The Migration and Economic Development Partnership (MEDP) concluded between the United Kingdom and the Republic of Rwanda in April 2022 represents one of the most far-reaching attempts by a European state to externalise responsibility for determining asylum claims. Under the partnership, certain asylum seekers arriving irregularly in the UK would be removed to Rwanda, where their protection claims would be determined under Rwandan law and, if successful, refugee status would be granted there rather than in the UK ([6]). The scheme was accompanied by substantial financial payments and framed by the Government as a means of deterring “dangerous and illegal journeys” across the Channel ([5]).¹

From a domestic legal perspective, the Rwanda scheme sat at the intersection of statutory powers of removal, the Human Rights Act 1998 (HRA), and the United Kingdom’s obligations under international refugee and human rights law.² Those powers were further re-configured by subsequent legislation, most notably the Illegal Migration Act 2023 (IMA) and the Safety of Rwanda (Asylum and Immigration) Act 2024 (Rwanda Act), which together sought both to mandate removal and to declare Rwanda safe as a matter of law.³ The resulting constellation of statutes, treaties, and non-binding arrangements generated acute questions about the limits of legislative and executive authority in the face of entrenched norms such as non-refoulement.

In R (AAA) v Secretary of State for the Home Department the Supreme Court unanimously held that removals to Rwanda under the original 2022 partnership would be unlawful,⁴ because serious deficiencies in Rwanda’s asylum system created a real risk of refoulement, contrary to Article 3 of the European Convention on Human Rights (ECHR) and the UK’s obligations under the 1951 Refugee Convention.⁵ The judgment has been widely welcomed as a robust affirmation of non-refoulement. Yet this reading understates its constitutional significance.

This article advances three claims. First, it argues that R (AAA) is best understood as a case concerning the interpretation and limits of statutory removal powers, rather than merely the unlawful implementation of a particular policy. The Court did not treat the case as turning on a factual misjudgement alone; instead, it framed the principle of non-refoulement as a structural constraint on the scope of Parliament’s delegation to the executive. Secondly, the article contends that the Court’s reframing of “safety” as a systemic rather than case-specific condition marks a significant development in non-refoulement law. By assessing whether the receiving state’s asylum system possesses adequate institutional capacity and procedural safeguards, rather than evaluating individual risk, the Court raised the threshold for safe third countries. This systemic approach has implications beyond Rwanda, potentially requiring governments to demonstrate functional, rights-respecting asylum infrastructures before transfers can proceed, thereby reshaping how courts scrutinise future deportation arrangements under domestic non-refoulement obligations. Thirdly, it develops an independent framework for identifying the structural tensions inherent in third-country transfer schemes and then uses that framework both to critique the Government’s legal case and to sketch the conditions under which any future externalisation scheme might be compatible with constitutional principle.

Methodologically, the article adopts a doctrinal approach, interpreting domestic statutes and case law in light of international norms.⁶ It treats R (AAA) and the Rwanda Act as a constitutional exchange between the judiciary and legislature over the legal status of non-refoulement and the allocation of institutional responsibility for its enforcement, examining how this interaction revealed fundamental tensions about constitutional boundaries, separation of powers, and the implementation of international refugee protection obligations in domestic law.

A. The pre-Rwanda framework

The statutory power to remove non-nationals from the UK has long been conferred in broad terms. The Immigration Act 1971 and subsequent legislation authorise removal to a variety of destinations, with relatively few express constraints on where individuals may be sent.⁷ Over time, however, these powers have been implicitly conditioned by the HRA 1998 and by the UK’s international obligations.⁸

The Nationality, Immigration and Asylum Act 2002 (NIAA) introduced the concept of “safe third countries” into primary legislation, allowing the Secretary of State to certify certain states as appropriate destinations for removal on the basis that they would consider asylum claims and respect non-refoulement.⁹ The system assumed that “safety” would be assessed through a combination of legal commitments and practice, but it did not specify the evidential standard or institutional mechanisms for such assessment.

A provision-by-provision reading reveals how “safety” operates as an implied condition at multiple points in the statutory framework. Under NIAA 2002 ss 80A-80C (as amended by the Nationality and Borders Act 2022 ss 15-17), the Secretary of State may declare an asylum claim inadmissible where the claimant has a connection to a “safe third country”, defined as one from which the person will not be removed in breach of the Refugee Convention or Article 3 ECHR, and in which the person’s claim will be determined in accordance with that Convention. Safety is therefore not extrinsic to these provisions; it is built into the definitional preconditions for their lawful exercise. Schedule 3 to NIAA 2002 further conditions removal on the third country’s compliance with the prohibition of refoulement and its provision of access to a fair procedure. The Nationality and Borders Act 2022 ss 15-17 updated this framework by introducing the concept of inadmissibility based on a “connection” to a third country, but retained safety as the operative threshold: the certificate of inadmissibility can only lawfully be issued where the receiving state meets the Convention-compliance requirements. It follows that the “reconfiguration of removal powers” effected by R (AAA) is not a judicial imposition on an otherwise unconstrained power: it is a recognition that the statutory text itself conditions removal on demonstrable safety in the receiving state, a condition the Court found was not satisfied on the evidence.

Judicial review in this context operated through two interlocking channels: 1) HRA-based challenges alleging that removal would breach Convention rights, particularly Article 3 ECHR; and 2) public-law challenges to the reasonableness or rationality of the Secretary of State’s assessment of safety.¹⁰ In practice, the former increasingly dominated, as Strasbourg jurisprudence on non-refoulement made clear that states could not avoid responsibility by relying on third states whose systems lacked effective protection.¹¹

B. The Rwanda agreement and its statutory “hooks”

The 2022 Memorandum of Understanding between the UK and Rwanda did not constitute a treaty capable of creating directly enforceable rights in domestic law.¹² Its legal operation rested entirely upon existing statutory removal powers, in particular those governing inadmissible claims and transfers to designated safe third countries. The Home Office maintained that the combination of these broadly framed statutory provisions and the inadmissibility rules introduced in 2021 provided a sufficient legal basis for removal to Rwanda, subject to the limited constraints imposed by the HRA in individual cases ([7]).¹³

Its procedural structure proved constitutionally significant. The Rwanda scheme initially relied on executive action under existing statutory removal powers rather than bespoke primary legislation explicitly designating Rwanda as safe. Consequently, in R (AAA), the Supreme Court was not confronted with a direct parliamentary command to treat Rwanda as safe, a scenario that would have raised complex questions of judicial deference. Instead, the Court’s task was to determine whether the existing removal powers, construed in light of the HRA and the United Kingdom’s international obligations, could lawfully authorise transfers in circumstances where substantial evidence indicated systemic risk in the receiving state. The Court held that they could not, effectively narrowing the scope of those powers by reading them consistently with non-refoulement principles. Crucially, this interpretive strategy allowed the Court to prevent the Rwanda transfers without issuing a declaration of incompatibility under section 4 HRA,¹⁴ thereby avoiding a direct constitutional confrontation with Parliament while constraining executive action through human-rights-informed statutory interpretation.

The subsequent enactment of the Illegal Migration Act 2023 and the Rwanda Act 2024 altered the legal framework by placing the assessment that Rwanda is a safe receiving state directly into primary legislation and by restricting the scope for administrative and judicial reconsideration of that assessment.¹⁵ Although those statutes post-date R (AAA), they are nonetheless relevant in illustrating the constitutional implications of the Court’s reasoning. The judgment had established that reliance on broadly framed executive removal powers left the scheme vulnerable to being halted through rights-consistent statutory interpretation grounded in the principle of non-refoulement. If removals were to proceed notwithstanding that conclusion, Parliament was therefore required to intervene legislatively and to do so in explicit terms. The later legislation thus reflects the shift in institutional responsibility brought about by the Court’s approach, and the increased constitutional salience of the policy choices involved.

A. Non-refoulement in international and regional law

Article 33(1) of the 1951 Refugee Convention prohibits the expulsion or return (“refouler”) of a refugee “in any manner whatsoever” to territories where their life or freedom would be threatened on specified grounds.¹⁶ Non-refoulement has been interpreted by UNHCR and much scholarship as having attained the status of customary international law, and as imposing obligations that are not contingent upon formal recognition of refugee status ([4]; [10]).

Parallel obligations arise under Article 3 ECHR, which prohibits torture and inhuman or degrading treatment, and under Article 3 of the Convention Against Torture and Article 7 ICCPR.¹⁷ The European Court of Human Rights has consistently held that removal in breach of these provisions violates the sending state’s obligations, irrespective of whether the ill-treatment is inflicted by the receiving state or through onward refoulement.¹⁸

In the context of third-country transfers, the key doctrinal development has been the shift from an individualised to a systemic lens. Cases such as MSS v Belgium and Greece and Hirsi Jamaa v Italy establish that states cannot rely on presumptions of safety where there is evidence of serious deficiencies in the receiving state’s asylum system.¹⁹ The sending state must assess whether the receiving state’s institutional framework provides real and effective protection against refoulement. A terminological caution is warranted here. In MSS, the Grand Chamber found Belgium in violation of Article 3 ECHR for returning the applicant to Greece, where he was exposed to conditions of destitution and a deficient asylum procedure that created a real risk of onward refoulement: the decisive legal finding was the violation of Article 3, grounded in evidence of Greece’s systemic failures. In Hirsi Jamaa, the Court found Italy in violation of Articles 3 and 4 of Protocol 4 for intercepting migrants at sea and returning them to Libya, holding that Italy could not relinquish Convention responsibility by transfer to a state whose treatment of migrants was inadequate. The word “systemic” appears infrequently in the Hirsi Jamaa judgment itself; the correct formulation is that both cases hold states responsible for transfer where the receiving state’s procedures and conditions create a real risk of treatment contrary to Article 3 of the ECHR. The expression “systemic deficiencies” is used in this article as a shorthand for that finding, while acknowledging that the primary legal ground in both cases remains Article 3. The concept of systemic or structural deficiencies as a legal trigger subsequently migrated into EU asylum law, where the Court of Justice of the European Union developed it further in the Dublin context. In NS v Secretary of State for the Home Department (Joined Cases C-411/10 and C-493/10), the CJEU held that an EU Member State cannot transfer an asylum seeker to another Member State under the Dublin II Regulation where it cannot be unaware that “systemic deficiencies” in the receiving state’s asylum procedure and reception conditions mean there are substantial grounds for believing the person faces a real risk of inhuman or degrading treatment. The CJEU’s formulation thus explicitly adopted “systemic deficiencies” as a legal term of art, deployed to rebut the Dublin presumption of safety between Member States. The parallel with R (AAA) is instructive: in both contexts, a general legislative or diplomatic presumption of safety yields to evidence of institutional failure in the receiving state. While the UK has now left the EU legal framework, the convergence of ECHR and CJEU doctrine in this respect confirms that the systemic-deficiency concept reflects a broad principle of international human rights law, not merely a domestic or regional peculiarity.

B. Domestic incorporation and the role of the HRA

In the UK, these international obligations are domestically operationalised through the HRA 1998 and principles of statutory interpretation. Section 6 HRA renders it unlawful for public authorities to act incompatibly with Convention rights, while section 3 requires primary and subordinate legislation to be read, so far as possible, compatibly with those rights.²⁰ As a result, removal powers are not free-standing. Their lawfulness depends on the compatibility of their exercise with Article 3 ECHR, and—through the presumption of conformity—with non-refoulement more broadly.²¹ The courts have repeatedly emphasised that broad statutory language will not be read as authorising violations of fundamental rights without clear words, and even then, such authorisation may have constitutional limits.²²

R (AAA) sits squarely in this tradition. The Court did not treat non-refoulement as an optional interpretive consideration; it treated it as a structural constraint on the scope of the power to remove. Where systemic risk exists, the power is—in practical terms—inapplicable.

It is important to distinguish two related but doctrinally distinct operations that the Court performed. The first is systemic-deficiency analysis: the inquiry into whether the institutional framework of the receiving state is so inadequate that transfers to it would create a real risk of refoulement at large, irrespective of any individual’s particular circumstances. The second is individualised Article 3 risk assessment: the fact-specific inquiry, required in every removal decision, into whether a particular person faces a real risk of treatment contrary to Article 3 in the receiving state. On the reading of R (AAA) advanced in this article, systemic risk is legally decisive on its own: a demonstrated systemic deficiency is sufficient to render the removal power inoperative without requiring proof of harm to a specific individual. This is because, where the receiving state’s system cannot be trusted to apply non-refoulement reliably, no individual assessment can adequately compensate for that structural failure—the risk of onward removal materialises not from what the decision-maker in the receiving state knows about a particular applicant, but from what the system will do in practice. That said, the two forms of analysis are not mutually exclusive: even where systemic risk is found to be insufficient to block transfers categorically, the individualised obligation under Article 3 persists and must be discharged in each case. The Court’s approach in R (AAA) operates primarily at the systemic level, but the continuing necessity of individual assessment ensures that the two strands of the doctrine are complementary rather than conflicting.

To move beyond synthesis of the case law, this article proposes a framework for understanding the structural tensions inherent in third-country transfer schemes. These tensions are not peculiar to Rwanda; they arise from the very logic of externalisation.

First, there is temporal displacement. Compliance with non-refoulement is often asserted prospectively, via promises of future capacity-building, training, or monitoring.²³ The sending state effectively bets on future institutional improvement to justify present removal, but non-refoulement is concerned with current risk at the point of removal, not speculative future conditions.

Secondly, there is evidentiary asymmetry. The state has privileged access to diplomatic communications, monitoring reports, and internal assessments; asylum seekers typically do not. Courts are asked to review safety assessments on the basis of an uneven evidentiary record, often heavily shaped by the executive ([8]).²⁴ This asymmetry is particularly acute when the receiving state has incentives to present an optimistic picture of its own compliance.

Thirdly, externalisation substitutes diplomatic assurances for legally enforceable guarantees. Individuals transferred to a third country usually lack direct access to the courts of the sending state once removed, and remedies in the receiving state may be weak or ineffective ([2]). The protection of rights therefore depends on the continued goodwill and capacity of the receiving state and the sending state’s willingness to enforce compliance diplomatically-mechanisms that sit uneasily with the individualised remedial structure of human rights law.

The Rwanda partnership exemplified all three tensions: heavy reliance on future reform, severe evidentiary disputes about Rwanda’s actual record, and a remedial structure built on diplomatic assurance rather than enforceable domestic rights. It is against this backdrop that the Supreme Court’s approach in R (AAA) must be assessed. To make this framework operational, consider how each tension generates a justiciable question.

The temporal displacement tension translates into the question: at the date of proposed removal, did the receiving state’s asylum system in fact provide adequate protection against refoulement, or was compliance with non-refoulement merely an aspiration projected into the future? In R (AAA), the Court decided that question adversely to the Government by examining the current state of Rwanda’s procedures at the time of the judgment ([94]–[108]), rather than accepting that future capacity-building under the MoU would remedy the present deficiency. The court’s refusal to credit prospective reform as satisfying a present-tense safety condition is the doctrinal resolution of the temporal displacement tension.

The evidentiary asymmetry tension raises the question: how should a court weigh safety evidence produced substantially by or at the instigation of the executive, particularly when the asylum seeker has no access to the underlying diplomatic communications? In R (AAA), the Court resolved this by placing decisive weight on independent evidence from UNHCR and expert witnesses, and by applying an intensive standard of review rather than a deferential one ([99]-[108]). The tension thus resolves into the procedural and substantive question of the intensity of judicial scrutiny and the sources of evidence that can lawfully be treated as sufficient to establish safety.

The remedial connectivity tension, finally, becomes justiciable as the question: does the removed individual retain effective access to a legal remedy capable of correcting a breach of the assurances given by the receiving state, and if not, does the absence of such access itself render the removal unlawful? In R (AAA), this question was resolved implicitly: the absence of any legally enforceable right for the transferred individual to invoke the assurances, combined with the weakness of Rwanda’s appellate structure, reinforced the finding that the risk of refoulement could not be adequately mitigated. The preconditions proposed in Section VIII below are designed, in part, to address precisely this gap.

A. The judgment’s core holding

The Supreme Court’s conclusion in R (AAA) can be stated relatively simply: on the evidence before it, there were substantial grounds for believing that asylum seekers removed to Rwanda faced a real risk of refoulement, and therefore removal would breach Article 3 ECHR.²⁵ That alone, however, does not convey the interpretive work the Court performed.

Unlike many Article 3 removal cases, R (AAA) did not concern an individual’s removal to their country of origin, but to a third state which itself had international obligations and domestic procedures.²⁶ Moreover, the statutory removal powers relied upon by the Secretary of State were general in character; Parliament had not explicitly legislated to authorise removal to Rwanda specifically.²⁷ The Court therefore had to decide how to relate these broad powers to the structural risk evidenced in Rwanda’s system.

The Court’s answer was to hold that the statutory powers, read compatibly with Article 3 and the UK’s non-refoulement obligations, simply could not lawfully be exercised in circumstances of systemic risk.²⁸ This is not merely a finding of irrationality. It is a conclusion about the legal limits of the power itself, derived from constitutional principle rather than from the text of the Immigration Acts alone.

B. Judicial method: safety as a structural condition

The Court’s method involved a significant choice: it treated “safety” as a condition precedent to lawful removal, not as a matter of executive judgment subject only to Wednesbury review.²⁹ This can be contrasted with a more deferential model in which the Secretary of State’s assessment of safety would be upheld unless no reasonable decision-maker could have so concluded.

In rejecting deference, the Court relied heavily on the absolute character of Article 3 and on the systemic nature of the risk.³⁰ It emphasised that where there is a real risk of refoulement arising from institutional deficiencies, assurances and monitoring cannot suffice unless and until those deficiencies are demonstrably remedied. Judicial review in such circumstances cannot be confined to process; it must extend to the substance of safety.

From a statutory interpretation perspective, this amounts to treating the absence of systemic risk as an implied legal precondition of the valid exercise of removal powers. The power remains on the statute book, but its lawful domain is narrowed: it is inoperative where institutional conditions in the receiving state fall below a certain threshold.³¹

The interpretive mechanism by which the Court reached this conclusion operates through a combination of three overlapping constitutional constraints. The primary mechanism is HRA 1998 s 6, which renders it unlawful for the Secretary of State, as a public authority, to act incompatibly with Convention rights: because removal to Rwanda would, on the evidence, expose individuals to a real risk of refoulement contrary to Article 3 ECHR, the exercise of the removal power in those circumstances is itself an unlawful act (R (AAA) [73]-[82]). This s 6 analysis is reinforced, but not duplicated, by s 3 HRA, which requires the removal powers in NIAA 2002 and the Nationality and Borders Act 2022 to be read, so far as possible, compatibly with Article 3 ECHR: where the statutory language is capable of a safety-conditioned reading—and, as argued above, the text of NIAA 2002 supports precisely such a reading—s 3 requires that interpretation to be preferred (R (AAA) [74]-[77]). Underpinning both HRA routes is the common-law principle of legality: the established constitutional presumption, applied in Simms and A (Belmarsh), that Parliament does not intend by general or ambiguous statutory language to confer power to violate fundamental rights. In R (AAA), that principle is engaged because the removal powers, broadly drafted, cannot reasonably be read as authorising removal to a state where non-refoulement cannot be assured ([99]-[108]). The three mechanisms are thus mutually reinforcing: s 6 HRA provides the immediate prohibition on unlawful executive action; s 3 HRA conditions the statutory interpretation of removal powers; and the principle of legality supplies the constitutional baseline against which both operate. Together, they explain why the Court was able to constrain executive action without issuing a declaration of incompatibility, and why the constitutional claim in this article rests on delegation limits rather than on formal statutory invalidity.

C. Domestic and international authority

The Court’s approach is consistent with, but goes beyond, earlier domestic and Strasbourg case law. In Soering and Chahal, the Strasbourg Court made clear that states cannot expose individuals to a real risk of ill-treatment through removal, even in the context of extradition for serious crimes.³² In MSS and Hirsi Jamaa, it held that systemic deficiencies in asylum systems and border procedures could render transfers unlawful even between states party to the same regional instruments.³³

Domestically, cases such as Simms and A (Belmarsh) established that broad statutory powers will be construed so as not to authorise serious rights violations absent clear language, and that even clear language may be subject to constitutional constraints.³⁴R (AAA) extends this logic into the domain of third-country transfers: it suggests that Parliament would have to legislate with exceptional clarity—and potentially in defiance of international obligations—if it wished to permit removals in the face of systemic risk.

A. Reconstructing the Government’s arguments

A serious doctrinal assessment must begin by reconstructing, rather than caricaturing, the Government’s legal case. At its core, that case rested on three propositions. First, that non-refoulement obligations can be discharged by transfer to a third state that is, in law and in principle, willing to comply with them.³⁵ Secondly, that diplomatic assurances, combined with monitoring and capacity-building, can mitigate systemic risk to an acceptable level.³⁶ Thirdly, that the breadth of the statutory removal powers, read together with the MoU and later the Rwanda Treaty, left a substantial margin for executive judgment, into which courts should not intrude save in cases of manifest unreasonableness.

These propositions were not legally trivial. State practice has long relied on assurances in extradition and expulsion, and domestic courts have sometimes accepted them where they are specific, credible, and accompanied by effective monitoring.³⁷ Moreover, the international law of state responsibility acknowledges that states may cooperate in fulfilling obligations, and does not per se prohibit burden-sharing arrangements ([1]).³⁸ The Government’s position thus reflected a plausible—if ambitious—reading of the relationship between non-refoulement and cooperative migration management.

The leading case on assurance reliability is Othman (Abu Qatada) v United Kingdom, in which the Strasbourg Court synthesised from earlier case law a non-exhaustive checklist of factors material to assessing whether diplomatic assurances provide a sufficient guarantee against ill-treatment. Applying those factors to the Rwanda scheme reveals systematically why the Government’s reliance on the MoU could not satisfy the legal threshold. First, Othman requires that the content of the assurances be practical and not merely theoretical, covering the specific risk at issue. The Rwanda MoU offered broad undertakings about non-refoulement in principle; it did not provide specific procedural guarantees addressing the documented deficiencies in decision-making that the Court found to create systemic risk. Second, the Othman checklist asks whether the assurances are given by the central government and are binding on it. The MoU was expressly stated to be non-binding in international law, a significant deficiency on this criterion. Third, Othman looks to the quality and independence of monitoring arrangements. The Joint Committee envisaged under the MoU was not yet operational at the time the Supreme Court considered the matter, and the Court found no adequate substitute mechanism capable of providing independent verification. Fourth, the Court in Othman weighs the general human-rights situation in the receiving state and whether there is a consistent practice of ill-treatment. The Supreme Court’s findings in R (AAA) on Rwanda’s past refoulement of asylum seekers, including to the Democratic Republic of Congo, amounted to precisely the “consistent practice” that the Othman framework identifies as weighing heavily against reliance on assurances. Mapping the Rwanda scheme evidence against the Othman checklist therefore reveals that the Government’s assurance-based case failed not merely on a broad evaluative judgement, but on specific legal and institutional criteria established by the very case law on which the Government itself relied. The defects were structural, not incidental.

B. Why the arguments fail

Nevertheless, R (AAA) exposes the fundamental limits of this position. The Supreme Court’s decisive finding was that Rwanda’s past record of refoulement, coupled with enduring deficiencies in its asylum procedures and the absence of robust safeguards for independence and legal representation, created substantial grounds for believing asylum seekers would face a real risk of refoulement ([12]),³⁹ the core test under Article 3 ECHR. This judgment rests on a pragmatic analysis of assurances: they are not self-executing. Their efficacy depends irrevocably on the receiving state’s institutional capacity and political willingness to comply, and, crucially, on the existence of independent, effective mechanisms for monitoring violations and enforcing correction. In the absence of such machinery, assurances remain non-justiciable diplomatic promises, incapable of mitigating a systemic, evidence-based risk.

From a domestic constitutional perspective, the Government’s case presented a profound difficulty. It asked the Court to infer that Parliament, through broad statutory removal powers, had authorised the Executive to send individuals to a state where their protection from refoulement rested entirely on non-binding instruments and aspirational, future-oriented reforms. Applying the principle of legality, the Court held that Parliament could not lightly be taken to have intended such a result absent express language. Given the absolute nature of Article 3 rights,⁴⁰ their displacement requires clear and specific statutory authority, not implication drawn from general ministerial discretion.

A critical reading, however, might question whether the Court’s reasoning elided the distinction between policy risk and legal incapacity. International law does not per se prohibit reliance on assurances; indeed, state practice in areas like extradition and deportation frequently operates through prospective undertakings backed by monitoring. The pivotal move in R (AAA) was therefore not the creation of a new legal rule, but the Court’s substantive evaluation, informed by extensive factual evidence, that Rwanda’s institutions, at the material time, lacked the requisite stability, independence, and adherence to rule-of-law norms to make those undertakings reliable. This entailed the judiciary making a predictive, evidence-heavy assessment of a foreign state’s capabilities, a function touching upon the executive’s traditional prerogative in diplomacy and foreign relations.

Consequently, the Government’s case failed not because extraterritorial cooperative schemes are inherently unlawful, but because the specific amalgamation of a substantiated refoulement risk, the absence of verifiable and enforceable safeguards, and broadly worded statutory language could not bear the weight of authorisation placed upon it. R (AAA) thus converts what might be viewed as contingent policy or diplomatic weaknesses into incontrovertible legal defects, a conversion that underscores the demanding domestic constitutional and evidentiary thresholds that now govern such arrangements, whatever their strategic or political appeal.

The Safety of Rwanda (Asylum and Immigration) Act 2024 can be read as Parliament’s attempt to reassert control over the safety determination by declaring Rwanda to be safe as a matter of law and constraining the ability of courts and tribunals to revisit that conclusion ([3]). Its significance can be understood through three interconnected lenses: the Act’s mechanism of legislative assertion, its impact on the courts’ supervisory role, and its interface with international obligations.

a. The Act’s Mechanism: Legislating Against Evidence

The Act’s core mechanism is to convert a complex, predictive assessment of risk, the very subject of the Supreme Court’s evidential scrutiny in R (AAA), into an irrebuttable legislative fact. Section 1 gives effect to “the judgement of Parliament that the Republic of Rwanda is a safe country”,⁴¹ and Section 2 conclusively directs all decision-makers, including courts and tribunals, to treat Rwanda as generally safe, effectively converting the systemic assessment into an irrebuttable legislative presumption applicable across all cases.⁴² By modifying the operation of the HRA and seeking to limit the effect of interim measures from Strasbourg (Sections 3-5), the Act constructs a bespoke domestic legal reality designed to insulate the removal policy from the type of judicial review that defeated it previously.⁴³

b. Constitutional Tensions and the Role of the Courts

Within the UK’s dualist system, Parliament possesses the sovereign authority to make such a declaration. The profound constitutional questions lie in its consequences:

i. The Judicial Role: The Act creates a potent tension with the courts’ inherent supervisory jurisdiction. While it ostensibly ousts their ability to re-examine the general safety of Rwanda, orthodox constitutional principle, holds that Parliament cannot, by general or ambiguous words, eliminate the courts’ fundamental supervisory jurisdiction over exercises of executive power that risk violating absolute rights.⁴⁴ The courts retain a residual jurisdiction grounded in the principle of legality. If the operation of the Act were shown to compel the Executive to act in a way that creates a real and imminent risk of a breach of an absolute right like Article 3 ECHR (non-refoulement) in an individual case, the courts would, in accordance with established principle, be required to assert their supervisory jurisdiction: under R (Privacy International), ouster clauses and limiting provisions are construed so as not to remove the courts’ inherent jurisdiction to review executive action that violates fundamental constitutional rights, absent clear and specific statutory language to that effect. Applied to the Rwanda Act, the conditional framework operates as follows: if (a) an individualised Article 3 risk is credibly pleaded on the basis of specific evidence not considered by the Act’s blanket safety declaration, and (b) no alternative route of challenge remains effective under the Act’s modified framework, then (c) the court, applying the principle of legality confirmed in Simms and the ouster-clause principles in Privacy International, would be entitled to grant relief to prevent removal. The Act narrows the avenues of challenge but does not, and under existing doctrine cannot by general words, extinguish this core constitutional safeguard.

ii. The International Law Interface: Within a dualist framework, the Act is formally valid domestic law. However, it places the UK in the position of asserting a domestic legal fiction directly at odds with the evidentiary requirements of its treaty commitments. The prohibition of refoulement under the Refugee Convention and Article 3 ECHR requires a fact-sensitive, individualised assessment of risk. The Act’s conclusive presumption seeks to bypass this requirement by parliamentary declaration. The Act does not breach international law directly, it creates conditions under which the executive would find compliance with those international duties exceptionally difficult, if not impossible, to verify, raising questions not of domestic validity but of international responsibility and practical enforceability.⁴⁵

A more sceptical reading would emphasise that the Act remains formally valid domestic law, and that parliamentary sovereignty doctrine offers no strict legal barrier to legislating factual presumptions, even conclusive ones. The constitutional tension lies not in the Act’s formal validity but in whether courts, when faced with its application in individual cases, can reconcile their duty to apply statute with their constitutional role as guarantors of fundamental rights. That reconciliation would have occurred, if at all, through strained interpretation or exceptional invocation of common law constitutionalism, not through any pre-existing legal prohibition on Parliament’s course of action.

c. Context and Significance: A Dialogue of Institutions:The Supreme Court in AAA applied orthodox public law principles, identifying a systemic risk that unenforceable assurances could not remedy. Parliament, prompted by the Executive, responded with the Rwanda Act, attempting to redefine the domestic legal baseline and substitute statutory judgement for the Court’s evidence-based findings. This legislative intervention highlights the tension inherent in the constitutional dialogue: while Parliament may legislate to create domestic legal facts, it cannot, by mere declaration, manufacture the institutional reliability, international cooperation, or political legitimacy necessary for a sustainable policy. The subsequent repeal of the Act confirms its fragility: the friction between statutory assertion and the practical, legal, and diplomatic realities proved insurmountable, underscoring the limits of legislative power when absolute rights and evidential constraints collide.

Although a full analysis of the Rwanda Act lies beyond the scope of this article, the Rwanda Act represents a landmark case study in the limits of parliamentary sovereignty when confronted with absolute rights. It confirms that Parliament can legislate to redefine domestic legal facts and constrain judicial review pathways. However, it also affirms that there remains a constitutional and practical hinterland where such legislation cannot safely operate: where it mandates state action that risks violating non-derogable rights, and where it requires the courts to suspend their fundamental protective function. The Act did not overturn R (AAA); it merely highlighted, in the boldest terms, the immense cost and profound difficulty of trying to legislate away a risk grounded in factual reality. In doing so, it underscored that sustainable compliance with the prohibition of refoulement cannot be achieved by statutory command alone but must be built on verifiable safeguards and accountable processes, the very prerequisites the judiciary exists to uphold.

If third-country transfer schemes are to remain legally conceivable after R (AAA), what conditions must they satisfy? On the analysis developed above, at least three sets of requirements emerge.

First, statutory clarity and constraint. Parliament must articulate with precision the circumstances in which removal to a particular state is permitted, rather than relying on open-textured delegations. The statute should explicitly condition removal on demonstrable compliance with non-refoulement and provide mechanisms for suspending removals when systemic risk emerges.⁴⁶

Secondly, institutional safeguards in the receiving state. Asylum decision-making bodies must be independent, adequately resourced, and legally required to apply non-refoulement consistently. There must be access to legal representation, interpreters, and meaningful appellate review ([11]).

Thirdly, remedial connectivity. Individuals removed under cooperative schemes should retain some form of effective access to the courts of the sending state, at least for the purpose of challenging systemic failures or breaches of assurances. This might require statutory recognition of extraterritorial jurisdiction for certain claims or the creation of a treaty-based supervisory mechanism with binding effect ([9]).⁴⁷

Fourthly, and cutting across all three conditions, post-return monitoring must be genuinely independent, systematically conducted, and capable of triggering suspension of the transfer scheme. The experience of the Rwanda MEDP illustrates the inadequacy of monitoring arrangements that are prospective, nascent, or dependent on the goodwill of the receiving state. For post-return monitoring to discharge the legal function required by Othman and confirmed by R (AAA), it must meet several requirements. The monitoring body must be institutionally independent from both the sending and receiving state governments. It must have unimpeded access to transferred individuals and to asylum decision-making records. Its findings must be communicated promptly to decision-makers in the sending state, and they must carry legal consequences: a finding of systemic failure should automatically trigger a duty to suspend further transfers and, where individuals already removed are at risk, to take steps to enforce the assurances or to seek repatriation. Finally, the monitoring framework must be capable of distinguishing between individual incidents and systemic failures, as the legal consequences of the latter are more far-reaching. Without these features, monitoring provides a veneer of accountability while leaving the structural deficiencies that give rise to refoulement risk unaddressed. Designing effective independent monitoring is therefore not merely a policy preference but a legal prerequisite for any future externalisation scheme that aspires to withstand judicial scrutiny.

Absent such features, externalisation schemes will continue to collide with the structural demands of non-refoulement as understood in R (AAA), and courts are likely to construe statutory powers narrowly to avoid sanctioning arrangements that rely on aspirational assurances rather than enforceable legal safeguards.

R (AAA) is often described as the case that “struck down” the Rwanda policy. In doctrinal terms, however, it did something more subtle and more consequential. It re-configured the relationship between statutory removal powers, international protection norms, and constitutional principle, by treating non-refoulement as a structural condition on the lawful exercise of those powers.

This article has argued that R (AAA) should be read not merely as an endorsement of UNHCR’s assessment of Rwanda, but as a significant development in the domestic law of statutory interpretation and constitutional limits. By resisting a deferential approach to executive assessments of safety and by insisting that systemic risk renders statutory powers effectively inoperative, the Supreme Court has reshaped the legal terrain upon which future externalisation schemes will be built.

The subsequent enactment and repeal of the Rwanda Act confirm that Parliament cannot easily legislate its way out of this terrain without openly confronting both international obligations and domestic constitutional norms. Whether the UK continues to pursue externalisation in other guises or instead redirects its legislative energies towards safer and more legally sustainable avenues for migration management, will be a central question for statute law—and for Statute Law Review—in the years ahead.

Two wider dimensions of the case deserve acknowledgement as areas for further inquiry. First, R (AAA) raises acute questions about the institutional competence of domestic courts to assess the capacity and reliability of foreign asylum systems. The judiciary’s task in such cases extends well beyond doctrinal interpretation: it requires evaluation of socio-political conditions, administrative cultures, and rule-of-law standards in foreign states on the basis of expert evidence and diplomatic materials. This is a function courts are formally equipped to perform through the standard adversarial process, but it sits uneasily with traditional doctrines of executive deference in foreign-affairs matters, and it risks courts substituting their evaluation of foreign institutional quality for the considered assessments of governments with direct diplomatic engagement. A fuller examination of whether specialist judicial institutions or standing expert panels might assist courts in this function falls beyond the scope of the present article but represents an important institutional question prompted by R (AAA). Second, the UK’s Rwanda scheme is not unique: comparable externalisation arrangements have been attempted or proposed by other states, including Australia’s offshore processing regime to Nauru and Papua New Guinea and the EU-Türkiye Statement of 2016, which similarly relied on a designated-safe-country framework. A comparative analysis of how courts in those systems have applied analogous doctrines would enrich understanding of the structural tensions identified in this article and reveal whether the UK Supreme Court’s approach in R (AAA) represents an emerging international consensus or a distinctively domestic constitutional response. Those comparative questions are reserved for future work.

¹HC Deb 14 April 2022, vol 712, cols 515-18 (Priti Patel).

²See e.g. Nationality, Immigration and Asylum Act 2002; Immigration Act 2014; Immigration Act 2016; Human Rights Act 1998.

³Illegal Migration Act 2023; Safety of Rwanda (Asylum and Immigration) Act 2024, c 8 (Rwanda Act). On the Rwanda Act’s repeal, see Border Security, Asylum and Immigration Act 2025, s 37 (repealing the Rwanda Act).

⁴R (AAA) v Secretary of State for the Home Department [2023] UKSC 42.

⁵ibid [113]-[121].

⁶On doctrinal public-law method, see M Elliott and R Thomas, Public Law (4th edn, OUP 2023) ch. 1.

⁷Immigration Act 1971, ss 3-5; see also Immigration and Asylum Act 1999, pt VI.

⁸HRA 1998, ss 3, 6; R v Lyons [2002] UKHL 44, [27] [28] (Lord Hoffmann).

⁹NIAA 2002, ss 80A-80C and Sch 3 (as amended). See generally M den Heijer, Europe and Extraterritorial Asylum (Hart 2012) 90-111.

¹⁰For HRA-based challenges, see e.g. EM (Eritrea) v Secretary of State for the Home Department [2014] UKSC 12, [2014] AC 1321; public-law review illustrated by R (Yogathas) v Secretary of State for the Home Department [2002] UKHL 36, [2003] 1 AC 920.

¹¹MSS v Belgium and Greece (2011) 53 EHRR 2; Hirsi Jamaa v Italy (2012) 55 EHRR 21.

¹²Rwanda MoU (n 1) para 1.6 (expressly non-binding in international law).

¹³See Nationality and Borders Act 2022, ss 15-17 (inadmissibility)

¹⁴R (AAA) (n 5) [74]-[82]. The Court did not issue a declaration of incompatibility under HRA 1998, s 4.

¹⁵IMA 2023, especially ss 2-4; Rwanda Act, ss 1-5. On the Rwanda Act’s purpose, see Govt of the UK, ‘Safety of Rwanda (Asylum and Immigration) Bill: Factsheet’ (Home Office, 11 January 2024). https://www.gov.uk/government/publications/the-safety-of-rwanda-asylum-and-immigration-bill-factsheets/safety-of-rwanda-asylum-and-immigration-bill-factsheet-accessible (accessed 27/5/2026).

¹⁶Convention Relating to the Status of Refugees (adopted 28 July 1951, entered into force 22 April 1954) 189 UNTS 137 (Refugee Convention), art 33(1).

¹⁷ECHR, art 3; Convention against Torture and Other Cruel, Inhuman or Degrading Treatment or Punishment (adopted 10 December 1984, entered into force 26 June 1987) 1465 UNTS 85, art 3; International Covenant on Civil and Political Rights (adopted 16 December 1966, entered into force 23 March 1976) 999 UNTS 171, art 7.

¹⁸Soering v United Kingdom (1989) 11 EHRR 439; Chahal v United Kingdom (1996) 23 EHRR 413.

¹⁹MSS (fn. 12); Hirsi Jamaa (fn. 12).

²⁰HRA 1998, ss 3, 6.

²¹On the presumption of conformity with international law, see Assange v Swedish Prosecution Authority [2012] UKSC 22, [2012] 2 AC 471 [122] (Lord Dyson); R v Lyons (n 9).

²²R v Secretary of State for the Home Department, ex p Simms [2000] 2 AC 115, 131 (Lord Hoffmann); A v Secretary of State for the Home Department [2004] UKHL 56, [2005] 2 AC 68.

²³Rwanda MoU (fn. 1) paras 10-16 (capacity-building, training, and funding).

²⁴On evidentiary asymmetry in immigration.

²⁵R (AAA) (n 5) [119]-[121].

²⁶ibid [54]-[60].

²⁷ibid [74]-[77].

²⁸ibid [118]-[121].

²⁹For the contrast, see Lord Hoffmann’s approach in Rehman v Secretary of State for the Home Department [2001] UKHL 47, [2003] 1 AC 153.

³⁰R (AAA) (fn. 5) [99]-[108].

³¹For discussion of powers rendered inoperative by constitutional constraints, see J Elliott, The Constitutional Foundations of Judicial Review (Hart 2001) 90-94.

³²Soering (fn. 20); Chahal (fn. 20).

³³MSS (fn. 12); Hirsi Jamaa (fn. 12).

³⁴Simms (fn. 24); A (Belmarsh) (fn. 24).

³⁵See the Secretary of State’s arguments summarised in R (AAA) (n 5) [60]-[65].

³⁶Rwanda MoU (n 1) paras 15-22; Second Reading debate, Rwanda Act, HL Deb 29 January 2024, vol 835, cols 100-105.

³⁷Othman (Abu Qatada) v United Kingdom (2012) 55 EHRR 1; Babar Ahmad v United Kingdom (2013) 56 EHRR 1.

³⁸For a nuanced view.

³⁹R (AAA) (n 5) [94]-[108].

⁴⁰Compare Simms (n 24) 131-32.

⁴¹ibid, s 1(2).

⁴²ibid, ss 2-5.

⁴³ibid, s 3(3)-(5).

⁴⁴R (Privacy International) v Investigatory Powers Tribunal [2019] UKSC 22, [131]-[132].

⁴⁵Goodwin-Gill (n 44) 638-639.

⁴⁶For similar proposals in EU law, see S Peers, EU Justice and Home Affairs Law (4th edn, OUP 2016) ch 14.

⁴⁷On extraterritorial human-rights obligations and jurisdiction.