Promises at the Border: Undertakings, Mirror Orders, and the Australian Case That Reframed Grave Risk

Executive Summary

Nearly half of all contested Hague refusals now turn on "grave risk" — and nearly every hard grave-risk case asks whether the danger can be managed, so the child can be returned safely. The tools are undertakings (promises to the returning court), mirror orders (the same protections entered in the home court), and safe-harbour packages. The High Court of Australia in DP v Commonwealth Central Authority; JLM (2001) held that grave risk means its ordinary meaning — no artificial narrowing — and is assessed against the real, protected situation on return; the case turned when the father reorganized his life to make his autistic son's return survivable. But research shows undertakings are often broken once the family is back, and the US Supreme Court in Golan v. Saada (2022) held that courts may consider protective measures but need not engineer a return at any cost, especially where domestic violence is found. This article is educational and not legal advice.

Introduction

Nearly half of all contested refusals in the Hague system now turn on one clause: Article 13(1)(b), the "grave risk" exception. And nearly every hard grave-risk case ends up asking the same practical question: can the risk be managed? If the child would face harm on return — a violent household, a carer facing arrest, a medical need — can the left-behind parent, or the home country's courts, put protections in place that make return safe?

Legal Background: return, not custody — and the tools that make return safe

A Hague return order does not decide custody; it returns a wrongfully removed child to their country of habitual residence, whose courts then decide the parenting questions. When a parent raises a genuine grave-risk defense, the question is whether protective measures can make the return itself safe enough to order. Three instruments, in ascending order of strength: undertakings — promises given by the left-behind parent to the returning court; mirror orders — the same protections entered as orders of the child's home court before return, enforceable where the family will live; and safe-harbour / protective packages — interim housing, support, and non-molestation or non-prosecution arrangements settled in advance. They are the hinge on which the modern Convention swings: between the Neulinger dead-end, where return is impossible because the carer cannot safely come (article #6), and the summary return the treaty promises.

What happened

The child at the centre of DP v Commonwealth Central Authority (referred to in the judgment as "M") was born in Greece in November 1994, to a Greek father and a Greek-born Australian mother. The marriage failed; in 1998 the mother moved with the boy to her parents' village nearby — and then, without the father's consent, to Australia. A Hague application followed.

What made the case extraordinary was the child's situation: M was autistic, and his condition anchored the mother's defense. In the region of Greece where the father lived, the specialized treatment M was receiving in Australia was, she argued, simply unavailable. Returning him would not merely relocate a child; it would sever a disabled child from the services his development depended on. If Article 13(1)(b) means anything, she argued, it means this.

The case, heard together with a companion appeal — JLM, in which a mother's defense was the grave risk of her own suicide if her child were returned to Mexico — reached the High Court of Australia, which ruled on 27 June 2001. Two holdings changed the field's grammar:

1. Grave risk means what it says — no extra-narrow gloss. Lower courts across the Convention world had developed a habit of reading Article 13(1)(b) with added restrictions — treating it as available only in the most extreme cases, "construed as narrowly as possible." The High Court rejected the gloss: the exception is to be applied according to its ordinary meaning. It is a demanding test on its own terms — the risk must be grave — but courts may not raise the bar beyond the treaty's text. (The same de-glossing move that the House of Lords would make for the discretion stage in Re M (article #5), and that Strasbourg would make for procedural scrutiny in X v. Latvia (article #3): the Convention works best when courts apply it rather than armour it.)
2. The risk is assessed against the real, managed situation on return — not the worst case. The matter was remitted for rehearing on the actual evidence: what would this child's life in Greece look like, with what services, under what arrangements?

And then came the detail that makes this a parable rather than a footnote: by the rehearing, the father had moved — from his village to Thessaloniki, Greece's second city, where expert evidence documented genuine facilities for autistic children. The grave risk, as pleaded, had been engineered away by a parent willing to reorganize his life around his son's needs. The return followed. No doctrine did that. A father did — and the legal system's job was to give his adaptation legal force.

Case Study Analysis — the uncomfortable evidence about promises

Undertakings work in courtrooms. The harder question — documented by the field's own research — is whether they work after the plane lands.

The UK charity reunite commissioned follow-up research on families after return (Freeman, The Outcomes for Children Returned Following an Abduction, 2003), and its findings remain the standard caution: undertakings given to secure returns were frequently breached once the family was back, and the returning court had no practical way to enforce them. A promise made in Sydney or London can evaporate at the arrivals hall in another jurisdiction; the receiving country's courts never made the order and may not recognize it. Practitioners know cases where the promised apartment, support payments, or non-prosecution never materialized — with the returned carer and child bearing the shortfall (see this series, article #29).

Two decades of system-building have answered with harder instruments: undertakings (fast, flexible, but only as good as the promisor's honour and the destination's attention); mirror orders (enforceable where the family will actually live); and safe-harbour packages (settled in advance, increasingly coordinated judge-to-judge through the International Hague Network of Judges, and endorsed as best practice by the HCCH's 2020 Guide on Article 13(1)(b) and the POAM domestic-violence framework).

The United States Supreme Court completed the doctrinal picture in Golan v. Saada (2022): where grave risk is established, a court may consider ameliorative measures but is not obliged to construct them — especially, Justice Sotomayor emphasized for a unanimous Court, where domestic violence is found; a judge need not engineer a return at any cost. Discretion, not mandate; protection, not formalism. Read together with DP, the working rule is now: courts assess the real, protected situation on return — and parties who make protection real change outcomes.

What This Shows About the Limits of the Hague Convention Alone

DP and Golan together show that the grave-risk clause is only as good as the machinery that surrounds it. The clause is right, and the ordinary-meaning reading is faithful to the text — but whether a child is actually protected on return depends on instruments the treaty does not itself provide: enforceable mirror orders, funded safe-harbour arrangements, judge-to-judge cooperation, and follow-up after the plane lands. Where that machinery exists, a genuine risk can be managed and a return made safe; where it does not, courts are left choosing between an unsafe return and a refusal the treaty hoped to avoid. The limit is not the exception's wording; it is the protective infrastructure behind it.

What Parents and Professionals Should Understand

For a left-behind parent, the strongest legal argument is often a changed fact: secure housing near needed services, funded support, mirror orders filed at home, and prosecution risks addressed before the hearing (the Neulinger boomerang, article #6). A protective package on the table can turn a grave-risk trial into a logistics discussion. For a taking parent with a genuine safety concern, specificity is credibility: documented risk (medical evidence, police records, service gaps) is weighed seriously, and courts increasingly are listening. For courts, the reunite findings are an indictment of promise-based practice — order what can be enforced, prefer mirror orders and follow-up, and remember that Golan permits refusing return where protection cannot be made real. None of this is legal advice; it is a map for a conversation with a qualified lawyer.

Limitations

This is a case study of two leading judgments and the surrounding research, not a comprehensive account of protective-measures law, which varies by jurisdiction. The disability and suicide-risk facts are reported only as the basis of the parties' genuine defenses, from the public record. Statistics are from the HCCH global study.

Conclusion

The special-needs and safety cases DP anticipated are no longer rare, and the 6.7-year-old at the centre of the average case increasingly arrives with a therapy schedule. The answer the system has is not a new exception but better protective-measures practice — undertakings backed by mirror orders, safe-harbour packages, judicial cooperation, and follow-up — funded and templated enough to be real. As DP showed and Golan confirmed: courts assess the real, protected situation on return, and the parties who make protection real are the ones who change what happens to the child.

Frequently Asked Questions

What is an "undertaking" in a Hague case? A promise — usually by the left-behind parent to the returning court — to provide protections on return, such as housing, support, or not pursuing prosecution. Research shows undertakings are often broken once the family is back and are hard to enforce across borders, which is why mirror orders are stronger.

What is a "mirror order"? An order entered by the child's home court that mirrors the protections promised in the returning court, so they are enforceable where the family will actually live — a sturdier alternative to a bare undertaking.

Does establishing "grave risk" mean the child won't be returned? Not automatically. Courts consider whether protective measures can make the return safe. After Golan v. Saada (2022), a court may consider such measures but is not required to engineer a return at any cost — especially where domestic violence is found.

Did these cases decide who the child should live with? No. A Hague case decides return to the child's home country; custody is decided there afterward. Protective measures are about making the return survivable, not about the final parenting outcome.