AG Emiliou: Lis Pendens, Mutual Trust, and the Cost to Choice-of-Court Agreements

Emilia Sandri is a doctoral candidate at the University of Antwerp.

Yet another victim could be claimed by the free circulation of judgments: this time, it is choice-of-court agreements that must face the test of mutual trust. On 23 April 2026, Advocate General Emiliou delivered his Opinion in Babcock Montajes: a lis pendens case raising uncomfortable questions about how far we are prepared to go in the name of the free circulation of judgments. While the Opinion has already been commented on, either questioning or cautiously accepting recognisability, this post examines what it reveals about the practical value of choice-of-court agreements and, more broadly, the limits of mutual trust as a foundational principle.

Background  

Companies A and B concluded a contract with an exclusive choice-of-court clause favouring Cologne courts. After a dispute arose, Company A sued in Madrid while Company B sued in Cologne. The Madrid court ruled first, declaring Spanish jurisdiction. The  Cologne court recognised this as a “judgment” under the Brussels I bis Regulation and declined jurisdiction. Company B submitted a request to Spanish courts to re-examine their jurisdiction, which was refused. On appeal, a higher Cologne court found that the Spanish ruling was not a “judgment”, and that German proceedings could continue as Spanish courts had breached the exclusive choice-of-court clause. The case was referred to the European Court of Justice, asking whether the Spanish decision constitutes a “judgment”, and if its recognition depends on its potential alteration during the proceedings.

Opinion of the AG

AG Emiliou first assessed whether breaching a choice-of-court agreement could justify refusing recognition, answering in the negative. He explained that, under the Regulation, the enforcing court may not review whether the issuing court correctly applied jurisdiction rules due to the principle of mutual trust (para. 40). Article 45(1) exhaustively lists refusal grounds for the recognition of judgments, none of which cover choice-of-court agreement breaches (para 42). This exhaustivity is unaffected by Article 31(2), a provision which allows a designated court to decide on its own jurisdiction even if it was not the first court seised, displacing other courts’ jurisdiction (para. 50).

The AG then addressed whether a decision affirming jurisdiction in breach of a choice-of-court agreement constitutes a “judgment”, answering affirmatively. Relying on Article 2(a)’s broad definition and case law qualifying as judgments any court decision open to adversarial proceedings (paras. 63-65), the AG saw no meaningful distinction between jurisdictional decisions and those on substantive rights (paras. 70-73). Including such decisions is necessary for lis pendens efficacy – otherwise, jurisdictional decisions would not trigger second-seised courts’ obligation to decline jurisdiction (paras. 74–77). Since Article 45(1) excludes lis pendens breaches, jurisdictional challenges are left to the issuing Member State’s remedies (para. 87). Moreover, allowing designated courts to proceed despite prior proceedings elsewhere would create parallel proceedings, contrary to the Regulation’s aims (para. 89). The AG concluded that effective judicial protection is preserved even if the first-seised court later revises its jurisdiction, since second-seised courts need only decline jurisdiction once the first-seised court’s jurisdiction is no longer contestable – at which point, it will likely rule on the merits too (paras 95-96).

The paper tiger of choice-of-court agreements      

The purpose of lis pendens rules is to prevent parallel proceedings. Under this mechanism, any court second-seised must stay proceedings until the first-seised court establishes jurisdiction and, if so, decline its own. The Regulation’s lis pendens regime, described in Section 9, comprises Article 31(2) on choice-of-court agreements. This provision takes precedence over general lis pendens rules, allowing a designated court to decide on its jurisdiction even if not seised first. Article 31(2) was introduced to prevent “torpedo actions” where, despite a choice-of-court agreement, a party seises a non-designated court first to secure priority under general lis pendens rules.

The AG’s approach potentially incentivises breaching this provision. Under his reading, a party obtaining a jurisdictional decision from a non-designated court first is essentially rewarded. In the best-case scenario, once that decision becomes final, the designated court must permanently step aside, meaning the “torpedo” party effectively chooses its own forum despite the exclusive jurisdiction clause. In the worst-case scenario, it succeeds in freezing the designated court’s proceedings while, at least temporarily, litigating in the preferred forum. The latter case is not a contrived scenario. The “Italian torpedo” – named after Italy’s notoriously slow courts – describes precisely this abuse: a party rushes to sue in a slow court not to win there, but to paralyse legitimate proceedings elsewhere. This problem concretised in 2003 with Gasser, where the Court held that even excessively long proceedings in the first-seised court could not justify derogating from lis pendens rules. As legal scholarship noted, this facilitated such abuse to the point that lawyers had to advise clients to sue immediately rather than attempt settlement, since being first was crucial to secure jurisdiction in a preferred forum. This Opinion risks reviving the same dynamic, as others have observed.

Article 31(2) was introduced precisely to address these situations, but the AG’s solution neutralises its effects. Company B concluded a valid choice-of-court agreement and sued before the designated court – yet, since Company A moved faster in breach of that agreement, Company B now faces frozen proceedings while the non-designated court proceeds. The AG justifies this by observing that Article 31(2) is also aimed at preventing parallel proceedings, so once the latter cannot be avoided, the first-seised-court rule should apply – a view shared by some commentators. But this proves too much. Article 31(2) exists because the legislature considered the general rule insufficient to protect choice-of-court agreements. Allowing it to be displaced whenever parallel proceedings have materialised makes it effective only where it is least needed: when the non-designated court voluntarily steps down. When a party acts in breach of the agreement – precisely when the provision matters most – it offers no protection. If a choice-of-court agreement can be neutralised simply by moving fast enough, what does it actually guarantee?

Mutual trust: a great premiss for a paradoxical application   

This Opinion illustrates how far mutual trust has become disconnected from how courts behave. Courts trust each other because, despite differences between their legal systems, they operate within the same EU law boundaries, as reflected in Opinion 2/13. The prohibition of substantive review of foreign judgments gives concrete expression to this presumption: if courts can be trusted to behave, there is no need to verify their results. By the same logic, courts are equal in applying jurisdictional rules, so there’s no need to question their assumption of jurisdiction.

However, mutual trust depends on both sides honouring their obligations. Article 45(1) acknowledges this, listing grounds for refusing recognition precisely because trust cannot be absolute. Lis pendens rules are not in that list: they seem to be subject to unconditional trust, and their breach must be addressed through the issuing Member State’s legal remedies. As confirmed in Liberato, such violations cannot justify refusal of recognition via public policy, as that would amount to a prohibited jurisdictional review.

Yet, the assumption that courts respect EU law cannot shield its breach. Here, jurisdiction was taken in violation of Article 31(2) – not through a difficult interpretative question, but by plainly ignoring a choice-of-court agreement. Company B sought review and Spanish courts refused, bearing no consequences while mutual trust preserved their decision. The only remedy for Company B is to appeal in the Spanish system: the one that already twice upheld its jurisdiction. It has been argued by some that Company B must accept this result, having failed to challenge jurisdiction within the first-seised court system. Yet Company B sought a re-examination of jurisdiction before Spanish courts and was turned away. The remedy was not practically available, which is precisely what makes reliance on mutual trust so difficult to sustain here.

Moreover, the case law might offer other avenues. In Charles Taylor Adjusting, the ECJ held that a ruling functioning as a quasi-anti-suit injunction may be refused recognition through public policy. The Court drew a distinction: while jurisdictional rules cannot be reviewed under public policy, judgments impairing the jurisdictional system’s proper functioning can trigger the exception. The quasi-anti-suit injunction crossed that line as it undermined the fundamental principle that each court must decide its own jurisdiction. If a ruling can be refused recognition because of its harm to the jurisdictional architecture, recognising one made in clear breach of Article 31(2) – a provision designed to protect that architecture – raises the same concern. This suggests an alternative, yet cumbersome, route for the enforcing court: letting the non-designated court decide on the merits, and then refusing recognition of the judgment through public policy – possibly even adjusting the ruling’s content, following Real Madrid. That such a convoluted path may be the only meaningful remedy is itself telling. Does the free circulation of judgments really benefit from all this complexity? And, more deeply, what practical value does mutual trust retain when its application rewards those whose disregard the rules on which it depends?

Conclusion      

The AG’s Opinion reveals a fundamental tension: mutual trust and free circulation of judgments risk rewarding non-compliance with EU law. By requiring automatic recognition of jurisdictional decisions violating choice-of-court agreements, the Opinion strips Article 31(2) of meaningful protection. The result is a system where speed matters more than legality, and where mutual trust shields the very breaches that should prevent its application. Confronted with this case, the Court can now decide whether that is the system that it actually intends.