Eswatini has received two more people deported from the United States under the countries’ bilateral third-country arrangement, according to the Eswatini government and reporting by Reuters and the Associated Press on 27 and 28 August 2026. The two arrivals were described only as Latin American nationals. Their names, nationalities, legal status and length of stay were not disclosed. That uncertainty is the legal story. A state cannot make a transfer lawful merely by calling it a deportation, outsourcing reception to another government, or describing the destination as a diplomatic partner. From a French and European perspective, the decisive questions would be different: what is the legal instrument, has the person been lawfully admitted to the receiving country, what individual risk has been assessed, who controls any detention, and can the person obtain a lawyer and an effective remedy? This article does not suggest that the latest Eswatini arrivals have already been found to be unlawfully detained or exposed to a prohibited risk. It explains the test that would apply if France attempted a comparable transfer, or if a person transferred through such a chain sought protection before French or European institutions.
I. What do Eswatini’s new third-country deportations actually change?
A. Is a US deportation to Eswatini an extradition or an administrative transfer?
The first mistake in any fast-moving discussion of the Eswatini flights is to treat every cross-border removal as an extradition. Extradition is a request by one state for a person to be surrendered for criminal prosecution or the execution of a sentence. It normally identifies the requesting authority, the offence, the warrant or judgment, the applicable criminal law and the procedural guarantees attached to the surrender. Deportation, by contrast, is ordinarily an immigration measure: the sending state asserts that the person has no right to remain and arranges departure to a country that is legally able to receive that person. The fact that the person may have a criminal history does not erase the distinction.
The French Code of Criminal Procedure makes the distinction visible. Article 696 of that Code places extradition inside the title on international judicial assistance and states that, unless an international convention provides otherwise, the conditions, procedure and effects of extradition are governed by the extradition chapter. Article 696-8 requires a request to be supported by a judgment, a prosecutorial act, a warrant or an equivalent judicial instrument, with an indication of the facts and the relevant criminal law. The text is not a general licence for an executive authority to send an unwanted migrant to a third state. It is a judicial-cooperation framework.
Article 696-4 is equally instructive. In the version currently displayed by Légifrance, Article 696-4 of the Code of Criminal Procedure begins with the exact words L’extradition n’est pas accordée
. It then lists bars including political purpose, extinction of the prosecution and the possibility that the requested person would be tried by a tribunal that does not provide “garanties fondamentales de procédure et de protection des droits de la défense”. Those safeguards belong to a surrender requested for a criminal case. They cannot be imported mechanically into a deportation file, but neither can a deportation label be used to avoid the principles that protect a person from torture, arbitrary detention or an onward chain of removal.
The latest reporting points to an administrative arrangement rather than a conventional extradition request. Reuters reported on 27 August that Eswatini said it had accepted two further “third-country nationals” from the United States, that the government described them as Latin American and that more than 30 people had been sent there since the previous year. Reuters also reported that the United States agreement was valued at $5.1 million and had been challenged by lawyers and rights groups who said that some earlier arrivals were detained in Eswatini after serving criminal sentences in the United States. The Associated Press reported the next day that the two latest arrivals were the fifth batch since July 2025, while their identities and the length of their stay remained undisclosed. The Associated Press account is available here.
Those facts establish a live cross-border legal question, not a final finding of illegality. The public record does not disclose the two individuals’ immigration histories, whether they consented to any onward movement, whether Eswatini issued visas or residence documents, whether a court order governs their custody, or whether an onward return to their countries of nationality is planned. A sound legal analysis must preserve those limits. It is possible to condemn secrecy as a rule-of-law risk without asserting facts that have not been verified.
There is also a jurisdictional limit. French courts do not act as a general appellate tribunal for every American removal decision. A French judge would become directly relevant if the person were in France, if French authorities were asked to remove or transfer the person, if a French administrative decision fixed a country of destination, if a French consular or police action contributed to the transfer, or if a European legal instrument governed the movement. The French comparison is therefore not a claim that French law automatically invalidates the United States–Eswatini agreement. It is a stress test: the comparison identifies the documents, guarantees and facts that a French administration would have to confront.
The distinction matters for public authorities and for counsel. A request that uses the vocabulary of “deportation” but seeks a person for criminal proceedings may be an extradition in substance. A flight arranged after a completed sentence may still involve immigration custody, but the sending state must identify the legal basis for continued detention and the receiving state must explain the person’s legal status. A bilateral memorandum may allocate costs and logistics; it does not by itself decide whether the person may enter, whether the person may be held, or whether the person may be sent onwards. Those questions remain attached to the individual.
B. What do the latest arrivals show—and what remains unproven?
The news value of the latest arrivals lies in the combination of repetition and opacity. This is not the first flight. The AP account describes a fifth batch since July 2025. Reuters lists earlier nationalities including Vietnam, Jamaica, Cuba and Yemen. Earlier reporting also described a Jamaican national being repatriated and a Cambodian national being released for repatriation after five months in a maximum-security prison. The current two arrivals were described as Latin American, but neither wire report identifies them. The pattern suggests an operating programme, while the missing data prevents a conclusion about the legal position of these particular people.
For a lawyer examining the case, the first question is not whether the political description is persuasive. It is whether the transfer file can be reconstructed. The file should show the original removal order, the notice given to the individual, the country decision, the receiving government’s acceptance, any travel document or admission authorisation, the custody decision, the medical assessment, the contact details for counsel and the route actually taken. If the person faces a further removal from Eswatini, the file should show who is expected to receive the person next and on what terms. A chain of transfers cannot be assessed one airport at a time.
The second question is the nature of custody. “Arrived in Eswatini” says nothing by itself about whether a person was free, subject to immigration reporting, held in a reception facility, imprisoned under a court order or detained under an executive arrangement. The distinction is essential. Immigration detention may be lawful under a national system with judicial review and a genuine removal prospect; detention as a substitute for criminal punishment raises a different set of questions. A person who has finished a sentence in the sending state does not become legally invisible at the moment of disembarkation.
The third question is legal access. The reports describe earlier litigation about access to lawyers and detention conditions. That history does not prove that the two latest arrivals have been denied counsel. It does show why the issue cannot be reduced to the existence of a diplomatic agreement. If a person cannot communicate confidentially with a lawyer, cannot understand the basis of custody, cannot challenge a removal decision or cannot present a fear of persecution before an onward transfer, the receiving arrangement creates a serious procedural vulnerability.
The fourth question is onward refoulement. A person may be flown from the United States to Eswatini even though Eswatini is not the person’s country of nationality. If Eswatini then sends the person to that country, the legal risk does not disappear because the first aircraft landed elsewhere. The sending state may have to account for a foreseeable chain if it knew, or should have known, that the intermediate destination would make a further removal likely. The receiving state’s formal promise is relevant evidence, but not a complete answer where actual practice, detention, access to asylum and judicial independence point in another direction.
The fifth question is individualisation. A country’s human-rights record does not automatically decide every person’s case, and a government’s assurance does not automatically cure every risk. The analysis should examine the person’s nationality, ethnicity, religion, political activity, prior detention, health, family links, prior asylum claim, criminal conviction, ability to obtain documents and exposure to onward removal. It should also ask whether the person’s procedural position has changed since the original removal decision. A risk assessment made before transfer cannot be treated as permanently valid if new detention facts or a new destination have emerged.
This is where the latest story has a direct legal and media consequence. The unknown identities are not a minor omission. They prevent independent testing of the risk. The absence of a public agreement prevents testing the receiving state’s powers. The absence of a custody order prevents testing the legal basis of detention. The absence of a lawyer prevents testing the reliability of consent or of any statement that the person accepts repatriation. Secrecy does not prove a violation, but it makes the central claims impossible to audit.
Reuters’ description of the $5.1 million arrangement also explains why courts would look beyond formal wording. A payment does not prove an unlawful bargain; states can reimburse reception, transport or administrative services. It does, however, make the institutional incentives relevant. If a receiving state is paid per person or under a capped programme, a court may ask whether the arrangement creates pressure to maintain custody, whether the person can leave freely, whether the state has a real legal interest in the person’s presence and whether the public authority has disclosed the conditions under which the transfer occurs. The legal test remains fact-specific, but the contract and its implementation become evidence.
Finally, the Eswatini story should not be confused with an American practical guide. The useful question for an international audience is how a European legal system would classify and control an analogous transfer. That is the gap between news reporting and legal analysis. The French answer begins with a separate destination decision, an individualised risk assessment and an effective remedy; it does not begin with the political convenience of finding a third country willing to receive a person.
II. What would French law test before a comparable transfer?
A. How do non-refoulement and the destination-country decision work?
French law separates the decision to remove a person from the decision identifying the country to which the removal may be carried out. That separation is the first practical safeguard against a “legal shortcut”. The Légifrance section containing Articles L. 721-3 to L. 721-5 of the Code governing entry and residence of foreigners and asylum states that the administration fixes, through a decision distinct from the removal decision, the country to which the person may be sent. Article L. 721-4 then provides three routes: the country of nationality subject to asylum-related limits; another country for which a valid travel document has been issued under a European or bilateral readmission arrangement; or, with the person’s agreement, another country in which the person is legally admissible.
The relevant statutory language is precise. Article L. 721-4 says that the administration may designate the country of nationality, a country covered by a valid document issued under a readmission arrangement, or, with the person’s agreement, another country where the person is legally admissible. It then states: Un étranger ne peut être éloigné à destination d’un pays
if the person establishes that life or liberty would be threatened there or that the person would face treatment contrary to Article 3 of the European Convention on Human Rights. The complete provision is available on Légifrance, Article L. 721-4 CESEDA.
This framework makes the Eswatini comparison concrete. A third country is not automatically an acceptable destination because it has agreed at government level to receive non-nationals. The administration would need to explain which statutory route applies. Is there a valid travel document under a readmission agreement? Has the person consented to the destination? Is the person legally admissible under the receiving country’s law? Is the person still seeking asylum or protected against removal? A diplomatic understanding may support an answer, but the administrative decision must identify the legal and factual bridge between the individual and the country.
Article L. 721-5 is equally important because it confirms that the destination decision is challengeable. The provision states, in exact terms, that La décision fixant le pays de renvoi peut être contestée
according to the procedure applicable to the removal measure it implements. It also addresses the situation in which the destination decision is notified later, including after the removal decision has become final. That provision is available at Légifrance, Article L. 721-5 CESEDA. The strategic consequence is simple: a person must not focus only on the order to leave. The country decision may contain the decisive defect.
Non-refoulement is not a general policy preference. Article 3 of the European Convention on Human Rights provides that no one may be subjected to torture or to inhuman or degrading treatment. The European Court of Human Rights treats that protection as absolute. The Convention text is available from the European Court of Human Rights. Article 3 applies to removal where substantial grounds show a real risk in the destination. It can also become relevant where detention conditions, prison treatment or the foreseeable next leg of the journey expose the person to a prohibited harm.
The same principle appears across the international framework. Article 3 of the Convention against Torture prohibits transfer where there are substantial grounds for believing that a person would be in danger of torture. Article 33 of the Refugee Convention protects refugees against return to territories where life or freedom would be threatened. Within EU law, Article 19(2) of the Charter of Fundamental Rights states that no one may be removed, expelled or extradited to a state where there is a serious risk of the death penalty, torture or inhuman or degrading treatment or punishment.
French administrative case law supplies the individualised method. In Conseil d’État, 7th and 2nd Chambers sitting together, 5 June 2025, no. 493675, the Court reviewed decisions fixing a country of return and referred to an exposure that must be “directe, actuelle et personnelle”. The exact wording appears in the decision’s reasoning. The Conseil d’État rejected the claim on the evidence before it, but the decision does not turn non-refoulement into a country-name checklist. It requires the court to ask what this person would face in this country on the evidence actually produced.
The evidence for a third-country transfer must therefore go beyond a generic statement that the receiving government will comply with human rights. Counsel would request the text of the agreement, the legal basis for custody, the person’s admission document, the applicable prison or immigration rules, the route for an asylum claim, the contact details for the independent monitoring body and the written conditions for any onward movement. If the receiving state offers diplomatic assurances, the court would have to examine their specificity, enforceability, monitoring and compatibility with actual conditions. An assurance that says “rights will be respected” is weaker than a document identifying release conditions, judicial access, counsel contact and a prohibition on onward removal without an individual process.
A foreseeable chain matters even where the immediate country appears safer. Suppose a person is sent to a state that does not prosecute the individual but is expected to return the individual to a country where the person faces torture or persecution. The French administration would have to address that likely next step. The absence of a direct flight to the feared country would not by itself answer the risk. Conversely, speculation about an entirely hypothetical future transfer would not be enough. The lawyer must show a real prospect through the agreement, public statements, prior flights, detention records or the person’s own documented circumstances.
The rule also protects against a different form of evasion: transferring a person into a system where the formal status is unclear and the person cannot reach a decision-maker. The ability to obtain asylum, counsel and a judicial review is part of the practical risk assessment. An intermediate state may be geographically distant from the person’s home country but still unsafe if it confines the person, blocks communication and treats the person as a security object rather than as an individual with a legal status.
French law also requires a realistic removal prospect before a person is placed under an alternative control measure. Article L. 731-1 CESEDA allows an administrative residence assignment where the person cannot leave immediately but removal “demeure une perspective raisonnable”. The quoted phrase is the statute’s own wording. The condition is relevant by analogy to a third-country arrangement: if no state has actually accepted the person, if the person cannot be legally admitted, or if the receiving state has not supplied the required documents, the administration cannot turn a theoretical route into an indefinite legal status.
The same analysis distinguishes political responsibility from legal responsibility. France would not be required to decide whether the United States acted lawfully in the Eswatini programme before applying its own law. If France received a person from the United States and then considered removal to Eswatini, the French decision-maker would have to make a fresh assessment under French and European standards. A previous American finding that removal was permissible would be relevant background, not a substitute for the French destination decision. That is the practical meaning of territorial legal responsibility.
For a related explanation of the French framework for international alerts and surrender, readers can consult the firm’s overview of Interpol Red Notices, arrest and extradition remedies in France. The comparison is useful precisely because an administrative third-country transfer must not be mistaken for a criminal surrender procedure.
B. What remedies and evidence can stop an unlawful removal?
The remedy depends on the measure and on where the person is held. A person facing a French removal order should identify at least four separate targets: the underlying removal decision, the decision fixing the country of destination, any refusal of a voluntary departure period or entry ban, and any placement or continuation of administrative detention. A single urgent filing may need to explain how the measures interact, but the legal grounds and competent judges are not identical.
For the destination and removal decision, the administrative court is the central forum. Article L. 721-5 CESEDA links the challenge to the procedure applicable to the removal measure. The applicant should state clearly that the proposed destination is not the country of nationality, identify the absence or weakness of consent and admissibility, and explain the individual Article 3 risk. If the administration relies on a bilateral arrangement, the applicant should request the instrument and challenge any refusal to disclose the factual basis of the decision. The court can assess whether the administration had enough evidence at the date of the decision and whether the reasoning addressed the actual destination rather than a generic country profile.
Urgency may require interim relief. Under Article L. 521-1 of the Code of Administrative Justice, the urgent applications judge may suspend the execution of an administrative decision when urgency is established and a ground creates serious doubt about legality. Under Article L. 521-2, the judge may order measures necessary to safeguard a fundamental freedom where a public authority has caused a serious and manifestly unlawful interference. The official text says: Le juge des référés se prononce dans un délai de quarante-huit heures
. A person about to be placed on a flight must present a concrete timetable, the flight or removal notice, the destination evidence and the risk material, not merely a general objection to outsourcing.
Detention generates a parallel judicial track. Article L. 741-1 CESEDA authorises an initial period of 96 hours where the statutory conditions are met, including a risk of evasion and the insufficiency of another measure. The text requires that aucune autre mesure n’apparaît suffisante
. That wording matters in a third-country case because the administration must explain why a controlled residence measure cannot secure removal and why a real prospect of departure exists.
After the initial period, Article L. 742-1 CESEDA states that Le maintien en rétention au-delà de quatre-vingt-seize heures
may be authorised by the judicial magistrate of the tribunal judiciaire on the administration’s application. The judicial judge examines the conditions for continued detention and the administration’s diligence within that jurisdiction. If the receiving state has not issued an admission document, if the planned route has collapsed, or if the administration is taking no useful step, those facts can become decisive. They do not necessarily allow that judge to rewrite the administrative destination decision, but they can affect whether continued detention remains justified.
The person can challenge the placement itself. Article L. 741-10 CESEDA provides that a detained foreign national may contest the placement before the judicial magistrate within 96 hours of notification. The statutory phrase is dans un délai de quatre-vingt-seize heures à compter de sa notification
. Article L. 743-3 CESEDA identifies the procedure for applications challenging placement and requesting extension or release. These are short windows. The person, family and counsel should preserve the notification time, the language used, the interpreter’s identity and every document given at the place of detention.
Access to communication is not a decorative guarantee. Article L. 744-4 CESEDA requires information, as soon as possible, about the right to an interpreter, counsel and a doctor, as well as communication with the consulate and a person of the detainee’s choice. The law says those details must be communicated “dans une langue qu’il comprend”. Article L. 744-6 CESEDA adds that, on arrival at the detention centre, the person receives notification of rights that may be exercised in an asylum claim, with legal and linguistic assistance. In an Eswatini-type scenario, the equivalent questions would be: was the person told why the receiving state held them, in a language they understood, and could the person reach counsel before an onward flight?
The division between judges is not a technicality. In First Civil Chamber of the Court of Cassation, 5 December 2018, no. 17-30.978, the Court held that the judicial judge cannot decide the legality or opportunity of the administrative choice of destination. The decision describes the assessment of “la possibilité ou de l’opportunité du renvoi” as outside the judicial judge’s powers in that detention litigation. The administrative court reviews the destination decision; the judicial judge reviews the detention framework and the administration’s diligence. A lawyer must therefore place the same factual story in the correct procedural channels rather than wait for one judge to supply every remedy.
The evidence package should be assembled for both tracks. It should include the removal order and country decision; the receiving state’s written acceptance; the travel document, visa or admission permit; proof of the flight and deadline; any notice of detention; evidence of prior detention or ill-treatment; medical and psychological records; family and residence links; asylum documents; certified translations; and a witness statement explaining what the person was told. Where onward removal is feared, the package should add evidence of the receiving country’s practice, the likely next destination, prior flights involving similarly situated people and any public or confidential assurance concerning release and legal access.
Evidence must also address credibility without turning the application into a political manifesto. A court will ask whether the risk is personal, current and supported. A report about a country’s general record may be important, but it should be connected to the individual: nationality, political profile, language, prior accusations, prison placement, health or family relationship. If the person has served a sentence, counsel should obtain the judgment and release record. If the receiving state claims the person remains wanted, counsel should request the warrant and test whether the claim is criminal, administrative or merely diplomatic.
One recurring danger is the substitution of consent for a real choice. Consent to leave the United States does not necessarily mean consent to be detained in Eswatini or to be sent from Eswatini to a country where the person faces persecution. A person who signs a document in custody, without an interpreter or access to counsel, may dispute the meaning of that signature. The legal question is not simply whether a form exists. It is whether the person understood the destination, the custody consequences, the option to challenge the transfer and the risk of any onward movement.
A second danger is the substitution of a payment arrangement for admissibility. French Article L. 721-4 speaks of legal admissibility, not of the receiving state’s commercial interest. A state may be willing to receive people while lacking a clear immigration status for them. That gap can produce airport limbo, custodial confinement or an onward transfer with no identifiable remedy. In a French file, the administration would need to show the actual legal status to be granted on arrival, the duration of that status, the person’s ability to leave the facility and the authority competent to review any deprivation of liberty.
A third danger is the substitution of assurances for verification. Diplomatic assurances can matter, especially where a state has a functioning judicial system and a monitoring mechanism. They should be precise enough to be tested: no onward removal without an individual decision; access to a named lawyer; confidential communication; independent medical care; notice of any immigration hearing; and a channel for urgent intervention. If prior cases show that people were held for months, unable to meet counsel or unable to challenge the basis of custody, the administration must explain why the new assurances are credible and how compliance will be monitored.
The French answer would therefore be neither “every third-country transfer is unlawful” nor “a bilateral agreement settles the matter”. It would be a staged inquiry. First, identify the measure and the legal basis. Second, establish that the destination is legally available for this individual. Third, test Article 3 and the international non-refoulement rules against the person’s actual circumstances, including the foreseeable chain. Fourth, secure a timely and effective remedy. Fifth, review detention separately and require a genuine removal prospect. That method is portable across borders and explains why the Eswatini news is relevant beyond the United States or southern Africa.
For international media, the point is also a warning about terminology. “Third-country deportation” sounds like a logistical category, but it can conceal four different legal events: removal from the sending state, admission to the intermediate state, deprivation of liberty there and onward removal to a final state. Each event has its own authority, document and remedy. Reporting should identify all four. Legal counsel should request all four. A court should not allow the gaps between them to become a zone without responsibility.
Conclusion
Eswatini’s receipt of two additional people deported from the United States is a hot cross-border legal story because it tests whether states can outsource the hardest part of removal—the person’s status, custody and protection—to an intermediate country. The latest reports confirm the arrival of two Latin American nationals, but they do not disclose enough to decide whether those individuals have been detained unlawfully or face a prohibited risk. That restraint is part of rigorous legal reporting.
French law supplies a demanding comparator. A removal decision and a country-of-destination decision are distinct. A third country must be legally available for the person. Article 3 of the European Convention and the non-refoulement rules remain engaged even when a flight stops somewhere other than the country feared. Detention requires a real removal prospect, communication rights and judicial control. The administrative judge and the judicial judge have different roles, so an effective challenge must be filed on both tracks when the facts require it.
The practical question for the next Eswatini flight is therefore not only where the aircraft lands. It is what legal status the passenger receives on arrival, who may hold the passenger, which lawyer can speak confidentially with the passenger, whether an onward removal is planned and which court can stop it in time. Until those questions are answered with documents rather than assurances, the programme remains a serious rule-of-law test for every state involved.