Extradition to China after Liu: is Spain the last holdout in Europe?

In Liu v. Poland of 6 October 2022, the European Court of Human Rights radically changed the terms of the debate on extradition to China.

The case concerned a person wanted for ordinary criminal proceedings, not a political dissident or a member of a persecuted minority. Yet the Court held that extradition to China would violate Article 3 ECHR because of the broader situation in Chinese detention facilities. Strasbourg found that the extent to which torture and ill-treatment were credibly and consistently reported in Chinese detention centres and prisons could be equated to a “general situation of violence” (§ 83). Once it was established that the requested person would be detained after extradition, he did not need to prove any further personal risk: “the applicant is relieved from showing specific personal grounds of fear, it being enough that it is established that, upon extradition, he will be placed in a detention centre or penitentiary” (§ 83)

That is the real novelty of Liu. The judgment does not simply say that extradition to China requires caution, which was already the standard approach at the domestic level. Rather, it says that, where detention in China is foreseeable, Article 3 is engaged at a systemic level and prevents extradition regardless of the existence of an individualized risk. For comments on Liu, see here at EJIL: Talk! and here at Giurisprudenza Penale.

Indeed, multiple reliable international sources (including the UN CAT, the US State Department and the EU) continue to describe systematic human-rights violations in China, particularly the widespread and tolerated use of torture, the use of coerced confessions as a basis for convictions, and the practical impossibility for independent bodies to verify actual detention conditions. In this regard, it is telling that, since the UN Special Rapporteur on Torture’s last country visit in 2005, during which it concluded that “torture remained widespread in China”, China has refused to extend any further invitation to UN experts, clearly showing its dissatisfaction with those findings and its continuing unwillingness to permit any international scrutiny.

Since Liu, a fairly clear European line has begun to emerge. Courts are no longer treating Chinese extradition requests as ordinary cooperation instances, but are asking whether the risk identified in Liu can be neutralized. And in most cases, the answer has been no.

The European line: no extradition where the risk is structural

One of the clearest post-Liu judgments is the Italian Court of Cassation’s decision no. 21125/2023.

The case concerned a Chinese citizen requested for prosecution for alleged illegal collection of public deposits. The Court of Appeal of Ancona had declared extradition admissible, relying on additional information and assurances provided by China. The Supreme Court annulled that decision without referral.

The Court of Cassation’s judgment is noteworthy because it stressed that the shortcomings identified by the Strasbourg Court in Liu had a systemic and generalized nature. From there, the conclusion was almost inevitable: generic assurances are not enough, especially where they merely reproduce domestic legal provisions on detainees’ rights. If the requesting State simply lists its domestic safeguards, but does not explain how they will operate in the concrete case and how compliance will be monitored, the Article 3 risk remains.

Since, Italian courts have constantly refused extradition to China (see, for instance, the Court of Appeal of Rome judgment of 20 April 2023).

Poland had already anticipated much of this reasoning before Liu.

In the Li Zhihui case, the Warsaw Court of Appeal refused extradition to China in March 2021. The requested person, a Swedish citizen of Chinese origin, was wanted for alleged fraud. The Court found extradition inadmissible under Article 604 § 1 points 5, 6 and 7 of the Polish Code of Criminal Procedure, read together with Article 3 ECHR. It relied on a combination of grounds: the risk of ill-treatment, the possible application of the death penalty, possible violations of fundamental rights, and the incompatibility of surrender with Polish and EU law.

What makes the Polish judgment particularly interesting is that it did not stop at prison conditions. The Court examined the functioning of the Chinese system more broadly, particularly whether China had an effective system of protection against torture, whether it cooperated with international monitoring bodies, whether life imprisonment could realistically be reduced, and whether the applicant’s links with Falun Gong could worsen his position after surrender.

Other European courts have moved in the same direction, even if through slightly different routes.

In Albania, in April 2023, the High Court overturned lower-court decisions that had authorised extradition to China in the Liangbin Chen case. The lower courts had approached the matter largely through the formal requirements of extradition: the existence of an arrest decision, double criminality, and the absence of domestic statutory bars. The High Court reversed course after the defence relied on Liu, Article 3 ECHR, and international reports on torture, arbitrary detention and fair-trial violations in China. The result was the rejection of the extradition request and the revocation of the extradition-related house arrest.

Cyprus has also joined the restrictive line. In the Ma Chao case, the District Court of Larnaca, with a judgment of December 2022, refused extradition to China on grounds that included Article 3 and Article 6 concerns. The importance of the Cypriot approach is that it treats torture and fair trial as connected problems. In a system where confessions are allegedly extracted through ill-treatment, the Article 3 risk cannot be isolated from the Article 6 risk. A trial based on coerced evidence may reach the high threshold of a flagrant denial of justice, especially where judicial independence and access to effective defence are also in question.

France had already shown similar caution before Liu. In 2021, the Bordeaux Court of Appeal gave an unfavourable decision on a Chinese extradition request. Its reasoning included political-purpose concerns, fair-trial issues, and the risk of treatment contrary to Article 3 ECHR.

Finally, the earlier Italian Cassation judgment no. 6769/2016 adds another strand to the same jurisprudence from the angle of grossly disproportionate punishments. Long before Liu, the Court had already been concerned by Chinese sentencing ranges for economic offences, especially where the maximum could reach life imprisonment and the criteria triggering the higher range were vague. That reasoning now sits naturally alongside the Sanchez-Sanchez (§ 89) and Hayes (§ 111) openings from the European Court on Human Rights on grossly disproportionate punishment as a potential bar to extradition.

Taken together, these cases do not all say the same thing. Some focus on Article 3, while others add the death penalty, irreducible life imprisonment, fair trial, political motivation, religious persecution, or gross disproportionality sentences. But the underlying logic is that extradition to China is refused because the risks are structural and cannot be dispelled through (generic) assurances.

The common thread: assurances are not enough if they cannot be verified

Assurances is where the European jurisprudence converges.

Under Othman (Abu Qatada) v. United Kingdom, assurances must be closely assessed in substance: their specificity, the authority giving them, their binding character, the requesting State’s record of compliance, and the existence of an effective monitoring mechanism. Later cases such as Yefimova v. Russia (§§ 202-203), Soldatenko v. Ukraine (§ 73), and, more recently, Kunshugarov v. Türkiye (§ 115), confirm the same idea: assurances are not self-validating.

A promise that fundamental rights will be respected – such as those submitted by Chinese authorities in the domestic extradition proceedings cited above – does not explain where the person will be detained, in what conditions, with what access to lawyers, with what protection against coercive interrogation, and with what independent mechanism to verify compliance after surrender. Likewise, a list of domestic provisions does not show that those provisions are effective and will be respected.

There is, however, an important caveat. The gravitational force of Liu appears to be largely confined to Europe, or at least to jurisdictions operating within the Convention framework. Outside the EU/ECHR space, the picture is different.

In parts of Asia – notably Thailand (here and here) and Cambodia (see here and here) – and more recently in South Africa, Chinese extradition requests continue to be treated primarily as ordinary law-enforcement cooperation, especially where the allegations concern fraud, gambling, cybercrime or scam-centre networks.

Morocco offers a different example: the extradition of a Uyghur man was initially approved, but later halted after an interim decision and, subsequently, a full decision on the merits from the UN CAT in August 2024 on non-refoulement grounds.

These cases suggest that, outside the ECHR system, Liu has little practical pull. The inquiry tends to be driven by treaty cooperation, reciprocity and transnational-crime enforcement, rather than by the systemic Article 3 analysis that is now shaping European case law.

Spain: the European exception

Spain stands apart.

Since Liu became final in January 2023, Spain has remained the only major European country to continue handing over persons wanted by China. As reported by El País, in 2023 and 2024, the Spanish Government reportedly authorised nine extradition proceedings to continue judicially and most of them ended up with a positive decision for extradition (see, inter alia, judgments of the Audiencia Nacional no. 165/2025 (súplica), no. 791/2025, no. 81/2024 (súplica), no. 504/2024, no. 44/2024 (súplica), and at least one person was surrendered to China in October 2024.

The Spanish position appears to rest on a narrower reading of Liu. The Audiencia Nacional and the Government do not treat the Strasbourg judgment as creating a practical bar to extradition while the general situation of violence in Chinese detention facilities persists. Instead, they continue to decide case by case and condition surrender on guarantees from China.

In practical terms, this turns Liu into a problem of assurance drafting. If China gives a note promising that the requested person will not be subjected to ill-treatment and that fundamental rights will be respected, then this is considered sufficient by the Spanish courts and surrender may be authorised.

That is precisely where Spain diverges from the emerging European line and, quite frankly, from the consolidated jurisprudence of the European Court of Human Rights.

For Italy, Poland, Cyprus, Albania and France, the key question is not whether China can promise compliance, but whether those promises can be trusted and verified in a system where independent monitoring is unavailable. Spain, instead, appears more willing to treat assurances as capable of neutralising the Article 3 risk, even where the underlying problem identified in Liu remains unchanged.

There is, however, a further problem in the Spanish approach: one of methodology which ultimately becomes one of substance.

The Audiencia Nacional first declares extradition admissible in principle after a review of the basic requirements, subject to China providing assurances within a fixed period of time. Only afterwards does China transmit the diplomatic guarantees. At that post-extradition stage, the requested person is allowed to challenge them only in writing, without a fresh oral hearing. The Audiencia Nacional then decides by way of a short order whether the condition has been satisfied. In most cases, it merely considers that it has, without any actual assessment of the substance and quality of the assurances.

This sequence is hardly reconcilable with the logic of Article 3 ECHR. Diplomatic assurances are not a mere administrative condition to be checked after the judicial decision has already been taken. They are the very material on which the Article 3 assessment depends and on which any decision to extradite should rest. If the risk of torture or ill-treatment is alleged, the court must examine the quality, specificity, reliability and verifiability of the assurances before deciding whether extradition is compatible with the Convention.

The Spanish method reverses that order. It front-loads the authorisation of extradition and back-loads the scrutiny of the guarantees. Once extradition has already been declared admissible, the later assessment of assurances risks becoming a formal verification exercise to merely check that guarantees were provided and that they correspond in general terms to what was requested. But that is not compatible with international human rights standards.

This procedural model also creates a subtle shift in the burden of persuasion. Instead of requiring the requesting State to dispel the Article 3 risk before extradition is authorised, it places the requested person in the position of having to undo a decision that has already gone against him. The court is no longer asking, with an open mind, whether extradition is safe. It is asking whether the later diplomatic note is enough to complete a process already set in motion.

Numerous judges of the Audiencia Nacional have expressed dissenting opinions precisely on this point. One dissenting view described Chinese guarantees as merely formal and insufficient to allow courts to conclude that the right not to be subjected to inhuman or degrading treatment would be protected. Another criticised the idea that the assessment of guarantees could effectively be left to the Government, calling it an unacceptable surrender of the judicial function to the political branch.

This is the real legal fault line.

Warning or wall?

The post-Liu landscape therefore turns on a simple question: is Liu a warning, or is it a wall?

Spain treats it as a warning, while the rest of the emerging European jurisprudence treats it as something much closer to a wall.

The second reading is more faithful to the structure of Article 3, at least as matters currently stand. The prohibition of torture and inhuman or degrading treatment is absolute. It does not permit a balancing exercise, and it does not allow courts to rely on diplomatic optimism where the risk is systemic and verification is absent.

Within Europe, Spain increasingly looks alone.

A counter move from Beijing?

A recent decision of the Court of Appeal of Brescia refusing extradition to China on the basis of the risk of the death penalty was overturned by the Italian Court of Cassation, with judgment no. 9195 of 9 February 2026, on procedural grounds, concerning the right of the requesting State to participate in the proceedings (for a comment on eXtradando, see here). The case was remitted for a fresh assessment.

The judgment is noteworthy not only in relation to the rights of the requesting State, but also for another reason: it appears that China appointed counsel to represent it in the Italian extradition proceedings. This is unusual and may signal a renewed interest by the Chinese authorities in trying to push back against the effects of Liu.

This renewed litigation strategy also fits within a broader diplomatic pattern. China has been steadily expanding and activating its extradition treaty network, not only by signing new agreements but also by ratifying older treaties and bringing pending instruments into force. Recent examples include the ratification of treaties with Armenia, Congo, Kenya and Uruguay at the end of 2022, with Zimbabwe in 2024, and with Serbia in 2025.