EU Harmonisation of Norms Regulating Detention: Is EU Competence (Art. 82(2)b TFEU) Fit for Purpose?

The EU has acquired the competence to harmonise individual rights in the field of criminal procedure (Art. 82(2)b TFEU). This was hailed as a positive development helping redress the unbalance towards a too security-oriented development of the Area of Freedom Security and Justice. This article discusses the breath of this competence and designs an analytical framework illustrating what requirements need taking into account to legitimate EU regulatory action in the field of detention. It argues that the wording of the provision, especially its utilitarian framing, strongly limits its potential for the EU to act in at least two areas, compensation for unjust detention and material detention conditions.


Introduction and Research Objectives
EU competence in the field of criminal justice is one of the areas which has most expanded across the years. It is among the most telling examples of policy areas which have gone from an intergovernmental governance structure to an almost fully supranational one (Fletcher et al., 2008, 58). But it has also expanded from a substantive point of view. The Maastricht version of the Treaty included competence to adopt cooperation instruments (Art. K.1.7 and K.1.9 TEU consolidated in Maastricht), while the Amsterdam and, most recently, the Lisbon versions of the Treaty, respectively, introduced a competence to harmonise definition of crimes (Article 31(1)(e) TEU consolidated in Amsterdam) and a competence to harmonise rules of criminal procedure (Art. 82(2) TFEU). This paper focuses on the novel Lisbon competence to set minimum standards in the field of rights of individuals in criminal procedure enshrined at Art. 82(2)b TFEU.
The introduction of this competence to harmonise procedural rights was hailed as a very positive development. It was hoped it would help to rebalance the focus within the Area of Freedom Security and Justice (AFSJ) which had long been criticized for being too security oriented (Mitsilegas, 2016). Nonetheless, this paper argues that there are few shortcomings in how this legal basis was drafted, thus making it more challenging for it to deliver the promise of addressing the AFSJ imbalances. In particular, the paper focuses on the shortcoming of the provision to function as a legal basis for the harmonisation of rules regulating detention.
The possibility, and desirability, for the EU to harmonise norms regulating detention, especially pre-trial detention, has often been discussed by the EU Institutions (Council 2009, European Commission 2011Parliament 2011) and by scholars (Baker et al., 2020, Martufi andPeristeridou, 2020). Advocate General Pitruzella also called for EU action in this field (Opinion of AG Pitruzzella., 2019) . Yet, this paper argues that it is questionable whether Art. 82(2)b can serve as a legal basis to regulate all relevant aspects of detention. The paper illustrates this point by focussing on two areas, where case law of the court of justice and national case law show the need of EU action, but where the possibility to adopt harmonisation measures on the basis of Art. 82(2)b TFEU can be debated. These two areas are material detention conditions and compensation for unjust detention. The aim of the paper is not to plead against harmonisation of EU action in these areas, but rather to show what constitutional challenges this poses.

Research Methods and Structure of the Article
The paper was commissioned to contribute to the special issue "An evidence-based approach to Pre-Trial detention" of the European Journal on Criminal Policy and Research to provide an analytical framework on how to assess when EU action regulating detention is constitutionally legitimate. It therefore does not include an empirical case study as such but provides a framework on how to interpret the relevant legal provision. In terms of specific methods, the research has been carried out combining desk research on scholarly work, and primary sources (legal provisions and policy reports), with elite interviews with practitioners and EU officials. The article is structured as follows. The "Art. 82(2)B TFEU: Three Requirements for Harmonisation of Detention Rules" section provides a detailed discussion of the text of Art. 82(2)b TFEU and of the requirements it sets for justifying EU legislative action on its basis. The following one which is the "The Limit of Art. 82(2)B TFEU: Compensation for Unjust Detention and Detention Conditions" section illustrates the limits of this provision in acting as a legal basis for harmonisation of rules on compensation for unjust detention, and material detention conditions.

Art. 82(2)B TFEU: Three Requirements for Harmonisation of Detention Rules
The text of Article 82(2) TFEU states that: To the extent necessary to facilitate mutual recognition of judgments and judicial decisions and police and judicial cooperation in criminal matters having a cross-border dimension, the European Parliament and the Council may, by means of directives adopted in accordance with the ordinary legislative procedure, establish minimum rules. Such rules shall take into account the differences between the legal traditions and systems of the Member States.
The provision identifies at least 3 specific requirements a legislative proposal must meet to be legitimately grounded on this legal basis (Coventry, 2017). First harmonisation of legislation on individual rights is only justified to the extent that is necessary to facilitate mutual recognition (hereinafter the "mutual recognition" requirement). Secondly, it must be confined to individual rights in the field of criminal procedure (hereinafter the "criminal procedure" requirement). Thirdly, it must concern scenarios having a cross-border dimension (hereinafter the "cross-borderness requirement"). The next three sub-sections provide an in depth discussion of the meaning of each requirement offering both restrictive and extensive interpretations of each criterion. Before discussing the meaning of requirements, it should be specified that the provision speaks specifically of harmonisation of individual rights. Therefore, when thereafter the expression harmonisation of rules on detention is used, reference is naturally made to those rules that regulate detention which contain individuals' rights.

The Mutual Recognition Requirement
To fully understand the first "mutual recognition" requirement, namely the need for harmonisation of individual rights in criminal procedures to be linked to mutual recognition, this sub-section starts by offering a broader contextualisation of the relation between harmonisation and mutual recognition.
From a theoretical point of view, one can distinguish between a deontological rationale for harmonisation of national norms and a utilitarian one (Wieczorek, 2020, pp. 76-93). In the first case-deontological rationale-the justification for the EU to constrain EU Member States' regulatory autonomy and set EU-wide regulatory standards is of a normative nature. That is, the EU steps in with the aim of ensuring that particularly desirable standards, which have a value in themselves, are adopted by all Member States. In the second case, utilitarian rationale, harmonisation is justified in light of a secondary policy goal to which the establishment of a regulatory level-playing field is instrumental. This ultimate policy goal can be the smooth functioning of the internal market, or of judicial cooperation in civil or criminal matters (Wieczorek, 2020, pp. 76-78).
The relation between harmonisation of definition of crimes and judicial cooperation has been the subject of a long debate. In particular, it is still discussed whether harmonisation definition of crimes on the basis Art. 83(1) TFEU should be justified on deontological or utilitarian rationales, the latter being its capacity to support cooperation (Wieczorek, 2020, pp. 106-119, Arroyo Zapatero & Munoz de Morales Romero, 2012, Weyembergh, 2013. Conversely, Art. 82(2)b TFEU undeniably sets a utilitarian function for harmonisation of criminal procedure as a requirement for its constitutional legality. Reliance on this legal basis is allowed only if it can be demonstrated that the resulting harmonising norms can contribute to the smooth functioning of mutual recognition instruments. Examples of such mutual recognition instruments can be the 2002 European Arrest Warrant Framework Decision (hereinafter EAW) or the 2014 European Investigation Order Directive. To understand whether harmonisation is justified on Art. 82(2)b TFEU is therefore necessary to unpack when can be said that harmonisation "facilitates mutual recognition".
Without making justice to the complexity of the principle, mutual recognition can be defined as requiring that Member States judicial authorities execute foreign decisions as their own but for few exceptional circumstances. The principle was introduced as the cornerstone of judicial cooperation, and its functioning was grounded on the assumption that all Member States could trust each of them would respect a basic level of fundamental rights (Willems, 2021;Xanthopoulou, 2020). Foreign decisions could therefore be considered normatively equivalent. Such basic level of fundamental rights was found in the 1950 European Convention on the protection of Human Rights and Fundamental Freedom (ECHR) and the 2012 Charter of Fundamental Rights of the European Union (EU Charter). Mutual recognition is therefore inherently connected to the twin principle of mutual trust.
The Court has been adamant that, as long as the standards in Member States live up to those of the ECHR and the EU Charter, Member States should trust one another (Stefano Melloni vs Ministerio Fiscal, 2013) . Therefore, as long as there is mutual trust, Member States authorities must comply with mutual recognition requests, for instance EAW requests and surrender suspects and convicted persons. They cannot halt surrender procedures on the basis that the state requesting the surrender has lower fundamental rights standard, if compared to domestic ones (Soo, 2020).
A different scenario is that in which Member States fundamental rights protection standards fall significantly below those set by the ECHR. The question of how to deal with surrender requests coming from states which have below ECHR fundamental rights standards was dealt by in Aranyosi and Caldararu (Pál Aranyosi and Robert Căldăraru v Generalstaatsanwaltschaft Bremen 2016). The case dealt with potential violations of Art. 3 ECHR (Art. 4 EU Charter). The Court established that if surrender towards a country where the fundamental rights of the requested person would be seriously at risk, mutual trust can be temporarily suspended. It then set down a number of steps Member States' authorities should take to verify the likelihood of such violations of fundamental rights in the individual case, and to decide whether surrender can be authorised or not. A subsequent case, LM (Minister for Justice & Equality v LM, 2018), extended this case law also to fair trial rights.
This interpretation of mutual recognition by the Court and of its relationship with mutual trust is relevant to the question of when harmonisation is legitimate on the basis of Art. 82(2)b TFEU. Based on what has been said so far, two interpretations of the "link with mutual recognition" requirement are possible.
A first, broad one, would go as follows. Harmonisation of individual rights based on Art. 82(2) b is justified when it generally contributes to creating a climate of mutual trust. Bringing national legislations closer can be assumed to be instrumental to such goal of mutual trust. More specifically raising the level of fundamental rights would hopefully create more trust. This interpretation of Art. 82(2)b, although plausible, has nonetheless been criticised for not complying with the Court of Justice established case law that the choice of legal basis must be grounded on objective factors which are amenable to judicial review (Commission v Council of the European Union, 2004) . Simply saying that harmonisation must generically foster mutual trust for it to be legitimately adopted on the basis of Art. 82(2)b TFEU was not considered a sufficiently objective factor for the choice of the said legal basis (Mitsilegas, 2016, 157). It should be noted, however, that in a recent case, the Court has acknowledged that harmonisation measures adopted on the basis of Art. 82(2)b TFEU are legitimate if they contribute to the creation of mutual trust (Criminal proceedings against Gianluca Moro 2019), argument picked up again by AG Pikamae in a subsequent case (Criminal proceedings against DR and TS, AG Opinion, 2021). 1 The case however concerned the interpretation of the "cross-borderness" requirement and is therefore further discussed below. It is unclear whether the Court considers it sufficient to say that harmonisation will increase mutual trust to say that the "mutual recognition" requirement is fulfilled.
A second, narrower, interpretation of the Art. 82(2)b TFEU "mutual recognition" requirement would not focus just on the capacity of harmonisation of individual rights to foster a climate of mutual trust. It would look at whether harmonisation can in practice eliminate concrete barriers to mutual recognition. One can think of a three-step enquiry (Coventry, 2017p. 56, Lööf, 2006. Firstly, one should evaluate whether there is empirical evidence that differences between national law and practice are a hindrance to mutual recognition, in that they cause delays, ill-execution and non-execution of requests (Sellier and Weyembergh, 2018) . Secondly, one should evaluate whether such hindrances are "legitimate" ones. If Member States refuse to comply with EAW requests because the issuing states' legislation differs from the domestic one and is below ECHR (and EU Charter) standards, such halting of surrender is legitimate. The Aranyosi and Caldararu case law, discussed above, allows to suspend surrender in these cases. In other words, the differences in national standards can be considered a "legitimate hindrance" to mutual recognition. Conversely, if Member States decide to suspend surrender because the domestic standards in requesting states are different from national ones, but still ECHR (or EU Charter) compliant, such a halting of surrender would not be necessarily legitimate. A faithful implementation of the Melloni case law discussed above would dictate that mutual recognition requests cannot be denied on the basis that legislation in the requesting state have a different standard, as long as they respect the EU Charter standards. In this context, the difference between Member States legislation cannot be considered a hindrance to mutual recognition. It is Member States' failure to comply with EU law obligations on how to navigate such differences that prevents effective cooperation.
In the event that differences in legislation and practice are a legitimate hindrance to mutual recognition, then the third step is to verify if EU regulatory intervention by way of harmonisation would be the best way to remedy the situation. If harmonisation is indeed the solution, in that it would make mutual recognition in practice easier, then harmonisation of individual rights on the basis of Art. 82(2)b TFEU is legitimate.
Applying this three-step test to our discussion on harmonisation of detention rules, the conclusions would be as follows: if differences exist across Member States as to the rules regulating detention; if such differences create legitimate hindrances to mutual recognition (i.e. certain Member States would not surrender because rules regulating detention are not drafted or implemented in compliance with the ECHR); and if EU harmonisation would remedy the situation, then harmonisation of detention rules on the basis of Art. 82(2)b would be legitimate.

The Criminal Procedure Requirement
The second requirement set by Art. 82(2)b is that harmonisation must be confined to individual rights applying to the phase of criminal procedure. For the purposes of our analysis, one should consider what rules on detention fall within criminal procedure. One should first distinguish between the criminal trial phase which culminates with the imposition of a sentence, and the phase of the sentence enforcement. Detention taking place during the trial is referred to as pre-trial detention and normally serves various purposes such as prevention of flight, preventing of tampering with evidence, and prevention of commission of further crimes. Detention occurring after the sentence has been imposed and is normally referred to as post-trial detention, it embodies the enforcement of a sentence, and it has a punishment rationale. EAW surrender requests can be issued both for the purposes of prosecution, that is during the duration of a trial, and for the purposes of sentence execution (Art. 1 EAW). Detention conditions in both contexts are therefore relevant for the functioning of mutual recognition. In other terms, differences in how member states regulate both pre-trial and post-trial detention can cause hindrances to mutual recognition which creates a prima facie case for Art. 82(2)b harmonisation. However, does the EU have the competence to harmonise rules on both pre-trial and post-trial detention? That is can the expression "criminal procedure" be considered applying both to the trial phase and also to the phase of sentence execution?
The European Council (2010) and the Parliament (European Parliament 2017, p. 92) arguably leave the door open to interpreting Art. 82(2)b TFEU as applying to both (Soo, 2020, p. 339). However, two of the procedural rights directives, the 2012 Directive on the right to information and the 2010 Directive on right to interpretation and translation, both adopted on the basis of Art. 82(2)b TFEU do not apply to the sentence execution phase. Moreover, during the negotiations on the 2016 Directive on the rights of children in criminal proceedings, the question of whether the norms on detention therein included should apply also to the sentence execution phase was raised. This was however strongly opposed by Member States (Soo, 2020, 334).
Assuming that only rules applying to the phase of pre-trial detention can be harmonised on the basis of Art. 82(2)b TFEU, a second question is what qualifies as pre-trial detention. The Commission and the Council of Europe speak of "detention up until a final conviction is imposed", including periods the person spends in detention waiting for an appeal (European Commission 2011, Council of Europe 2006). This position would be consistent with the interpretation that procedural rights directives apply until the final determination of guilt (Mitsilegas, 2016, p. 171). However, there does not seem to be a consensus among Member States on the temporal definition of pre-trial detention (Martufi and Peristeridou, 2020). Moreover, the Strasbourg case law defines pre-trial detention, for the purposes of applying the guarantees of Art. 5(3) ECHR, as only detention up the first instance conviction (Wemhoff v. Germany, 1968, Labita v. Italy, 2000 .Incidentally, this is also the interpretation of the UN documents (UN Report on Congress on the United Nations, Eighth United Nations Congress on the Prevention of Crime and the Treatment of Offenders, 1990) . Naturally, the EU can develop an autonomous concept of pre-trial detention. However, the Commission position must find support in the Council, thus within the government representatives of the Member States. Moreover, coherence with the Strasbourg Court would be important, especially in light of future plans of accession (Coventry, 2017, 46). 2 Finally, a third question is whether material detention conditions, like the ones at stake in Aranyosi and Caldararu such as "housing and services," can be considered part of criminal procedure. These are technically not rules of procedure. Coventry indeed argues against it, considering aspects such as length of detention, or grounds for detention as more pertinent (Coventry, 2017, 47).

The Cross-borderness Requirement
If the "criminal procedure" requirement delimits the "temporal" scope of application of legislation harmonised on the basis of Art. 82(2)b TFEU, the cross-borderness requirement limits the type of disputes rules harmonised on the basis of Art. 82(2)b can apply to. This requirement does not appear, at first sight, particularly controversial. The most obvious interpretation of a proceeding having a cross-border dimension is that of a proceeding with a "foreign" element. Such foreign element can be the nationality of the suspect, or of the victim, which are not those of the state where the proceedings are taking place. But it can also be a piece of evidence which is located abroad (European Commission 2008, p. 8). The EU would only be able to adopt legislation on the basis of Art. 82(2) TFEU which applies to this type of proceeding. The challenges lie more into the implication of enforcing this requirement in practice. Firstly, it has been observed that the cross-border character of certain proceedings might not be known at the start of the proceedings. It might surface at a later stage when investigation or questioning takes place (Peers, 2011, 670).
If EU harmonised rules must only apply from the moment that proceedings become cross-border, then this can create problems from a legal certainty perspective. Moreover, a rigorous application of the cross-border requirement raises the risk of reverse discrimination. That is individuals involved in cross-border proceedings will benefit from individual rights, while individuals involved in purely domestic proceedings would not. This issue has been raised in particular for what concerns individual rights of people in detention and is therefore relevant here (European Commission, 2011;Coventry, 2017, 53). 3 It would appear however that legislative practice and recent case law of the Court of Justice has rendered these discussions on the meaning and the implementation of the crossborder requirement moot. Indeed, the directives adopted so far on the basis of Art. 82(2) b TFEU do not explicitly limit their scope of application to cross-border cases. 4 The Court upheld this choice in a case concerning the 2012 Directive on the right to information in criminal proceedings which was also mentioned in the previous section. The Court's decision is grounded on the fact that application of harmonised standards on matters of criminal procedure can contribute to create mutual trust (para. 34 of the case). It would appear that following this case law of the Court, the question on how to interpret Art. 82(2) b TFEU cross-borderness requirement has lost its practical relevance. Therefore, it will not be further discussed.

The Limit of Art. 82(2)B TFEU: Compensation for Unjust Detention and Detention Conditions
Having clarified the interpretation of the various requirements listed by Art. 82(2)b TFEU, this section turns to the core of the paper's discussion. It assesses the potential of Art. 82(2)b TFEU to function as a legal basis to harmonise various aspects of detention rules.
Regulation of detention is a wide policy area which ranges from norms regulating the material conditions in prison (e.g. cells' square metres, access to different types of activities, visits) to norms regulating the grounds for pre-trial detention, the permitted length of pre-trial detention and the duty to review, or norms on compensation for unjust pretrial or post-trial detention. Admittedly, some regulation of pre-trial detention can be found in existing instruments, as for instance the EAW (Art. 4(6), Art. 5(3), and Art. 12), 2016 Directive on right to child in criminal proceedings (Art. 4 and Art. 6, among others). Instruments providing alternative to detention also exist, such as the 2002 European Supervision Order Framework Decision. And rules relevant to post-trial sentence execution are included in the text of the 2008 Transfers of the execution of sentences Framework Decision. However, there is no text specifically devoted to the harmonisation of rules on detention. And actually, EU harmonisation arguably enjoys a different degree of legitimacy depending on the aspects of detention at stake. For instance, the grounds and length of pre-trial detention are an aspect where harmonisation appears justified. There is evidence that Member States are reluctant to surrender to those states where there are long delays in pre-trial detention, and which are in breach of ECHR standards (Sellier and Weyembergh, 2018, p. 87). In other terms, different legislation and practices constitute legitimate hindrances to mutual recognition which harmonisation can correct. And, following what was said in the "The Criminal Procedure Requirement" section, norms on pre-trial detention can be classified as falling within the notion of "criminal procedure" as they concern indeed the pre-trial phase. There are however other aspects of detention where harmonisation would be intuitively needed, but whose harmonisation can be difficult to justify on the basis of Art. 82(2)b TFEU. The next two subsections discuss two of these examples, namely rules on compensation for unjust detention and material detention conditions. The challenges to justify their harmonisation on the basis of Art. 82(2)b TFEU in these two cases stems, respectively, from the need to respect the mutual recognition requirement and the criminal procedure requirement.

The Mutual Recognition Requirement and the Challenge to Harmonise Compensation for Unjust Detention
A first example of policy area where EU harmonisation would be needed but which is challenging to justify on the basis of Art. 82(2)b TFEU is compensation for unjust detention. In this context, there is a clear deontological need for EU norms, which is however difficult to reconcile with Art. 82(2)b TFEU utilitarian approach. In other terms, transnational administration of justice has created gaps in protection which did not exist before and which needs remedying for fairness reasons (deontological rationale for harmonisation). Yet, it is difficult to argue that harmonisation in this area is necessary for the functioning of mutual recognition, as required by Art. 82(2)b TFEU (utilitarian rationale for harmonisation).
There exist ECHR standards on the question of compensation of unjust detention both in a national and a transnational context. Detention in the context of criminal justice proceedings is regulated firstly at Art. 5(1)a (detention after conviction) and Art. 5(1)c (pretrial detention), naturally binding on the state where the criminal trial is taking place and where the person is detained. Secondly, also Art. 5(1)f (detention in a requested state for, among others, extradition purposes) is relevant and it sets standards binding the state where the person is physically present and where he is detained before being extradited. Art. 5(5) ECHR establishes a right to compensation for unjust detention and it defines it as detention in violation of Article 5(1) thus including both Art. 5(1)a, Art. 5(1)c, and Art. 5(1) f. Notably, compensation must be granted both in case of full acquittal but also in case of acquittal for lack of evidence (Tandem v Spain, 2010) . Naturally in national proceedings, the responsibility for compensation lies on the state carrying out the proceedings. In extradition cases, the Court admitted that requesting states can be held responsible for unjust detention under Art. 5(1)c, and thus their obligation to grant compensation can be engaged if detention occurs abroad but as a result of their extradition request. The requesting state has the same obligations flowing from Art. 5(1)(c) as if detention occurred on its territory (Stephens v Malta, 2009, and Toniolo vs San Marino and Italy, 2012) . This does not negate the requested state responsibility under Art. 5(1)f naturally. Moreover, it can be assumed that if the requested state law/conduct is solely responsible for the unlawfulness of the detention, naturally its responsibility under Art. 5(1)f will be engaged, while the requesting state responsibility under 5(1)c will not (Sørensen, 2015, p. 203).
Member States legislation on compensation in cross-border cases varies (Sørensen, 2015, p. 194). Some states include this possibility, however very restrictively. For instance, Spain was reported as granting compensation only after acquittal (Sellier and Weyembergh, 2018, p. 124). This rules out many other scenarios which can occur in a cross-border case where there is unjust detention. A scenario not covered can be for instance the withdrawal of an EAW, or an arrest based on a mistake on the person made by issuing authorities, which executing authorities could not be aware of.
The UK is also known for having a particularly restrictive regime on compensation for unjust detention. 5 This is still relevant after Brexit. Indeed, the Trade and Cooperation Agreement regulating extradition relations between EU Member States and the UK establishes a system quite similar to the one in the European Arrest Warrant (Grange et al., 2021) , and a steady flow of extraditions will still take place between EU Member States and the UK. EU citizens surrendered to the UK and unjustly detained could still be in need of compensation.
Other States do include compensation in extradition cases but can have restrictive requirements, for instance only granting compensation if they are the executing state, or only if they are the issuing state, but not in both cases. When national grounds for compensation are not complementary, individuals can be caught in the middle and not receive compensation. For instance, a case was reported of Austrian authorities issuing a SIS alert, and Germany having arrested the person. Shortly after, Austrian authorities informed that the alert had been revoked. Germany (executing authority) refused compensation, and it is uncertain whether Austria (issuing authority) provided it (Sellier and Weyembergh, 2018, p. 125). A similar scenario could occur in cases involving Germany, which grants compensation in EAW cases only if it is the issuing state, and the Netherlands, which admits compensation but only when acting as an executing state (Sellier & Weyembergh, 2018, p. 125). This might imply that someone suffering unjust detention in Germany following a Dutch request could not have access to compensation. Other problematic scenarios could be the following. What if an individual is detained in the executing states which nonetheless refuses to surrender, and frees the individual, due to mandatory grounds for refusal, e.g. Ne Bis in Idem (Art. 3(2) EAW)? What if the issuing state was not aware of the previous conviction but the executing state was, and put the person in detention, nonetheless? How should the responsibility be shared? Should the result change if we are talking of an optional ground for refusal, e.g. double criminality? 6 How much diligence should (or can) the executing state put into verifying the information provided by the executing state, for instance about the identity of the person, before putting them in detention? What if a person is put in detention following a disproportionate EAW request? Our interviewee confirmed how difficult it is to foresee during the investigations whether issuing an EAW is proportionate, or rather a European Investigation Order would be more advisable. This makes it difficult to assess how the responsibility should be shared (Interview 1). Or whose responsibility for compensation is it, if the issuing state EAW request is proportionate, considering the crime the person is accused of, but the executing state decision to put them in detention is not, e.g. there is no risk of flight? Or what if pre-trial detention in the issuing state goes over the limits after surrender, but due to delays in the executing state?
Finally, even when the law does exist and individuals might be entitled to compensation, protection might not be granted due to practical issues such as language barriers and unfamiliarity with the law. A case was reported of a Slovak citizen who was arrested and surrendered to the Netherlands. The charges were dropped, and the person was released. Yet he received no support in how to ask for compensation in the Netherlands (Sellier and Weyembergh, 2018, p. 126).
The need for a supranational intervention in this context is clear. It was suggested that an EU instrument should impose to Member States a duty to introduce compensation schemes in EAW cases that an EU wide fund could be created, allowing individuals to ask for compensation in their own country of nationality, and that specific criteria should be identified on how should Member State contribute to the fund (Sellier and Weyembergh, 2018, p. 128;Sørensen, 2015, p. 203). The Parliament similarly raised this issue (European Parliament 2011). However, the practitioners we spoke to have considered it highly unlikely that these gaps in protection from unjust detention would lead to halting surrenders (Interview 2). While there has not been, at least to our knowledge, a systematic comparative study of domestic case law refusing EAW and compensation, this argument sounds intuitively convincing. The prospect that at the end of the whole procedure it might turn that the individual was in fact unjustly detained, and that yet they might not get compensation, sounds too much of a farfetched and hypothetical violation of fundamental rights, to have executing authorities suspending surrender on these bases. It certainly would not be a scenario in which suspending surrender would be allowed, following the Aranyosi and Caldararu case law. And actually, a case was reported in which Member States invoked mutual trust as a reason not to pay compensation. Italian authorities issued a request with incorrect information to Belgium. Belgian authorities failed to verify the information and arrested the wrong person, who was then detained unlawfully. Belgium denied compensation on the basis that it was simply executing a request for mutual cooperation and implementing its duty of mutual trust (Sellier and Weyembergh, 2018, p. 128). Our interviewees confirmed that indeed they are unlikely to suspend surrender in cases in which there might be doubts as to, for instance, the person's identity, because they would trust the foreign authorities (Interview 1, and Interview 2).
If one interprets, Art. 82(2)b utilitarian rationale for harmonisation in a narrow way as defined in section I(a)-harmonisation being legitimate only when a level playing field is necessary for the actual functioning of mutual recognition-it is hard to justify EU regulatory action on compensation regimes on those bases. Mutual recognition seems to be working just fine without harmonisation. One could rely on the broader interpretation of the mutual recognition requirement listed in section 'The mutual recognition requirement', and say that raising fundamental rights standards would generally improve mutual trust. However, as it was stated above, our interviewees do not see diverging regimes on compensation as affecting mutual trust. In truth, it is more from a deontological point of view that EU action is needed, to correct the gaps that mutual recognition has created, or at least accentuated. Harmonisation would be used to address the negative effects of mutual recognition, rather than to facilitate it in a strict sense (Mitsilegas, 2016, p. 158).
Interestingly, EU instruments which envisage compensation rights for individuals already exist within the AFSJ (2003 Framework Decision on Freezing of Orders, the 2006 Framework Decision on confiscation orders, the 2014 European Investigation Order, the 2000 SIS Regulation, and the 2012 Directive on Victims' rights). The Victims' rights directive was interestingly adopted on the basis of Art. 82(2)b TFEU, despite the link between mutual recognition and harmonisation also being hard to trace if interpreted in a strict sense. I have argued elsewhere that this instrument has a bigger individual rights focus, if compared to its predecessor, the Framework on Victims' rights, which conversely saw the victims very much as potential witness, in a law enforcement perspective (Wieczorek, 2011). The Victims' rights Directive was, nonetheless, a strongly criticized instrument (Mitsilegas, 2016(Mitsilegas, , p. 197, Öberg, 2015 showing that while the EU legislator might be open to a broader interpretation of the relevant legal basis, this is not without controversy.

The "Criminal Procedure" Limit and the Challenge to Harmonising Material Detention Conditions
A second area in which EU harmonisation is intuitively needed, but where Art. 82(2) b TFEU might not be an adequate legal basis, is that of material detention conditions. These are the physical conditions of detention (e.g. the quality of "housing and services"), as opposed to those related to the rules regulating (pre-trial) detention (e.g. grounds, deadlines, review proceedings) (Coventry, 2017, 47, Soo, 2020, 338, Baker et al., 2020. In this case, there is a clear utilitarian need for harmonisation. That is harmonisation is necessary for the smooth functioning of mutual recognition, as requested by Art. 83(2) b TFEU. The three-step test devised in the subsection 'The mutual recognition requirement' to identify if harmonisation can support cooperation requires to look at whether the differences between legislations and practice hinder mutual recognition; to assess if these hindrances are legitimate; and to see if harmonisation would solve the problem.
The Aranyosi and Caldararu case provides an answer in the positive to the first two limbs of the test for what concerns material detention conditions. Indeed, the judgement was not only a landmark one in that for the first time it identified some limits to the principle of mutual trust. But it is also paramount because it brought to light that one of the most urgent problems among Member States concerns the differences between detention material conditions. Naturally, evidence that some Member States fell below ECHR standards already existed in the form of Strasbourg case law, and reports of the Council of Europe Committee for the prevention of torture. 7 And Member States were already hesitant to comply with surrender request when these came from jurisdiction where poor prison conditions were documented. The UK has traditionally been among the countries more prone to refuse extraditions requests, including semi-automatic mutual recognition requests issued following EAW pre-Brexit, where there were fundamental rights concerns. The Florea judgement, where surrender to Romania was refused on the basis of overcrowding risk, is a good example of this (Romania, 2014). The trend is bound to continue with respect to surrenders under the Trade and Cooperation Agreement. Similar case law can be found in Germany, Romania, and Greece (Wieczorek, 2019). With Aranyosi and Caldararu, the Court had established that executing authorities could lawfully suspend surrender procedures if serious risks of human rights violations existed for the surrendered person. In the relevant case, such human rights violations were directly linked to the prison conditions. What the Court tells us is that differences in material prison conditions so serious to fall below the existing internationally harmonised standards are a legitimate hindrance to mutual recognition. The judgement was well received by national courts who were quick to implement it Gigengack, 2020, Bard &Van Ballegooij, 2019) . Against this background, it seems fair to argue that there is clear empirical evidence that differences in prison conditions among Member States do hinder mutual recognition, which is the answer to the first step of the test. Secondly, in several states, the conditions fall below the existing EU harmonised standards, and therefore Member States are legitimate in suspending surrender, which is the answer to the second step of the test. One has then to move to the third step of the test: namely, would EU regulatory intervention improve the situation? If what is needed is more clarity and uniformity on what prisons standards are acceptable to meet the Aranyosi test, then adoption of EU law on Art. 82(2)b TFEU basis is clearly justified. But even if the problem is not at a regulatory but at an enforcement level (Baker et al., 2020) , the argument in favour of further EU Law is easily made. EU law can secure higher enforcement rate, if compared with ECHR law. It can rely on a wider range of enforcement tools. Next to centralised infringement proceedings, there is also the possibility to resort to direct and indirect effect, as well as the duty of consistent interpretation. Moreover, through the evaluation reports, the Commission can also assess the state of the law in action in Member States and make recommendations accordingly (Coventry, 2017, pp. 62, Soo, 2020, pp. 333, 339, Mitsilegas, 2016). The first requirement of Art. 82(2) b TFEU, the link with mutual recognition, is thus met. It is questionable, however, if the second one, namely limiting harmonisation only to aspects of criminal procedure, is also met. Subsection 'The criminal procedure requirement' clarified that it is unclear what the expression "criminal procedure" refers to, and in particular if it can extend to standard pertaining material conditions. Even if it did, legislative practice seems to suggest that in any case harmonisation of rules on the post-trial execution phase is excluded, and harmonisation can only extend to pre-trial detention. This would be an important limitation. EAW can 7 Italy is a good example for instance having made the object of the pilot Torreggiani judgement on prison overcrowding (Torreggiani and Others v Italy 2013). The yearly reports of the CPT on Italy can be accessed here: https:// hudoc. cpt. coe. int/ eng#% 22sort% 22: [% 22CPT Docum entDa te% 20Des cendi ng,CPTDo cumen tID% 20Asc endin g,CPTSe ction Number% 20Asc ending% 22] ,% 22CPT State% 22: [% 22ITA% 22. The latest 2020 one still highlights issues with increase in prison population and the fact that prisoners are not given the 4 m2 space in multioccupancy cells. Romania and Hungary are another example of Member States whose prison conditions have been object of a Strasbourg Pilot judgement. (Rezmiveș andothers v. Romania 2017, Varga andothers v Hungary 2015). be used both for prosecution and execution purposes. Leaving standards on detention for execution purposes untouched would mean maintaining the risk of important hindrances to mutual recognition in the case of EAW for execution purposes.
Briefly, the impression one gets is that Art. 82(2)b TFEU and the specific reference to the rights of individuals in criminal procedure were introduced having fair trial rights in mind. This was a policy area where the Commission had long pushed to have some harmonisation on. Detention was not an aspect on which the EU anticipated to take immediate action at the time. The initial Commission proposal for a Framework Decision on procedural rights mainly focussed on fair trial rights (European Commission 2004). And the 2009 procedural rights roadmap, which followed, encouraged the Commission to propose a number of legislative measures in the field of fair trial rights. But it only envisages for the Commission to adopt a Green Paper in the field of detention (Council 2009). Yet, it came to light that a key area in which harmonisation was necessary was not only fair trial rights, but also detention conditions. The relevant Treaty provision however, unless broadly interpreted, arguably proves inadequate to support this endeavour.

Conclusions
This contribution has illustrated the limits of Art. 82(2)b TFEU as a legal basis for the harmonisation of detention rules. The Treaty set as an objective the establishment of an AFSJ in the context of which both security but also freedom and justice must be guaranteed. The argument of this paper is that it is debatable whether the EU has given itself the necessary legal basis for achieving this objective. EU action in the field of detention is necessary from a deontological perspective, namely, to ensure that a certain level of fundamental rights is upheld across the EU, and to correct the imbalances or the gaps that mutual recognition has created. Yet, Art. 82(2)b TFEU only admits harmonisation of national law if three requirements are fulfilled. Firstly, the mutual recognition requirement must be met, i.e. only harmonisation measures aimed at supporting cooperation can be adopted on the basis of Art. 82(2) TFEU. This requirement can be broadly interpreted, i.e. any harmonisation that fosters mutual trust is justified. Or it can be narrowly interpreted, i.e. only harmonisation that eliminates concrete barriers to mutual recognition is admitted. Secondly, the criminal procedure requirement must be met. Legislative practice shows that the expression "criminal procedure" has been interpreted as not including the sentence execution phase. Moreover, it is unclear if "criminal procedure" also includes the rules on material detention conditions.
The limited potential of this legal basis is illustrated in this paper by reference to the area of material detention conditions and compensation for unjust detention. In this context EU action is intuitively needed but its justification on the basis of Art. 82(2)b TFEU is not straightforward.
To really exploit Art. 82(2)b TFEU potential as a legal basis in raising fundamental rights standards and correct the unbalances in the AFSJ, the provision should be interpreted broadly. That is, one should firstly embrace the broader understanding of the mutual recognition requirement, sketched in subsection 'The mutual recognition requirement'. This gives a sort of a carte blanche to harmonisation on the basis of Art. 82(2)b TFEU on grounds that raising fundamental rights standards increases mutual trust. It also requires adopting a broad interpretation of "criminal procedure" which includes both pre-trial and post-trial detention, and goes from rules on the procedures governing detention to those concerning the material standards. However, having an interpretation of the expression "criminal procedure" which differs from the one included in existing legislation based on Art. 82(2)b TFEU could raise questions in terms of legal consistency.
Admittedly, broad interpretations of Treaty articles and especially of EU competence are nothing new, including in the criminal justice field. However, identifying the different possible interpretation of Art. 82(2)b TFEU seemed important first from an analytical perspective, as it illustrates the tension still existing between a deontological and a utilitarian understanding of harmonisation of procedural rights within the EU. Moreover, from a more practical perspective, expansive interpretation and competence creep are constitutionally problematic, and not always politically viable as Member States have to provide their support to draft legislation in the Council. Member States might have been happy to support a broad interpretation of Art. 82(2)b TFEU for the purposes of adopting directives on fair trial rights. However, they might not be equally keen on supporting a similar broad reading in the field of detention. Indeed, any expansive reading of the Treaty provisions means that there is more scope for the EU to regulate, and less room for Member States' regulatory autonomy. Ambitious proposals which do not get the relevant support among Member States, which can also rely on textual argument such as the content of Art. 82(2)b TFEU in this case, risk of being watered down leading to a result with limited regulatory impact.
An alternative would be to incorporate specific fundamental rights standards, including for instance a right to compensation, within mutual recognition instruments themselves, for instance by amending Framework Decision on the European Arrest Warrant. This would open a different issue though. It is reported that Member States are quite resistant to revise such legislation as they fear it might open a Pandora box (Interview 3). This option might therefore be less politically viable.
A rethinking and redrafting of Art. 82(2)b TFEU is probably the most difficult option, at least politically, but the most constitutionally elegant and legitimate one in this context. Funding This work was supported by the Wiener Anspach Foundation. It was presented at the workshop "An Evidence Based Approach to Pre-Trial Detention in Europe", Maastricht University (NL), 31 March to 1 April 2021, and at the workshop "What role for the EU as an Agenda Setter of Human Rights Standards in the field of Detention?" Fitzwilliam College, Cambridge (UK), 16 September 2019.

Conflict of Interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.