The Good and Faithful Servants

One of the foundational assertions of judicial approaches to legislation in the UK comes from Willes J in Lee v Bude and Torrington Railway Co:

We sit here as servants of the Queen and the legislature. Are we to act as regents over what is done by Parliament with the consent of the Queen, lords and commons? I deny that any such authority exists.

The idea of the courts as servants to other constitutional actors has fallen decidedly out of fashion as judicial independence has become an increasingly valued feature of the UK’s constitutional order. But it still inhabits some of the darker corners of jurisprudence. Lord Denning’s account of the courts “loyally” following Parliament’s precepts where security is at stake in Hosenball comes to mind. The notion of the judiciary as servants seems to speak to constitutional quiescence, rejecting the imposition of common law constraints upon the terms of statutes unless they are ambiguous.

This vision of service accords with a discourse over judicial activism and conservatism which presents these as opposite poles. Activism, the debate in US legal scholarship once went, involved the invention of constitutional constraints upon the legislature and the executive by progressive judges. In recent decades, however, there has been an increasing recognition of conservative activism, where “politicians in robes” use result-oriented jurisprudence to undermine certain rights protections.

The UK Supreme Court’s decision in Dillon demonstrates an intense form of regressive judicial activism at work in UK constitutional law. Arising, as it does, at the intersection of issues around the Windsor Framework and Northern Ireland’s conflict legacy legislation, which do not attract the attention of many UK public lawyers, this decision is unlikely to receive the scrutiny it warrants. But here the periphery is projecting an urgent warning to the core of constitutional law discourse. The UK Government had made rights commitments to the people of Northern Ireland under the UK-EU Withdrawal Agreement that ministers thereafter found to be getting in the way of their policy priorities, and the Court has worked assiduously to hollow out those commitments.

The Windsor Framework Commitments and the Legacy Act 2023

The Dillon case concerns the controversial Northern Ireland Troubles (Legacy and Reconciliation) Act 2023 (the “Legacy Act”), the Conservative Government’s instrument for dealing with the challenges posed by ongoing inquests and litigation relating to the Northern Ireland conflict. This legislation, in curtailing the operation of civil and criminal justice processes in conflict‑related cases in favour of an alternative mechanism for addressing legacy issues, the Independent Commission for Reconciliation and Information Recovery (ICRIR), has generated profound rights concerns.

The Conservative Government was unable to issue a parliamentary statement that this legislation was in conformity with the UK’s European Convention obligations. It has faced multiple challenges, including under the Human Rights Act 1998 and through an inter‑state application brought by Ireland before the European Court of Human Rights. Given how expansive this litigation is, this post focuses primarily on the challenges grounded in Article 2, paragraph 1 of the Windsor Framework:

The United Kingdom shall ensure that no diminution of rights, safeguards or equality of opportunity, as set out in that part of the 1998 Agreement entitled Rights, Safeguards and Equality of Opportunity results from its withdrawal from the Union, including in the area of protection against discrimination, as enshrined in the provisions of Union law listed in Annex 1 to this Protocol, and shall implement this paragraph through dedicated mechanisms.

This was an essential element of the Brexit deal which the UK Government conceded early in the Brexit negotiations. If it had not, it would not have been able to maintain that Brexit conformed with the ‘letter and spirit’ of the Belfast/Good Friday Agreement 1998, because that document’s Rights, Safeguards and Equality of Opportunity (RSEO) Chapter established a new framework for post‑conflict governance in Northern Ireland grounded in the rights of ‘everyone in the community’, with many of those protections at the time (or developing thereafter) being derived from EU law. On its face, the non-diminution commitment in Article 2 commits the UK Government to maintaining those rights and equality protections derived from EU law in force in Northern Ireland law at the point of Brexit as if Northern Ireland continued to be part of an EU member state.

This aspect of the litigation is of particular importance because the Withdrawal Agreement provides for the disapplication of primary legislation to the extent that it conflicts with relevant elements of EU law. These requirements are covered by Article 4 of the Withdrawal Agreement, which stipulates that provisions of the Agreement, and the EU law continuing to apply within the UK by virtue of it, are to be afforded the same remedies as would apply to breaches of EU law by Member States. That obligation was reflected in domestic law by section 7A of the European Union (Withdrawal) Act 2018, as confirmed by the UK Supreme Court in Allister.

Both the Northern Ireland High Court and a unanimous panel in the Northern Ireland Court of Appeal used the Article 2 Windsor Framework mechanism to disapply the immunity provisions of the Legacy Act, insofar as they conflicted with the Victims’ Directive’s requirement that individuals be able to challenge decisions by the authorities to refuse to prosecute. This posed a serious constraint on law and policy making on a UK-wide basis; the UK had committed to a higher rights baseline for Northern Ireland, and would have to meet those requirements in any legislation of general application across the UK, or see it disapplied with regard to Northern Ireland law. In its judgment the UK Supreme Court undertakes a series of highly inventive steps which, taken together, eviscerate these obligations.

Direct Effect

The case turns upon the application of the rules for direct effect of EU law. The Court acknowledges that Parliament did set out that Article 2 of the Windsor Framework should have direct effect, or else various terms in the Withdrawal Agreement legislation would be subverted:

The new section 78C empowers the Human Rights Commission and the Equality Commission to bring judicial review proceedings in respect of an alleged breach or potential future breach of article 2(1) of the Windsor Framework or to intervene in such proceedings. The new section 78D empowers those Commissions to assist persons who have brought proceedings in respect of an alleged breach or potential future breach of article 2(1) of the Windsor Framework. The provisions in section 78C and 78D are evidence of the view of Parliament that article 2(1) is capable of giving rise to a directly enforceable right in certain circumstances.

The key to the judgment lies in the phrase ‘certain circumstances’, and indeed, in the way it injects such stringent circumstances into these arrangements that very few cases become feasible. To do so, the Court, at [112], imposes direct effect requirements on the RSEO Chapter of the Belfast/Good Friday Agreement of 1998, under the Demirel test:

The test for direct effect of an international agreement between the EU and non Member States was established by the Court of Justice of the European Union (“CJEU”) in Case 12/86 Demirel v Stadt Schwabisch Gmund [1987] ECR 3719 (“Demirel”) at para 14 … : “…a provision in an agreement concluded by the European Union with a non-member country must be regarded as being directly applicable when, regard being had to its wording and to the purpose and nature of the agreement, the provision contains a clear and precise obligation which is not subject, in its implementation or effects, to the adoption of any subsequent measure …”

Because, the Court proceeds at [113], Article 2 ‘relates to rights, safeguards or equality of opportunity “as set out in” the RSEO chapter’, it must consider whether ‘those provisions read together impose a clear and precise obligation which satisfies the test for direct effect’. Direct effect requirements that should have been applied with regard to specific EU law measures being invoked under the Windsor Framework are instead being applied to the 1998 Agreement. The 1998 Agreement is not an EU association agreement, as was at issue in Demirel. The application of Demirel is thus entirely result-oriented jurisprudence. The 1998 Agreement provides the context in which EU law continues to apply under the Windsor Framework. It is not a manifestation of EU law. But the Supreme Court used this test to rule that the aspects of paragraphs 1, 11 and 12 of the RSEO chapter that the respondents relied upon did not have direct effect and therefore their action could not succeed.

This position was backed up by a very narrow account of the Victims’ Directive, which, it was found at [135], ‘does not address and does not seek to regulate broader questions of policy as to when prosecutions should be pursued, such as a national policy on immunity for the purpose of achieving reconciliation following conflict’. But the Court acknowledged that this was a subordinate position; it had already concluded, at [133], that the relevant part of the RSEO chapter ‘cannot in itself be regarded as having direct effect’. In other words, there was a way to deal with this case without gutting Article 2, and the Supreme Court’s decision is deliberately maximalist.

The only element of EU law left where the Article 2 commitments clearly remain able to meet the direct effect requirements relate to the six EU directives relating to anti-discrimination law covered by Annex 1 of the Windsor Framework (at [118]):

Before us, it was common ground between the Secretary of State and the applicants that article 2(1) might operate in conjunction with a directly effective provision of one of the Annex 1 Directives so as to give rise to a directly effective obligation not to diminish such a right. We agree.

To reach this conclusion, however, the Court had to confront the fact that the express wording of Article 2 is expansive; the UK makes a commitment ‘including’, but not limited to, these provisions. It does so at [125]:

While the provisions of paragraphs 1, 11, and 12 of the RSEO chapter do not themselves have direct effect, article 2(1) of the Windsor Framework may be capable of having direct effect in conjunction with other EU instruments falling within the ambit of the RSEO chapter if the Demirel requirements are satisfied in respect of the obligation imposed.

This is an exercise in unicorn-making. The Court holds out the possibility of some element of EU law might remain available to actors in domestic courts, whilst (as addressed below) imposing such stringent conditions as to make this a functional impossibility.

Promises, What Promises?

The imposition of direct effect upon the RSEO Chapter in this way is not simply wrong at law. It flies in the face of the promises made by the UK Government with regard to the Windsor Framework. In an extensive Explainer document, published in 2020 to set out the UK Government’s account of the non-diminution commitments it had taken on under Article 2, there was an explicit recognition of both the ability to rely on the non-diminution commitment as a basis for litigation in the domestic courts and, at [13], the scope of this commitment:

It should be noted that the directives in Annex 1, which have a future-facing dimension, represent only a subset of the protections in scope of the wider ‘no diminution’ commitment. There are other pieces of EU law that are relevant to the “Rights, Safeguards and Equality of Opportunity” chapter of the Agreement which have been implemented in retained EU law – via the EU (Withdrawal) Act 2018 – or domestic law in Northern Ireland. These include, but are not limited to, the Victims’ Directive, the Parental Leave Directive and the Pregnant Workers’ Directive, as well as specific measures aimed at protecting the rights of persons with disabilities. We do not envisage any circumstances in which those rights would be rolled back. However, provided that the rights in question are relevant to the aforementioned chapter of the Agreement, they are in scope of the UK Government’s commitment that there will be no diminution of rights as a result of the UK leaving the EU.

This document sat on the homepage of the Northern Ireland Office website for nearly six years, being decorously removed the day after the Dillon decision. The Northern Ireland Court of Appeal was not engaged in a flight of fancy in connecting the Victims’ Directive to the non-diminution commitment. It was relying on the UK Government’s public account of its obligations. It was not asserting an expansive or open-ended range of commitments, but drawing on a limited list of promises (even if the language of the above paragraph is not exhaustive).

For the UK Supreme Court, this inconvenient language was of no great importance. But to arrive at that conclusion, at [124], it had to adopt an unduly narrow account of the rules of treaty interpretation in international law:

[S]ubsequent practice can only be an aid to interpretation, ie to establish the meaning of the treaty provisions. The essential question here is not one of interpretation but one of the effect of the relevant provision, ie whether it has direct effect. Article 4 of the Withdrawal Agreement and section 7A of the 2018 Act make clear that the question of direct effect is governed by EU law. In particular, article 4 of the Withdrawal Agreement provides that persons shall be able to rely directly on its provisions “which meet the conditions for direct effect under Union law”. It is therefore only if it satisfies the requirements for direct effect in EU law that a provision of the Windsor Framework can be relied upon before a domestic court. That cannot be changed by practice establishing the agreement of the parties as to its interpretation.

This is an exercise in waving away the UK Government’s commitments. It had accepted that the Victims’ Directive could be used in this way, and now it wants to back away from those commitments. As to it being subsequent practice, or alternately at [124], ‘a unilateral statement of opinion by one party to the Withdrawal Agreement and Windsor Framework as to their effect’, the problem with such assertions is that Article 2 is an obligation only binding upon the UK. It is not a mutual obligation. The stated position of the party on whom the obligation rests matters.

Narrowing the Scope of the 1998 Agreement

There is a level of tragic comedy about the continuation of the Dillon litigation to the UK Supreme Court. The UK Government was fighting a case about the application of the Victims’ Directive to the legacy measures, when it is committed to removing the amnesty from those measures (which is the only issue to which the Victims’ Directive applies). This only makes sense as part of a broader effort to remove the constraints imposed by the Windsor Framework from wider policy making. In particular, in NIHRC’s Application, the non-diminution commitment in Article 2 was found to impact upon the application of the Illegal Migration Act 2023 in Northern Ireland, with implications for future immigration policy.

And, once again, the Supreme Court obliges the Government with a series of expansive obiter dicta. These are contestable, but expressed with such certainty by the five judges of the panel that it is unlikely that any lower court would question them. The court is predicting and responding to the UK Government’s desires. The key paragraph is 119:

[W]e note that the RSEO chapter is one section of an instrument, the Belfast Agreement, aimed at establishing peace in Northern Ireland after decades of sectarianism and civil conflict. Paragraph 1 speaks at a high level of generality of civil rights and religious liberties against the background of what, in 1998, was a recent history of communal conflict. The rights listed with bullet points are all concerned with ending sectarian conflict. These include freedom of political thought, freedom of religion, the right to pursue national and political aspirations, the right to seek constitutional change by peaceful and legitimate means, equal opportunities and freedom from sectarian harassment.

This approach to paragraph 1 of the RSEO chapter, according sufficient specificity only to the listed rights, neglects the fact that they were expressly presented as mere examples of civil and religious rights. The assertion that they were ‘all concerned with ending sectarian conflict’ is equally dubious. An assertion of the rights of ‘everyone in the community’ does not derive from a narrow sectarian basis and ‘the right of women to full and equal political participation’ marks this list of rights out as a basis for a tolerant and vibrant society after a conflict. This interpretive approach invents a restrictive account of the purposes of the 1998 Agreement and uses it to circumvent key elements of the text. 

Once again, the clear terms of Article 2 gives a lie to this approach. The Annex 1 Directives do not cover merely discrimination on ethno-nationalist grounds. Rather, they cover a broad sweep of EU anti-discrimination law (which remains applicable in the context, for example, of protections for trans people in Northern Ireland). The non-diminution commitment, moreover, is not dated to 1998, but instead covers any relevant developments of EU law up until the UK’s departure from the EU. The 1998 Agreement’s rights provisions were future facing; they were not defined by the conflict. The Court’s narrow approach therefore has the purpose of thwarting reliance on Article 2 in a broad swathe of cases.

The irony is that the application of the Victims’ Directive to the Legacy Act provides the clearest possible link between the EU law and addressing the “sectarian conflict”. When the implementing legislation was enacted in Northern Ireland, its relevance to helping to fulfil the 1998 Agreement’s commitments to victims was noted. If the Supreme Court’s framing of the issue is so constricted as to deny any connection between the Agreement and the Victims’ Directive, then there is no example of EU law that could conceivably be covered by Article 2 beyond the Annex 1 Directives. The clear language of the non-diminution commitment is rendered surplusage.

When Quasi-Justiciability is more dangerous than non-Justiciability

This move to actively undermine rights commitments demonstrates radical subservience of the Supreme Court to the UK Government’s agenda. This is facilitated by the litigation arising in the context of the legacy of the Northern Ireland conflict, with the Supreme Court in recent years making it quite clear, in cases such as McQuillan, Dalton and Thompson, that it would like to see the legacy cases taken off its agenda, and repeatedly narrowing the basis for applicable rights arguments. The very first paragraph of the judgment adopts the position on the conflict of the UK’s institutions of government being bystanders to 30 years of violent Irish atavism, and the judgment proceeds with this detached basis as its theme:

There have been longstanding conflicts in Ireland, including over the partition of the island. A key aspiration of Irish nationalists has been to bring about a united Ireland, with the whole island forming one independent state. This aspiration conflicts with that of unionists in Northern Ireland, who want that region to remain part of the United Kingdom (“the state”, in what follows). These conflicting aspirations together with many other pervasive and serious communal differences resulted in violence during what have commonly come to be known as “the Troubles”, leading to many tragedies with a deep and profoundly regrettable legacy of suffering. More than 3,500 people were killed during the Troubles, with approximately 40,000 injured. Around 1,200 cases relating to killings remain unsolved.

Whereas the Northern Ireland Court of Appeal was flagging extensive concerns with the legacy process, concluding that in many cases it could not operate according to effective human rights standards, the UK Supreme Court sought to neatly tidy up some of these concerns and punt others down the line for specific cases. One egregious example of this comes around the “security veto” within the Legacy Act. Schedule 6 of the 2023 Act gives the Secretary of State the power to prohibit disclosures by ICRIR on national security grounds. For the Court of Appeal, at [224], this amounted to “an effective veto”, undermining the independence of ICRIR in the legacy process. For the UK Supreme Court, at [225], the existence of judicial review over the ministerial decision was sufficient to inject independence into this process:

[I]t is misleading to categorise the Secretary of State as “the sole decision-maker.” The Secretary of State is restrained in the exercise of the power to prohibit disclosure, and any decision to do so is subject to challenge by way of judicial review. The Strasbourg court does not require absolute independence. … Absent a concrete examination of the circumstances of a specific case, it cannot be said that the restrained nature of the power, together with supervision by the courts in judicial review proceedings, does not provide adequate hierarchical independence for the ICRIR in any case …

This is an act of judicial gaslighting. We have, supposedly, moved on from the era of national security claims being non-justiciable. But as the Thompson case demonstrated, just a few months ago, the UK Supreme Court adopts an almost entirely hands-off approach to such security assessments around disclosures. Here quasi-justiciability is worse than non-justiciability. It ties families in knots pursuing legal claims where tests are weighed against them and that the government defends to the last ditch, and the Supreme Court seeks to present this rigmarole as providing effective independent oversight.

The human rights claims against the operation of ICRIR are treated by the Court as an ab ante challenge. For such a challenge to succeed on human rights grounds, the UK Supreme Court has, since Bibi, adopted the position that the breach of rights must be apparent ‘in all or almost all cases’. This explains the wide framing of the conflict by the Court. ICRIR primarily exists because of a series of complex and controversial cases involving the activities of state bodies or allegations of collusion between them and paramilitary groups leading to convoluted litigation. In those challenging cases, the issues raised by the Court of Appeal remain live, and to be litigated down the line. But because they are but one part of ICRIR’s overall workload, the UK Supreme Court excludes ab ante review of these issues. Note that, had the Windsor Framework arrangements not been curtailed, it would have been difficult to maintain this position because, in regard to Article 2, Parliament has expressly granted Northern Ireland’s statutory Human Rights and Equality Commissions the ability to intervene in the context of any ‘potential future breach’ of the Windsor Framework.

In adopting such a facilitatory stance, the UK Supreme has tied its credibility to the UK Government’s legacy policy. The Northern Ireland Court of Appeal was highlighting, on multiple levels, how the legacy scheme did not secure minimum rights standards. It therefore gave the distinct impression of being created to serve the UK Government’s desire to draw a line under the conflict and prevent further embarrassing revelations about the conduct of the police, military and security agencies in the courts. Just four days after the Supreme Court refused to entertain these concerns, the UK Government released a review by Sir Peter May that found that ICRIR to be a deeply dysfunctional organisation, insufficiently funded, divided, and undermined by inadequate legislative underpinnings. It gives the distinct impression of being a body set up to fail victims and their families, and the Supreme Court just gave the process more rope, further undermining the credibility of its oversight in legacy cases.

Judicial Fixers

The difficulty with all of the fulmination over “judicial power” in the UK is that the supposed powers which are asserted to be so problematic are very often not inherent to the judiciary, but have been granted to it by Parliament. The Northern Ireland courts’ approach to the Windsor Framework’s rights provisions is not the product of some swashbuckling pursuit of justice: it  has been impeccably grounded in the wording of the relevant provisions. In Dillon, the Court of Appeal noted, at [58], that ‘[o]ur role is simply to address the legal question put before us by means of interpretation of the [Withdrawal Agreement] and the relevant statutory provisions’. If the disapplication of a statute was at issue, it was because Parliament had provided for this. The Court adopted a particularly narrow approach to claims based around the Charter of Fundamental Rights.

It is the UK Supreme Court which strains to arrive at the interpretation desired by the Executive, True, the judges did not deny these plain effects of the Withdrawal Agreement. Instead they introduce direct effect hurdles where they have no place in the process, cast a blind eye to the UK Government’s promises, reframe the whole dispute to minimise the UK Government’s agency and cling to fictions of effective judicial review to achieve the requested result. In Dillon, the current UK Supreme Court, in its own fashion, has demonstrated that it is not a conservative court. Anything but. It is a radically regressive body where rights protections are at issue.

(Suggested citation: C. Murray, ‘Hollowing out Northern Ireland’s Post-Brexit Rights Protections in Re Dillon [2026] UKSC 15’, U.K. Const. L. Blog (14th May 2026) (available at https://ukconstitutionallaw.org/))