments by Australia’s security agency, ASIO. Their detention cannot be challenged in an Australian Court. The HRC held that Australia’s indefinite detention and lack of procedural rights for the applicants violated international human rights law.[21] Australia’s externalised ‘safe third country’ practices failed to guarantee Refugee Convention rights (Article 2-34) and penalised refugees for their mode of arrival. Australia’s regional processing centre model has now been emulated by the UK in the UK-Rwanda MOU. It remains to be seen what the High Court will make of the general and specific challenges in those cases.

6. Conclusion

Many of the ‘externalisation’ and ‘safe third country’ practices discussed above have been held to be incompatible with international refugee law. Some arguments (‘pullbacks’ by proxy) have yet to be tested and scholarly opinion is divided on the availability of legal accountability. However, ‘externalisation’ policies and ‘safe third country’ practices can potentially be compatible with international refugee law. Perhaps the time has come to acknowledge the need or “moral weight” for State practices which are constrained by obligations under the Refugee Convention (Articles 2-34)?




[1] Cantor, D., Tan, N. F., Gkliati, M., Mavropoulou, E., et al., ‘Externalisation, Access to Territorial Asylum, and International Law’, International Journal of Refugee Law, 28 June 2022, at p23

[2] Foster, M. ‘Protection elsewhere: the legal implications of requiring refugees to seek protection in another state’, Michigan Journal of International Law 28(2) 2007, pp.223–286, at 285

[3] Costello, C. ‘Safe country? Says who?’, International Journal of Refugee Law 28(4) 2016 pp.601–622, at p606-607

[4] Freier, L.F., Karageorgiou, E. and Ogg, K., ‘The Evolution of Safe Third Country Law and Practice’, Chapter 28, ‘The Oxford Handbook of International Refugee Law’ (Oxford University Press 2021) (“Oxford Handbook”), pp518-534 at 520

[5] Cantor, D., Tan, N. F., Gkliati, M., Mavropoulou, E., et al, op. cit., at  13

[6] Ilias and Ahmed v Hungary (Application no. 47287/15), European Court of Human Rights, Grand Chamber, 21 November 2019, at [159-161]

[7] N.D. and N.T. v Spain (Applications nos. 8675/15 and 8697/15), European Court of Human Rights, Grand Chamber, 13 February 2020, at [231], [242], [244]

[8] N.D. and N.T. op. cit. at [232]

[9] ibid., at [207], [209-210], [212], [227]

[10] ibid., at [29-30]

[11] Moreno-Lax, V., ‘Protection at Sea and the Denial of Asylum’, Chapter 26, Oxford Handbook, op. cit., pp483-501 at 484, 488 – 489

[12] Koh, H. H., YLS Sale Symposium: Sale’s Legacies, 17 March 2014 http://opiniojuris.org/2014/03/17/yls-sale-symposium-sales-legacies/

[13] Foster, M. op. cit., at 285

[14] Hirsi Jamaa and Others v Italy (Application no. 27765/09), European Court of Human Rights, Grand Chamber, 23 February 2012, at [73-75]

[15] See Hirsi Jamaa and Others v Italy, op. cit.,[80]; see Medvedyev and Others v. France, (Application no. 3394/03) European Court of Human Rights, Grand Chamber, ECHR 2010, at [66-67]

[16] Moreno-Lax, V. ‘Hirsi Jamaa and Others v Italy or the Strasbourg Court versus extraterritorial migration control?’, Human Rights Law Review 12(3) 2012, pp.574–598, at p581

[17] Hirsi Jamaa and Others, op. cit., at [76-79], [81], [133-138]

[18] Moreno-Lax, V., The Architecture of Functional Jurisdiction: Unpacking Contactless Control—On Public Powers, S.S. and Others v. Italy, and the “Operational Model” German Law Journal (2020), 21, pp. 385–416, at 390-391

[19] Gammeltoft-Hansen, T and Hathaway, J. C., ‘Non-Refoulement in a World of Cooperative Deterrence’ (2015) 53 Colum J Transnat’l L 235 – 284, at 277

[20] ibid., at 279

[21] McAdam, J. ‘Australia and asylum seekers’, International Journal of Refugee Law 25(3) 2013 pp.435–48, at 444