The Second Forum on Domestic Violence and the Hague Convention: One Step Forward, Two Steps Back

In June 2024, the Permanent Bureau of the Hague Conference on Private International Law (Hague Conference – Conférence de La Haye, or HCCH) convened the first-ever Forum on Domestic Violence and the Related Operation of Article 13(1)(b) of the Hague Convention on the Civil Aspects of International Child Abduction (the “1980 Convention”). That gathering, held in Sandton, South Africa, was a historic moment. For the first time, the HCCH brought judges, “Central Authorities”, advocates, psychologists and, most importantly, survivors of domestic violence into the same room to discuss whether the 1980 Convention was serving the children it was designed to protect. This author covered that Forum in the Child Relocation 2024 Global Practice Guide, noting the cautious optimism it generated:

• a recognition of the different forms domestic violence can take; the harm it causes to children; and
• a growing judicial willingness, as reflected in the US Supreme Court’s 2022 decision in Golan v Saada, to reduce the reliance on protective measures as a condition of return.

The optimism was reinforced by the Secretary General Christophe Bernasconi’s closing observation that success under the Convention “is not about the sheer number of returns but really about the correct application of the Convention”. The First Forum did not resolve the complexities at the intersection of domestic violence and the Convention. What it did, however, was acknowledge the problem and open the door to a productive dialogue.

That dialogue continued approximately sixteen months later, this time in Fortaleza, Brazil. From 27 to 30 October 2025, the HCCH and the Government of Brazil co-hosted the Second Forum on Domestic Violence and the Operation of Article 13(1)(b). Nearly 500 participants registered, more than 150 of them in person. The Second Forum should have advanced the work begun in Sandton and moved toward resolving the practicalities of how to best protect victims within the framework of the Convention. Unfortunately, the opposite happened. While the evidence assembled at Fortaleza was even more detailed and compelling than what had persuaded Sandton’s participants of the need for a serious conversation, the institutional posture with which that evidence was received appeared noticeably narrower than the one Mr Bernasconi had articulated sixteen months earlier.

This article examines that reversal. It is offered from the perspective of a practitioner who views the 1980 Convention as essential, but believes that a course correction in the way it is implemented has become necessary. The Convention’s original purpose was to protect children from the harmful effects of international abductions by promptly returning them to their home country. With 46 years of evidence, we now understand that not every “abduction” is harmful and not every return is protective. A more nuanced implementation of the Convention is needed to accomplish its true goal of protecting children.

The data presented in Fortaleza fleshed out the problems identified in Sandton

The data and other evidence presented in Fortaleza was clear, specific and detailed, albeit bleak. It filled in the gaps left open in Sandton.

Professor Merle Weiner (University of Oregon) presented research, drawing on journalist Olivia Gentile’s analysis of United States cases between July 2022 and June 2024, showing that 77% of return petitions in that period were filed by fathers against mothers, consistent with international data compiled separately by Professor Nigel Lowe and Victoria Stephens. Among respondent mothers, 79% alleged domestic violence or child abuse in defending against return; only 17% of those who raised the Article 13(1)(b) defence at trial succeeded in proving it. Meanwhile, 55% of fathers accused of abuse obtained their children’s return. Of the mothers Professor Weiner’s research team interviewed whose children were returned, 59% reported severely adverse custody outcomes, including losing their relationship with their children entirely, whether through inability to safely return themselves, deportation, minimal or supervised visitation, or the other parent’s non-compliance with the terms of return.

The picture was similar outside the return context. Dr Nishat Hyder-Rahman presented findings from a GlobalARRK survey of 165 respondents on the lived experience of relocation proceedings: of those who answered the survey’s question on domestic violence, 93.2% reported having experienced it. Yet, those who disclosed the abuse in their proceedings mostly reported that it was given limited weight, or ignored outright, by the court. Some had been advised by their own lawyers not to disclose the abuse at all, for fear of appearing uncooperative with contact arrangements.

Dr Braden Clark cited a 2020 inquiry by the government of New South Wales, Australia, finding that coercive control was present in 99% of intimate partner homicides in that jurisdiction, with no physical violence previously reported in approximately 25% of those cases. As is now well-understood, separation is statistically the most dangerous moment for a victim: 77% of Australian women killed by their partners had separated or were in the process of separating.

And on the issue of protective measures, the data, assembled at scale, was consistent across jurisdictions. Ruth Dineen and Roz Osborne, presenting joint research from FiLiA Hague Mothers and GlobalARRK, reported that 71.4% of mothers who returned under protective measures faced further abuse despite these measures, a figure consistent with a 2003 Reunite International study finding two thirds of undertakings in the United Kingdom cases were never implemented, and that even mirror orders were honoured in only one out of five cases.

The protective measures problem – revisited

A full session during the Second Forum was devoted entirely to protective measures. Speaker after speaker – Ms Osborne, Ms Dineen, Judge Myriam de Hemptinne and Justice Steven Arnold Majiedt – walked through the same conclusion from different jurisdictions: undertakings are rarely enforced, mirror orders are rarely honoured, and the practical safety net that return orders assume is often not there in any meaningful sense. Dr Adrienne Barnett (Brunel University of London), presenting on behalf of Hague Mothers, situated the protective measures problem within a broader institutional pattern, observing that the assumption that protective measures can mitigate risk “is underpinned by [a] physical incident model of domestic violence” and “fails to acknowledge high level of recidivism and breaches, and broader risk and harm”. She pointed to the recent English decision in J. v. K. [2025] EWHC 2125 (Fam), which had quoted with approval Cobb J.’s direct formulation in Re T (Abduction: Protective Measures: Agreement to Return) [2023] EWCA Civ 1415 that “Protective Measures need to be what they say they are – protective. To be protective, they need to be effective”. Ms Dineen summed up the implication for cross-border cases: “if protective orders fail domestically, they cannot be expected to work across borders”.

The institutional response, articulated by Christian Höhn on behalf of Germany’s Central Authority, was to double down. Return is the rule, non-return is the exception, and protective measures are the tool ensuring “safe” returns despite the risk of grave harm. That this remains the mainstream approach across the majority of the contracting states in the face of the mountain of evidence that shows protective measures do not work is truly baffling. Even more concerning, it is also the approach the HCCH continues to endorse in its Guide to Good Practice.

The reform that is not coming

Similarly, a full session was dedicated to reviewing the HCCH’s Guide to Good Practice on Article 13(1)(b). Such a review was timely, if not overdue, given the five years that have passed since its publication. Boni de Moraes Soares, an attorney at Brazil’s Attorney General’s Office, presenting on behalf of Brazil, questioned whether three of the Guide’s central features reflect either the current understanding of domestic violence or the text of the Convention itself. The Guide’s own language is instructive. It provides that “harm to a parent, whether physical or psychological, could, in some exceptional circumstances, create a grave risk” to the child (emphasis added). It directs that the grave risk analysis turn on “such considerations as the nature, frequency and intensity of the violence”, and states that “evidence of the existence of a situation of domestic violence, in and of itself, is therefore not sufficient to establish the existence of a grave risk to the child”. Mr Soares questioned why sexual abuse, for instance, should require any assessment of frequency or intensity to qualify as grave risk, an obviously fair question, given that the Guide itself elsewhere states there is no separate, lower test for any particular category of harm: “the Convention does not provide for different tests to assess a grave risk on the basis of the type of risk”.

It would be inaccurate to suggest that the Guide ignores domestic violence. Its companion volume on mediation devotes an entire chapter to screening for domestic violence and safeguarding victims who go through mediation, and the Special Commission has separately confirmed that a child need not be the direct target of violence for the exception to apply. The concern Mr Soares and others have identified is deeper. The burden to avoid return – ie, the “exceptional circumstances”, assessed by “nature, frequency and intensity” – demonstrates a fundamental lack of understanding of the pervasiveness of domestic violence and the frequency of institutional failures when responding to it.

The Guide’s treatment of protective measures compounds this concern. Even when a court finds that returning a child would create a grave risk of harm, the Guide recommends considering whether “adequate arrangements”, most commonly undertakings from the taking parent or mirror orders in the requesting state, can be put in place to ameliorate that risk and allow the return to proceed. As set out above, the empirical record on the effectiveness of such measures is appalling: the 71.4% figure from FiLiA Hague Mothers and GlobalARRK, the 2003 Reunite findings on undertakings and mirror orders, and the consistent testimony of practitioners across jurisdictions all point in the same direction. See The Guide to Good Practice on Article 13(1)(b): Domestic Abuse Perspective.

In his closing remarks, Mr Bernasconi was candid about the institutional runway: any revision to the Guide will depend on a decision by CGAP (the HCCH’s Council on General Affairs and Policy), which in turn will want to be informed by a future meeting of the Special Commission, a body that meets every six years. On that basis, the Secretary General noted that work on revising the Guide, or on drafting a separate instrument specific to domestic violence, “cannot realistically begin within the next five years”.

That timeline had already been set. At its March 2025 meeting, seven months before the Second Forum, CGAP had resolved that “the second Forum, like the first, will not result in any HCCH Conclusions and Recommendations”, and that “any proposal for an additional Forum involving the HCCH... shall be presented to CGAP for consideration”. Before a single participant travelled to Brazil, the question of what would be done with the evidence they would assemble had been answered: nothing binding, and nothing further, without another round of approval from the same body that had just foreclosed the current round. This was perhaps the precise moment at which the door opened in Sandton began to close.

The label problem

A recurring theme through the Second Forum’s discussion, surfacing across multiple sessions, in different formulations and from different speakers, was whether the Convention’s labelling of “wrongfulness” in cases of domestic violence accurately describes the problem for the population of cases to which the term is most frequently applied.

Under the Convention, a child’s removal in breach of the left-behind parent’s custody rights is deemed “wrongful”. The label itself presents certain difficulties. Several speakers questioned whether the treatment of a removal by a primary-caregiver parent fleeing documented domestic violence should be conceptually and evidentially identical to the treatment of a removal calculated to defeat the other parent’s lawful custody rights in the absence of any such factor. The two removals plainly have different purposes, raise different concerns and reflect different motivations, but the Convention does not allow for this nuance.

This is not a new question and it did not originate at Fortaleza. But the empirical evidence reaffirmed at the Forum – that 77% of respondents were mothers, 79% raised domestic violence allegations and 71.4% of returned mothers faced continued abuse despite protective measures – makes the definitional question harder to defer.

This question surfaced during the Secretary General’s opening remarks, and it is here that the distance between the two Forums is perhaps the most visible. At Sandton, in his closing remarks, Mr Bernasconi had offered what many participants took as the animating principle of the Forum’s project: that success under the Convention “is not about the sheer number of returns but really about the correct application of the Convention”. That formulation did not prejudge outcomes. It made the analysis, rather than the disposition, the measure. It was the sentence that most obviously kept the door open – a door through which the empirical evidence, the reform proposals and the definitional questions raised at Fortaleza were meant to walk.

Yet, in his opening at the Second Forum, the same Secretary General took a noticeably narrower stance. Responding to what he described as a recent line of argument suggesting the Convention should distinguish between “harmful abductions” and “protective abductions”, Mr Bernasconi said: “I respectfully disagree with this distinction. Let us be clear: when custody rights are violated, the removal or retention is wrongful. An abduction inflicts harm. Domestic violence inflicts harm. Both are wrong. Both cause trauma”. The Convention, he emphasised, “does not begin with labelling some abductions as “protective” – it begins with the rule that abductions are wrongful, followed by the question of whether an exception to the principle of a return applies”.

The two statements of the Secretary General are not immediately or obviously compatible. If it is the correct application rather than the number of returns that is the true measure of the Convention’s success, then the language used to describe the removals that trigger it is not incidental to the analysis. It is an essential part of what makes the analysis correct or incorrect. A category that treats fleeing domestic violence as identical, at the threshold, to retaliatory removal is not a neutral doctrinal starting point; it is a choice that shapes how correct application is defined and by whom. Labels, in other words, matter, and terminology is not the opposite of substance. It is one of the tools through which substance is either honoured or eroded.

This point has been developed recently in scholarly terms as well. Jeffrey Edleson and Emma Katz, in a forthcoming article in the Journal of Gender-Based Violence, argue that Article 13(1)(b)’s explicit references to “exposure” to grave psychological harm and to “intolerable situations” on their face already provide the textual foundation for denying return in cases of documented coercive control. All that is needed is to give these terms their due.

The institutional defensiveness is particularly striking considering that it stands apart even from the position now advanced by the Convention’s own rapporteur. In a February 2024 interview published in the Uruguayan newspaper La Diaria, Elisa Pérez-Vera, author of the 1981 Explanatory Report whose interpretation of the exceptions has for decades guided the Convention’s application, acknowledged that “gender-based violence was not taken into account… we had not become aware that it was a phenomenon” when the Convention was drafted. Pérez-Vera went further, offering a contemporary reinterpretation: “I believe that, almost 50 years later, we need to reinterpret the letter of the Convention in light of the new social realities in which it has to be applied”, and insisting that “from the point of view of interpreting Article 13, domestic violence and gender-based violence should be a factor to be taken into consideration both by the judges in the requested state, who have to decide whether or not to return the child, and by the judge who naturally, according to the Convention, would have to decide on the custody relationship”.

Mr Bernasconi’s own post-Forum footnote, appended to the published text of his opening address after “valuable discussions during the Forum”, acknowledged that some removals “may indeed be prompted by a genuine and instinctive – often motherly – desire to protect the child”, while maintaining that “the Convention does not distinguish between types or motives of abduction”. That the footnote was added at all suggests the Forum moved something. That it maintained the definitional rule intact suggests how much of the Sandton framing – and how much of the interpretive space Pérez-Vera herself is now inviting – is yet to be recovered.

Where this leaves the Convention

None of this is intended as criticism of the Hague Convention itself, much less as an argument for weakening it. The 1980 Convention remains an essential and effective tool for protecting children from the harmful effects of international abduction, and its goals are as important today as they were 46 years ago. Victims of domestic violence also depend on the Convention’s correct application, including when they seek the return of children wrongfully removed by abusive partners. The evidence presented in Fortaleza does not call for changes to the Convention’s structure, but for changes in how it is interpreted and applied:

• careful, fact-sensitive attention to the dynamics of domestic violence, including coercive control;
• consistent trauma-informed training for judges who apply the Convention; and
• a willingness to let definitions and interpretive frameworks evolve in light of 46 years of practice and research, as the Convention’s own rapporteur now suggests they must.

As Dr Barnett put it, what is needed is a “systemic and fundamental change in the understandings, perceptions, attitudes and approaches of all those working to apply and implement the law”. That is a cultural shift in service of the Convention’s purpose.

The Convention touches thousands of children and families each year. Implementing these changes would directly recognise and honour its protective purpose. The distance between Sandton and Fortaleza has become its own data point: evidence of how difficult this cultural and institutional shift remains. The next Forum, whenever it is held, will inherit both the evidence presented at Fortaleza and the unresolved question of which framework will guide the response – the openness of Sandton or the retrenchment of Fortaleza.

Mr Bernasconi’s Sandton observation – that correct application, not the number of returns, is the true measure of the Convention’s success – remains the most useful sentence the HCCH has produced on this subject. Whether the institution is prepared to be held to that standard is the question this Forum left for the next.