The literature on section 2 of the Human Rights Act 1998 has largely been preoccupied with a single direction of travel: how far domestic courts may go beyond the Strasbourg case law, or against it. Those are the questions behind Ullah, behind the refusal to follow Strasbourg in Horncastle, and behind the long debate over whether there is a ‘ceiling’ on domestic rights (see L. Graham, ‘The Modern Mirror Principle’ [2021] PL 523, and the work of Roger Masterman on the domestic–Strasbourg relationship). A Reference by the Attorney General for Northern Ireland of a devolution issue under paragraph 34 of Schedule 10 to the Northern Ireland Act 1998 [2026] UKSC 16 (‘the AGNI Reference’) runs in the other direction. Sitting as a panel of seven, the Supreme Court used the 1966 Practice Statement to overrule its own decision in Surrey County Council v P; Cheshire West and Chester Council v P [2014] UKSC 19 (‘Cheshire West’), holding that the ‘acid test’ for deprivation of liberty under article 5 ECHR had never been adopted by the European Court of Human Rights and was wrong in principle. Craig Wells has read the decision through a rule-of-law lens, stressing the consequences for those who lack capacity and the risk of arbitrary power (UKCLA, 15 June 2026); I approach it from another angle, though, as I suggest at the close, the two meet. My claim is narrow. The constitutionally distinctive feature of the AGNI Reference is not simply that the Court narrowed a Convention right, or even that it overruled itself to do so. It is the reason the Court gave for making that correction itself rather than leaving the matter to Strasbourg: a public authority cannot ask the European Court of Human Rights to correct an over-protective domestic reading of the Convention. What follows traces that move: its basis in precedent, the statutory link to article 5, the Article 34 asymmetry that made domestic correction necessary, and its 2012 pedigree, before turning to why it matters.

The mirror principle, reversed

The substantive holding — that ‘valid consent’ under article 5 is an autonomous Convention concept, not to be equated with legal capacity in domestic law — will no doubt be analysed from a number of perspectives. The move that interests me lies in the Court’s reasoning on precedent (paras 41–51). Lord Bingham’s formula in Ullah was that national courts should keep pace with Strasbourg, ‘no more, but certainly no less’. The usual version of that idea is prospective: should a domestic court follow Strasbourg, anticipate it, go beyond it, or refuse to go against it? Manchester City Council v Pinnock [2010] UKSC 45 illustrates the familiar upward version: there the Supreme Court departed from recent House of Lords authority in Qazi and Kay in order to bring domestic law into line with a clear Strasbourg position. AGNI is the reverse. The Court did not depart from precedent because Strasbourg had moved on; it did so because it concluded that its own earlier domestic law had moved too far beyond Strasbourg. The ‘no more’ limb therefore became a retrospective technique of self-correction through precedent.

The Court did not take the step lightly: it acknowledged (para 44) that it invokes the 1966 Practice Statement only rarely and sparingly, and never merely because a later panel would have resolved a question differently — the restraint set out by Lord Bingham in Horton v Sadler. That a seven-member Court should unanimously depart, twelve years on, from what had been a majority holding in Cheshire West only sharpens the point.

Two features make the departure more striking still. First, it was common ground that there had been no relevant change in the Strasbourg jurisprudence since 2014 (para 47). So this was not a case of Strasbourg moving and the United Kingdom Supreme Court following; it was the Court revisiting, and disowning, its own earlier reading of Strasbourg. The baseline to which it returned is thus, in part, the Court’s own reconstruction of what Strasbourg requires. The point must be put with care, because it cuts unevenly across the two elements of the ‘acid test’. As to the objective element, the claim to be returning to Strasbourg is relatively straightforward: the judgment presents the multifactorial line running from Engel and Guzzardi through cases such as Stanev as a settled approach from which Cheshire West had departed. The subjective element is more difficult. The reworked concept of ‘valid consent’ is more contestable, and the charities and the Official Solicitor disputed that Strasbourg required any such reworking at all. ‘Fidelity’ to Strasbourg therefore cannot be claimed evenly across the whole decision; that unevenness is part of what makes AGNI constitutionally interesting. Second, the correction ran in the direction that narrows the protective reach of the right, moving some — and potentially many — people who lack relevant capacity outside the automatic article 5 safeguards that Cheshire West had brought them within.

A statutory mirror?

Here, though, an objection arises. Both the Mental Capacity Act (Northern Ireland) 2016, section 306, and the Mental Capacity Act 2005, section 64(5), define ‘deprivation of liberty’ by direct reference to article 5 ECHR. It might therefore seem that statutory incorporation, rather than section 2 HRA, is what ties the domestic test to Strasbourg — so that the case is not really about the mirror principle at all.

That is not how the Court reasoned. In paragraph 46 it placed the exercise within the mirror-principle canon, citing Ullah,Pinnock, R (AB) v Secretary of State for Justice and Elan-Cane. It treated the direct statutory link not as displacing the mirror principle, but as a reason to apply that principle a fortiori. The obligation to track Strasbourg was, the Court said, stronger here than under section 2 HRA, which requires only that Strasbourg case law be taken into account.

Whether that is right — whether statutory incorporation of a Convention concept truly engages the logic of the mirror principle, or only resembles it — is contestable. For present purposes, the label of the mirror principle tracks how the Court itself reasoned.

Why the correction had to be domestic

Paragraph 49 carries the distinctive institutional point, though it is not the whole reason for the result: the Court also relied on the direct statutory link between the mental capacity legislation and article 5 (para 46), on the confusion produced by divergence between Cheshire West and Strasbourg (para 48), and on the practical consequences and expense of maintaining the earlier test (para 50). But paragraph 49 explains why the Supreme Court, rather than Strasbourg, had to make the correction. If Cheshire West had over-read article 5, that error could not be corrected in the ordinary way by the European Court, because the public authorities burdened by the over-reading have no standing to take the point to Strasbourg (the Court citing R (AB) v Secretary of State for Justice [2021] UKSC 28, [56]–[57]). From that asymmetry the Court drew a positive conclusion: the Supreme Court has ‘a particular responsibility’ to keep domestic article 5 law aligned with what it understands the Strasbourg jurisprudence to require. The asymmetry, in other words — normally invoked (if at all) as a reason for caution about going beyond Strasbourg — here becomes a reason to come back to it.

From Sales 2012 to AGNI 2026

That justification has an important antecedent. In substance, it restates an argument that Lord Sales, then Sir Philip Sales, made extra-judicially fourteen years ago. In ‘Strasbourg Jurisprudence and the Human Rights Act: A Response to Lord Irvine’ [2012] PL 253 — a defence of the mirror principle against Lord Irvine’s lecture ‘A British Interpretation of Convention Rights’ [2012] PL 237 — he identified precisely this imbalance: if a domestic court is too cautious, the aggrieved individual can apply to the European Court and the error can be corrected; if it is too generous, the losing public authority ‘has no right of application to the ECtHR’, and the error cannot be corrected. He traced the point to Lord Brown in Al-Skeini, and tied it to the Article 34 ‘victim’ requirement that section 7(7) of the HRA imports. The continuity is real, but it should be stated precisely, because the inference has changed. In 2012 the asymmetry counselled prospectivecaution: a reason not to create an over-expansion that no one could later undo. That premise is entirely compatible with never overruling anything — it argues for restraint going forward, not for retrospective correction. AGNI draws a different and more interventionist conclusion from the same premise: that an over-expansion which has already occurred ought to be undone. The shared premise is Sales’s; the step from ‘be cautious in future’ to ‘overrule what already exists’ is the Court’s own. The continuity should not be overstated in the other direction either: AGNI is a unanimous judgment, jointly given by Lord Sales and Lady Simler, with the other five Justices agreeing. But it is striking that an argument once advanced by a single judge, writing extra-judicially in defence of the mirror principle, should resurface as the unanimous reasoning of the Court.

How far can the technique travel?

On this view, the interest of AGNI lies less in any label than in how three bodies of doctrine, usually analysed apart, are made to combine: the mirror principle and section 2 HRA; the doctrine of precedent and the 1966 Practice Statement; and the structural asymmetry of access to Strasbourg under Article 34. Strip out the asymmetry and little remains that is distinctive — every overruling of a precedent thought to have misread the law is retrospective, and every overruling of a rights-expanding precedent narrows a right. What the asymmetry supplies is something narrower and, so far as I am aware, novel: not a reason why Cheshire West was wrong, but a reason why any correction, once warranted, had to be made at home rather than in Strasbourg. It is that institutional allocation — an apex court treating itself as the only forum that can pull its own over-protective reading back into line — that raises the real question: its constitutional legitimacy and its limits.

The same technique can be described in two ways. On one view, it is a discipline of fidelity to Strasbourg as an external standard. On another, it is a judicial power to retract domestic rights protection without any intervening Strasbourg correction. Which description fits turns on two questions I take up in a forthcoming article: whether AGNI is really a section 2 phenomenon at all, or an exercise in reinterpreting an incorporated concept; and whether ‘fidelity to Strasbourg’ can bear any weight when, at least as to valid consent, what Strasbourg requires was itself contested. This is also where the present reading meets Craig Wells’s. His rule-of-law concern is that the decision enables arbitrary power over those who lack capacity. The structural concern set out here is different, but related: it asks whether the Court may legitimately use precedent to retract rights protection in the name of a Strasbourg baseline it reconstructs for itself. If I had to commit provisionally, I would say the technique is easiest to defend in the narrow case the Court actually faced — a domestic statutory concept tied directly to article 5 — and harder to defend if generalised beyond that setting, for example to an ordinary HRA case in which there is no statutory definition incorporating a Convention concept, but the Court nevertheless treats the Article 34 asymmetry as a freestanding reason to cut back an earlier domestic rights-expansion. What I do here is narrower: I name the move and locate its pedigree — downward self-correction, a court using its freedom over precedent to bring its own rights jurisprudence back down, resting at the decisive point on a justification that one of the judgment’s own authors had set out extra-judicially in 2012.

Hleb Buziuk is an Independent Scholar.

(Suggested citation: H. Buziuk, ‘Downward Self-Correction: The AGNI Reference and the Mirror Principle in Reverse’, U.K. Const. L. Blog (30th June 2026) (available at https://ukconstitutionallaw.org/))