Exploring mutual trust through the lens of an executing judicial authority: The practice of the Court of Amsterdam in EAW proceedings

In the groundbreaking decision Aranyosi and Căldăraru, the Court of Justice of the European Union (CJEU) recognised that in exceptional circumstances, the risk of a possible breach of the right not to suffer inhuman or degrading treatments may qualify as a ground to suspend a European arrest warrant (EAW) and, ultimately, bring the surrender procedure to an end. In this judgment (and in the subsequent decision in LM, dealing with the right to a fair trial), the Court has devised a two-tier test to assess the real risk of a violation of fundamental rights after surrender. Yet the Court has left significant discretion to executing authorities in conducting their assessment of risk, thus raising questions as to how the two-tier test would be implemented at the national level. To address some of these questions, this article examines the practice of the executing authority for the Netherlands (the District Court of Amsterdam) concerning decisions on EAWs that may entail a real risk for fundamental rights. To do so, we analyse the judicial reasoning of decisions issued between June 2016 and June 2020 which implement the tests designed by the CJEU in Aranyosi and LM. The results of this analysis indicate that the Court of Amsterdam has gradually shifted the emphasis from mutual trust to fundamental rights. However, the Dutch court resists automaticity and scrutinises the relevance of any information attentively. This attitude indicates a readiness to engage in a dialogue with the issuing authorities together with resistance to indulge in ‘blind trust’.


Introduction
The Court of Justice of the European Union (CJEU) has long relied on the notion of mutual trust to cement the effectiveness of judicial cooperation within the European Union (EU). In its earlier case law, the Luxembourg Court has consistently emphasised the importance of this principle for the interpretation of the Framework Decision on the European arrest warrant (FDEAW). 1 In the CJEU's view, the notion that Member States' authorities should trust each other would preclude an extensive application of the FDEAW's grounds for refusal. 2 The Luxembourg judges made clear that the grounds listed in the FDEAW had to be regarded as exhaustive, despite the absence of a general ground for refusal that would apply in the event of risks of a violation of the requested person's fundamental rights. The CJEU went as far as to interpret the principle of mutual trust as a 'presumption of compliance', that is, a presumption that while applying EU law, national authorities comply with fundamental rights. 3 Yet, in the landmark decision Aranyosi and Ca˘lda˘raru the Luxembourg Court took a more nuanced approach. 4 In essence, the CJEU ruled that if a real risk exists for a requested person to suffer inhuman and degrading treatments in the issuing State, surrender may be postponed and, where no supplementary information can discount such risk, eventually brought to an end. By doing so, the CJEU has espoused an innovative interpretation of the relevant EU legal framework, expanding on the existing rules of the FDEAW to create a new procedure dealing with specific fundamental rights concerns. In order to assess whether a surrender may expose the requested person to a risk of inhuman and degrading treatment, thereby violating Article 4 of the Charter of fundamental rights of the EU (the Charter), the executing authorities need to engage in a delicate test.
Firstly, if these authorities are in possession of evidence pointing to a risk of ill-treatment, they need to verify whether a conceptual risk exists as a result of the general detention conditions in the issuing State: this is referred to as a risk in abstracto, in that it may be triggered by widespread or structural issues affecting the prison system of the issuing State or some parts thereof. The CJEU has indicated the sources of evidence an executing authority may rely on to find the existence of such a risk. This authority can only use information that is 'objective, reliable, specific and properly updated' concerning detention conditions 'prevailing' in the issuing State. These sources of information should prove the existence of deficiencies which may be systemic or generalised, or which may affect certain groups of people, or which may affect certain places of detention. 5 In the paragraphs that follow, we refer to the finding of an abstract risk as 'step 1'.
Secondly, if general detention conditions point to the existence of a risk in abstracto, the executing authorities are under an obligation to engage in a 'dialogue' with the issuing authority, by requesting supplementary information on the basis of Article 15(2) FDEAW. This information should focus on the conditions in which it is envisaged that the requested person will be detained. As the CJEU stated, the finding of an abstract risk based on the general conditions of detention in the issuing State cannot lead, in itself, to refusing an EAW. Rather, this finding creates an obligation for the executing judicial authority to conduct a further assessment and verify whether the individual concerned will be exposed to a risk in concreto upon surrender. 6 If a more specific risk is found, the executing authority must postpone the proceeding for a reasonable time and request further information that might discount such risk. In the paragraphs that follow we refer to the finding of a substantial risk as 'step 2'.
The CJEU's stance in Aranyosi has been widely regarded as an attempt to offer greater protection to fundamental rights while maintaining a formal adherence to the principle of mutual trust. 7 Far from accepting an indiscriminate deviation from this principle, the Court has confirmed that derogations from a 'presumption of compliance' with fundamental rights may only be allowed in 'exceptional circumstances'. This explains why the Court has formally ruled out the use of existing grounds for refusal to halt a surrender proceeding. If, after taking step 2, the executing authority finds that a real risk of ill-treatment cannot be discounted within a reasonable time, the procedure may 'be brought to an end' but cannot be formally refused. The distinction between this solution and the use of an ordinary ground for refusal may appear elusive. 8 Yet the Court's approach implies that a decision to discontinue surrender should only be taken as a measure of 'last resort'.
Subsequently, the CJEU has accepted further exceptions to the principle of mutual trust. In LM, the Luxembourg Court has devised a similar two-tier test to deal with risks of violations of a requested person's right to a fair trial. 9 This risk-assessment process borrows some features of the Aranyosi test. A real risk for the right to a fair trial should be established on the basis of a general assessment of the justice system in the issuing Member State, with particular attention to the independence of courts ('step 1'). Yet, a finding of systemic and generalised deficiencies in this respect does not suffice to discontinue the proceeding. Rather, it creates an obligation to request further information and verify whether the individual concerned might be exposed to a risk in concreto ('step 2'). If a risk exists that the requested person's right to a fair trial may be infringed, the EAW must be postponed and could eventually be brought to an end.
Both Aranyosi and LM seek to establish an equilibrium between opposing interests: the need to protect fundamental rights and the effectiveness of judicial cooperation. While the possibility to deviate from mutual trust is explicitly recognised by the CJEU, this may not lead to an outright refusal to cooperate. Rather, executing authorities may only discontinue a surrender proceeding after engaging in a dialogue with their counterparts in the issuing State. Ultimately the decision to allow a surrender or bring the procedure to a close is left in the hands of the executing authority. It is submitted that while engaging with the various steps of the test, the executing authority conducts 6

284
New Journal of European Criminal Law 11(3) a balancing exercise between the need to uphold fundamental rights and its commitment to secure effective judicial cooperation in the name of mutual trust. This warrants a closer look into the practice of judicial executing authorities. The case study in this article is the District Court of Amsterdam (CoA), the executing authority for the Netherlands under the terms of the FDEAW. In order to provide an overview of this court's practice, we conducted an analysis of decisions issued from June 2016 (in the aftermath of the CJEU's ruling in Aranyosi) to June 2020. In order to apprehend the way in which the Dutch court finds a balance between the need to protect fundamental rights and the effectiveness of judicial cooperation, we have undertaken an analysis of all these decisions to highlight the reasoning used by the CoA. Note, however, that only a sample of this data set is referred to in the footnotes. We begin by presenting the executing procedure in the Netherlands in order to explain how the Aranyosi and LM tests were implemented in the Dutch legal framework. We then move on to discuss the reasoning of the CoA in more detail by analysing the legal arguments behind each step of the tests.
The executing procedure and the implementation of the two-tier tests in the Dutch legal order In the Netherlands, the FDEAW has been primarily implemented by means of the Dutch Surrender Act ('Overleveringswet'; OLW). 10 In order to ensure the efficiency and legal uniformity of decisions, the Dutch legislator chose to appoint the CoA as the only competent authority to deal in first and final instance with the execution of EAWs. 11 The choice of establishing a single specialised authority to deal with EAW requests was meant to streamline the decision-making process and avoid diverging practices across the Netherlands. The CoA was given its role as executing authority in view of its existing competences in the field of extradition and judicial cooperation in criminal matters. Attached to this court is a specialised chamber, the 'Internationale Rechtshulpkamer' (International legal assistance chamber; IRK), competent on matters of extradition and surrender. 12 The first stage in the executing procedure begins with the receipt of an EAW. The EAW, supplemented by a translation in one of the languages accepted by the Netherlands (Dutch or English), is sent or forwarded to the public prosecutor at the Public Prosecutor's Office in Amsterdam. The prosecutor carries out a first scrutiny by assessing whether major obstacles exist that prohibit surrender. 13 It also verifies the completeness of the information in the EAW form and may ask the issuing authority to complete or verify it. 14 In addition, if the requested person accepts his 10 or her immediate surrender, the prosecutor may directly execute the EAW. 15 By contrast, if the concerned individual does not accept to follow this procedure, the prosecutor submits the EAW to the CoA within three days of its receipt. 16 The second stage of the procedure commences when the prosecutor submits the EAW to the court. Along with the EAW the prosecutor files a request with the CoA to decide on the case by presenting the available information. Upon receiving the request, the court's president sets a date for the first hearing and summons the individual to appear before the court. At the first hearing, the CoA conducts an in-depth analysis of the possible grounds for non-execution of the EAW. If none of the grounds listed by the OLW and the FDEAW can be invoked, the executing authority is under the obligation to allow the surrender. 17 After a final verdict has been rendered, Dutch law provides no opportunity to appeal a decision allowing surrender. 18 The OLW has implemented all grounds for refusal listed in Articles 3, 4 and 4a of the FDEAW. Yet, the Dutch legislature transposed all of the optional grounds for refusal as mandatory. 19 Furthermore, some of these grounds appear to fall beyond the scope of the FDEAW. One of the grounds exceeding the scope of the FDEAW concerns the risk of fundamental rights violations upon surrender to the issuing Member State. Article 11 OLW prohibits the surrender of a requested person when this would lead to a flagrant violation of one or more fundamental rights guaranteed by the European Convention on Human Rights (ECHR). Arguably, while this provision is meant to prevent the risk of future violations, it might also be invoked in the case of 'past or completed' breaches of fundamental rights. 20 While Article 11 OLW is meant to guarantee a wide range of fundamental rights, the CoA did not rely on this provision to incorporate the Aranyosi test into its practice. Before the judgment in Aranyosi, this provision had been invoked by the defence to halt surrender on grounds of possible violations of Article 3 ECHR and could therefore appear a natural solution to give effect to the CJEU's ruling. 21 While one can only speculate as to why the CoA decided to follow a different path, it is clear that Article 11 OLW lays down rather stringent requirements. Surrender must be refused if a 'real risk' exists that the surrender would lead to a 'flagrant violation' of fundamental rights of the person concerned as laid down in the ECHR. 22 This provision thus identifies a high threshold by referring to the notion of 'flagrant violation' and does not apply prime facie to the rights protected under EU law by the Charter.
In addition, refusal may only take place after the executing authority has conducted an exhaustive examination at the hearing. Under Dutch law, a refusal is conclusive and leaves little room to acquire further evidence. The decision to trigger a ground for refusal presupposes that the request made by the prosecutor while submitting the EAW is admissible and supported by sufficient information. 23 On the contrary, the Aranyosi test only allows to end the proceedings after

286
New Journal of European Criminal Law 11 (3) requesting supplementary information to no avail, thereby implying a lack of information that would hinder the execution of the EAW. When no additional information is provided, a different procedural outcome should be preferred. The CoA has therefore concluded that a lack of information allowing to discount the risk of a violation would cause the prosecution to be declared 'inadmissible' ('niet-ontvankelijkheid van het Openbaar Ministerie'). 24 This indicates that the prosecutor was not entitled to bring the case to court, thereby preventing the CoA from deciding on the merits of the case. 25 The structure of the risk-assessment process is different depending on the fundamental rights at stake. We will begin with an analysis of the procedure followed to assess the risk of inhuman and degrading treatments (the Aranyosi test). When the CoA has found a real risk of violation in previous decisions on EAWs originating from the issuing State, it may confine itself to acquire information ex officio. This is done to verify whether the assessment conducted on previous occasions may be reversed by some new information on the prison or court system of the issuing State. By contrast, if no risk for fundamental rights has been established in prior cases with regard to a specific Member State, it generally falls to the defence to put forward 'objective, reliable, specific and properly updated information' indicating that a requested person might be exposed to a violation of fundamental rights in the issuing State. 26 Before engaging with the two-tier test, the CoA may formulate a preliminary assessment, which was not contemplated by the CJEU's ruling in Aranyosi. For the purpose of this article, this additional stage has been labelled 'step 0'. Arguably, this preliminary step is taken when the available information did not meet the threshold set by the CJEU. If this information cannot be regarded as sufficient, the CoA is not allowed to take step 1. Nevertheless, if any piece of information gives reason to assume that a real risk exists, the CoA may ask the issuing authorities for supplementary information or assurances in order to establish the existence of such a risk. 27 After this information has been received, the CoA may choose to: (i) ask for additional information; (ii) take step 1 and ask for more information, or; (iii) allow the execution of the EAW without taking any step. 28 If the CoA finds that the sources of information are deemed sufficient, it may take step 1. As indicated above, this amounts to acknowledging that a 'real risk' of ill-treatment exists in abstracto. After taking step 1, the CoA entrusts the public prosecutor to request more information on the basis of Article 15(2) FDEAW. According to the CJEU's ruling in Aranyosi, the issuing authority is in principle obliged to reply to this request. 29 The information provided on this basis will be used to decide whether step 2 of the test should be taken. As explained earlier, after taking step 1 the CoA is under an obligation to assess whether in the case at hand substantial grounds exist to believe that the requested person will be exposed to a risk of ill-treatment in the issuing State. This amounts to assessing the existence of a risk in concreto. If satisfying information is provided which discounts the risk for the concerned individual, the CoA allows the execution of the EAW without taking step 2. 31 The CoA may come to the same conclusion if assurances are provided that the requested person will not be exposed to a risk of illtreatment. While the provision of assurances was not explicitly envisaged by the CJEU in Aranyosi, 32 the possibility to provide guarantees against the risk of ill-treatment has been explicitly admitted by some of the Luxembourg Court's most recent judgments. 33 In practice, the provision of assurances often reflects the commitment to avoid placing the requested person in one or more facilities where an abstract risk of ill-treatment exists.
When dealing with claims concerning the lack of judicial independence (the LM test), the CoA follows a slightly different procedure. First, the court ascertains whether a risk in abstracto exists on account of systemic or generalised deficiencies which endanger the independence of the judiciary of the issuing Member State (step 1). 34 Next, the CoA examines whether a risk in concreto exists. In light of this assessment, the Dutch court analyses whether those deficiencies might impact the courts in the issuing Member State with jurisdiction over the proceedings to which the requested person will be subjected. 35 The CoA has labelled this assessment as 'step 2' but in order to avoid confusion with step 2 of the Aranyosi test it will be referred to as step 2A.
Arguably, although this additional step has not been explicitly devised by the CJEU, it was implicit in the two-tier test designed in LM. 36 However, this stage of the procedure does not lead to find a risk in concreto for the requested person. The Dutch court might only come to this conclusion after taking step 2B (or 'step 3' as it is commonly referred to by the CoA). In taking step 2B, the CoA ascertains whether substantial grounds exist for believing that the requested person's right to an independent court will be violated and, as a result, whether the right to a fair trial will be affected. In this assessment, the personal situation of the requested person, the nature of the offence and the factual context that forms the basis of the EAW might constitute grounds for the Dutch court to find the existence of a risk in concreto, after which the procedure must be brought to an end.
Assessing the reasoning behind the different steps of the Aranyosi test: The use of 'step 0' In this paragraph, we rely on the analysis of the CoA's reasoning to highlight how the Dutch executing authority has reconciled the obligation to protect the right not to suffer inhuman and degrading treatments with the principle of mutual trust. We conduct an analysis of the various stages of the Aranyosi test as explained in the paragraph above. A preliminary remark is in order: despite our attempt to single out each of the steps of the test for a separate analysis, most of the decisions do not articulate a clear-cut division between the different stages. This is due to the CoA's practice of referring to its deliberations on both step 1 and step 2 in previous cases regarding

288
New Journal of European Criminal Law 11(3) EAWs issued by the same Member State. As a result, the assessment of risks in abstracto or in concreto is sometimes conducted implicitly. As already pointed out, the CoA may reject a claim put forward by the defence if it takes the view that the provided information does not qualify as 'objective, reliable, specific or properly updated'. For example, with reference to an EAW issued by the Czech Republic, the defence claimed that in the issuing country persons with Polish nationality were discriminated against and regularly placed in detention facilities where prison conditions fell below the standards of Article 4 of the Charter. However, the CoA did not find sufficient information to substantiate this claim. It thus rejected the application of the Aranyosi test and subsequently allowed the execution of the EAW without taking step 1. 37 In some cases, even though the available information does not meet the requirements set by the CJEU, the CoA may nevertheless stay the proceeding and ask questions on the basis of a 'step 0'. For example, in one of the first decisions applying the Aranyosi test, the CoA had to decide whether repeated strikes of prison staff in certain Belgian prisons would expose the requested person to risks of ill-treatment. Significantly, the CoA held that the sources of information provided by the defence (mostly reports by media outlets) were not sufficient to assert a risk in abstracto. 38 Yet it decided to stay the proceeding and request more information. Though refraining from a direct acknowledgment of an abstract risk, the CoA could not rule out that such danger would eventually materialise. In this instance, the CoA showed sympathy for the concerns raised by the requested person even though the information provided was not sufficient to uphold the finding of a systemic risk.
In other cases, the use of step 0 would reflect the CoA's caution to take step 1. Significantly, when in July 2018 new strikes took place in prisons all over Belgium, the CoA still refrained from taking step 1. Most notably, despite the claim made by Belgian authorities that due to reduced service of prison staff they could not guarantee the rights protected by Article 4 of the Charter during the strikes, the CoA resisted from declaring the existence of generalised risks. 39 Instead, it decided to adjourn the hearing until more clarity could be provided regarding the situation. 40 Although the reasoning behind step 0 is hardly explicit, its use in recent cases concerning EAWs issued by Belgium may be regarded as an attempt to keep concerns of ill-treatment at bay while avoiding the negative effects that may derive from declaring the existence of 'systemic' deficiencies in the issuing State.
Step 1: Systemic deficiencies, precedents and the role of assurances As reminded above, the CJEU has indicated three major factors that may prove a risk in abstracto: (i) the existence of deficiencies that are systemic and generalised; (ii) that affect specific groups of prisoners, or; (iii) that affect certain places of detention. In the majority of cases, the CoA associates a risk in abstracto with the presence of 'systemic and generalised' issues. Mostly, these deficiencies relate to a lack of personal space in prison. This is not entirely surprising and reflects a development in the context of the jurisprudence of the European Court of Human Rights (ECtHR) on Article 3 ECHR. As we will explain, in the Strasbourg case law the 'space factor' features prominently as one of the criteria to adjudicate on the existence of inhuman and degrading treatments in prison. 41 In the context of decisions dealing with EAWs from Romania, Bulgaria and Hungary, the CoA has determined the existence of a risk in abstracto primarily on the basis of previous ECtHR's judgments against these countries. 42 In addition, the finding of an abstract risk is often demonstrated by CPT reports. These reports contain accurate overviews of detention conditions in a Member State and are also often relied upon as evidence in ECtHR's judgments on Article 3 ECHR. 43 Arguably, when the ECtHR identifies a 'structural problem' giving rise to systemic violations of Article 3 in one Member State, the CoA is more inclined to conclude that a risk in abstracto exists. In these cases, the CoA may even gather the relevant information ex officio and take step 1 by referring to its previous decisions on EAWs issued by the same Member State. 44 However, on certain occasions the CoA has critically scrutinised the content of ECtHR's judgments as sources of evidence. In a recent case concerning Hungary, the Dutch court verified whether its previous finding on systemic overcrowding was still supported by 'properly updated' evidence. 45 The CoA stated that the most recent rulings showing widespread violations of Article 3 ECHR in Hungarian prisons had been issued in July 2016. Since then, all further applications to the Court of Strasbourg had been declared inadmissible. 46 Moreover, the latest relevant CPT report dated from April 2014 and was based on a visit in 2013. As a result, the CoA found that this information could no longer be regarded as 'properly updated'. In the absence of more recent information, it decided to allow surrender.
The CoA may also decide to take step 1 on the basis of information concerning 'specific places of detention'. Here, details provided by CPT reports are relied on as they provide specific data about single detention facilities. With regard to Portugal, in recent decisions the CoA referred to the findings of the latest CPT report to state that a real risk could be found in relation to prison facilities in Lisbon, Caxias and Setúbal. 47 The CoA thus accepted that, in abstracto, prisoners could be exposed to inhuman or degrading treatments within those facilities and took step 1 in relation to several EAWs issued by Portuguese authorities. 48   however refrained from finding a risk in concreto on the basis of the assurances provided by the issuing State. With regard to Lisbon, it was guaranteed that the requested person would not be detained in any of the sections where the CPT had found inadequate detention conditions. Moreover, Portuguese authorities guaranteed that the requested person would not be placed in the facilities of Caxias and Setúbal. 49 The defence may also point out the existence of risks associated with deficiencies affecting 'certain groups of people'. This may occur when specific aspects of prison regimes are believed to negatively affect certain categories of vulnerable prisoners. In one case, the CoA was faced with an EAW issued by Sweden for a person suffering from mental health problems. This prompted the Dutch court to ask more questions to the issuing Member State in light of a CPT report which indicated the frequent use of severe restrictions (e.g. isolation and confinement) against pre-trial detainees. 50 According to the CoA, the restrictions applying to Swedish pre-trial detainees could pose a 'general risk' of ill-treatment to persons with 'psychiatric illness and suicidal tendencies'. The Dutch court therefore stayed the proceeding on the basis of step 1 and asked specific questions about the treatment faced by the requested person.
All in all, the CoA seems to follow the 'playbook' devised by the CJEU in Aranyosi. The decision to take step 1 is based on well-documented evidence and subject to a strict scrutiny by the Dutch court. The finding of a risk in abstracto leads the CoA to stay the proceeding and engage in a dialogue with the issuing authority. Yet, the existence of previous judgments highlighting 'systemic and generalised' deficiencies in the issuing country often allows the CoA to treat the abstract risk in the issuing State as a 'notorious fact'. 51 Where structural problems like prison overcrowding have already been highlighted and the relevant information is not outdated, the CoA takes step 1 almost automatically upon receiving the EAW. 52 The CoA tends to operate similarly while dealing with risks affecting 'certain places of detention' as it relies on its own reasoning in previous decisions in order to find evidence of a risk in abstracto. 53 In the event of risks affecting specific prisons, the CoA is prepared to accept the assurance that the requested person will not be detained in one or more facilities where an abstract risk of illtreatment has been ascertained. In practice, assurances are often blended into information that issuing authorities provide upon request of the Dutch court. Yet, the CoA appears increasingly willing to surrender requested individuals on the basis of assurances granted by the issuing State. These guarantees tend to focus on the available personal space 54 or on general prison conditions in the facility where the person is likely to be detained. 55 The provision of assurances often prevents the CoA from taking step 2. After all, as the CJEU has stated, when an assurance is given (or endorsed) by a judicial issuing authority for the purpose of an EAW, the executing State 'must rely on it'. 56 Step 2: The presumption of risk and the 'simplified proceeding' The CoA may in principle postpone an EAW if information provided by the issuing authority points to the existence of a risk in concreto. The question therefore arises as to what kind of evidence the Dutch court may use to conclude that a substantial risk exists. When the 'risk' involves a lack of personal space in the prisons of the issuing State, the CoA applies the criteria developed by the ECtHR to measure the seriousness of prison overcrowding. 57 The reference goes to the principles of the Grand Chamber's ruling in Muršiç. 58 According to this judgment, the space factor alone would give rise to a 'strong presumption' of inhuman and degrading treatment if the personal surface allocated to each prisoner falls below the minimum threshold of 3 m 2 in a multioccupancy cell. The ECtHR's criteria have been endorsed by the CJEU in its recent case law as a way to provide guidance in the context of EAW proceedings. 59 The presumption of a violation may only be rejected if the issuing authorities provide evidence that the reduction of personal space will be compensated by the following counter-balancing factors: (i) the reductions of personal space are 'short, occasional and minor'; (ii) the 'freedom of movement outside the cell' is guaranteed, along with the provision of 'adequate out-of-cell activities', and; (iii) general prisons conditions are 'appropriate' with no other aggravating aspects. Importantly, national courts need to assess these factors cumulatively.
As a result, the CoA has often asked the issuing authority to provide measurable indications of the number of square metres available within the penitentiary facilities where the requested person is likely to be detained. 60 When this information is unable to offer conclusive guarantees that the personal space will not fall below 3 m 2 threshold, the Dutch court asks for a 'reasoned and detailed statement' of the 'compensatory circumstances' that may rebut the presumption of a violation. 61 These may include details on the applicable detention regime (e.g. semi-open regime) or on outdoor and working activities offered as part of the rehabilitative treatment in prison. Unless these 'compensatory circumstances' are sufficient to refute the presumption based on the 'space factor' the CoA will take step 2. 62 On the other hand, if the 'risk' found under step 1 relates to the possible detention in one or more specific prisons in the issuing State, the CoA would ask detailed questions about the place where the requested person will be detained upon surrender. For example, in a case concerning an EAW issued by France it appeared that, according to the latest CPT report on that country, detention conditions in Fresnes, Nîmes or Villepinte might be in breach of Article 3 ECHR due to serious

292
New Journal of European Criminal Law 11 (3) overpopulation. 63 The CoA therefore went on to ask whether the concerned individual might be detained in one of these facilities and if so, under which conditions. 64 The replies provided by French authorities were relied on (in a subsequent case) to assess whether a substantial risk existed that the requested person would be detained in Nîmes. 65 The CoA found that the information provided for this facility was not sufficient to rebut the presumption of violation triggered by the lack of personal space. 66 As explained above, the way in which the CoA takes step 2 may vary depending on the circumstances. When the Dutch court already has a considerable amount of information about prison conditions in an issuing State, it seems inclined to refer to the findings included in some of its previous decisions. To the best of our knowledge, this happens when EAWs issued by the same country have been repeatedly dismissed as a result of both systemic and substantiated concerns regarding the risk of ill-treatment. Over time, the CoA has developed what could be referred to as a 'simplified proceeding' in order to deal more rapidly with repeated claims of inhuman and degrading treatment. This allows the CoA to decide on the execution of an EAW in a single hearing, that is, without postponing the surrender proceeding.
The use of this 'simplified proceeding' is a visible pattern in some recent cases concerning EAWs issued by Romania. Firstly, in these cases the CoA briefly refers to relevant decisions of the ECtHR in light of its own 'established case law' concerning the existence of 'generalised and systemic deficiencies' in the issuing State. 67 Secondly, these cases show a considerable involvement of the public prosecutor in collecting all necessary information before bringing the case to court. Arguably, the prosecutor acts on the assumption that, if previous proceedings towards the same Member States have been discontinued, a new request to execute the EAW may only be filed in light of updated and specific information. 68 On this basis, the prosecutor may directly argue that its request should be declared inadmissible at the first hearing, thus allowing the CoA to swiftly end the proceeding. 69 In any event, once an EAW is submitted to the CoA the information previously gathered by the prosecutor allows the executing court to decide immediately on step 2. In addition, in some recent cases the CoA concluded that, in light of the information provided by the issuing authority before the hearing, it could 'not be expected that relevant additional information will be provided in the foreseeable future in order to discount risks' for the concerned individual. 70 In a handful of cases, this finding led the CoA to terminate the proceeding. 71 As reminded above, according to the CJEU a surrender can only be brought to an end after postponing the surrender 'for a reasonable time'. Yet, in light of the information provided by Romanian authorities 72 and collected by the prosecutor before the hearing, the CoA found no reason to allow the issuing authority a further extension.
The 'simplified proceeding' summarised above is arguably a deviation from the CJEU's ruling in Aranyosi. The choice of dismissing an EAW without postponing it seems at odds with the choice made by the Luxembourg Court in its case law as the obligation to postpone the EAW is meant to involve the issuing authority before turning down its request. Admittedly, the executing court may decide to disregard any further information which does not discount the risk of ill-treatment. Yet, in the CJEU's view this decision may only be taken ex post on the basis of the most recent information available. On the contrary, with the 'simplified proceeding' the CoA dismisses the execution of the EAW ex ante, that is, on the basis of information already available to the court and without leaving the issuing State a right to reply. In doing so, the CoA shifts the balance in favour of the individual's fundamental rights.
Waiting for tomorrow? The interlocutory findings of the LM test and the right to a fair trial All cases applying the LM test focus on EAWs issued by Poland. 73 Initially the CoA only applied the test to EAWs issued for the purpose of conducting a criminal prosecution. 74 More recently, the court has decided to expand the scope of the test to EAWs concerning the execution of a custodial sentence issued after 1 October 2017 75 or when possibilities for appeal and/or retrial are listed in the EAW. 76 It should be noted that, among the cases covered by our survey, none has so far resulted in a decision to terminate the proceeding. Yet the findings of the Dutch executing authority are by no means less relevant. Since the first case in which the CoA applied LM, 77 one conclusion has been unambiguous: the structural deficiencies affecting Poland's judicial system are of such order as to put at risk the independence of the Polish judiciary.
At the end of 2017, several amendments entered into force which were widely regarded as a threat to judicial independence in Poland. The election method of the National Council of the Judiciary (NCJ) was modified, resulting in an increase of political influence on this body. Through a separate piece of legislation, the retirement age of judges sitting on the Supreme Court was lowered which effectively shortened the term of office of its judiciary by 40%, including its President. Moreover, a 'Chamber of Extraordinary Control and Public Affairs' was established, and far-reaching adjustments were passed to alter the functioning of the Disciplinary Chamber of the Supreme Court. Although some of these amendments were subsequently dialled back, most of the legislative changes remained in place. 78 In October 2018, the CoA found that these reforms could give rise to systemic or generalised deficiencies and accordingly took step 1. 79 The CoA based its decision on a large number of sources, for example, the Opinion of the Venice Commission, 80 the reasoned proposal of the European Commission (EC) on the basis of Article 7(1) TEU, 81 and the report by the special rapporteur on the independence of judges and lawyers of the United Nations Human Rights Council. 82 In its assessment, the CoA concluded that these changes empowered the legislative and executive powers to interfere with the administration of justice, thus posing a serious threat to the independence of the Polish judiciary. Moreover, not only had the amendments been implemented but they were also being put into practice. In all subsequent decisions the CoA therefore chose to take step 1, ascertaining the existence of systemic and generalised deficiencies in the Polish judiciary. 83 In a number of follow-up decisions, the CoA moved on to verify the existence of a risk in concreto by examining whether the identified deficiencies might result in negative consequences for Polish courts with jurisdiction over the proceedings to which the requested person would be subjected (step 2A). In line with LM, the CoA engaged in a dialogue with Polish authorities on the basis of which information was received concerning: the number of (vice)presidents of the relevant courts laid off, the number of new judges nominated and appointed, how criminal cases were assigned to judges, whether any judges had been summoned before the Disciplinary Chamber and the possibilities for extraordinary appeal procedures. 84 The Dutch court then moved on to verify whether a specific risk existed that the person's right to an independent court could be breached upon surrender (step 2B). 85 For this assessment, the CoA normally considers the information provided by the requested person, taking into account his personal situation, the nature of the offence for which he is being prosecuted and the factual context surrounding the EAW.
In line with LM, the CoA places the burden of proof on the requested person, who needs to substantiate that the right to a fair trial would be breached if the EAW is executed. Significantly, although the requested persons argued in numerous cases that a risk in concreto existed, thus far the CoA has not been able to find sufficient information to take step 2B and bring the procedure to an end. 86 The burden of proof sets a very high threshold for the requested person, one that is very hard to reach in practice as detailed information on the organisation of the judiciary may be challenging to collect for the defence. This procedural rule also makes it extremely hard for the CoA to formally postpone the proceeding on the basis of step 2B. The absence of evidence, which may not be gathered ex officio, has led the court to execute most of the EAWs received from Poland.
For example, in a decision adopted in January 2020, the CoA concluded that new disciplinary measures adopted in Poland (see below) might threaten the independence of judges with jurisdiction over the case of the requested person. 87 All information provided on these measures confirmed the impact on the independence of the Polish judiciary and, as a result, on the right to a fair trial of the concerned individuals. Subsequently, the CoA decided to refrain from asking further questions for the purpose of step 2A. 88 This effectively led the CoA to halt the dialogue with Polish authorities. However, this decision had no effect on the burden of proof, which remained incumbent on the individual in light of step 2B. In the absence of any information provided by the requested person demonstrating a substantial risk of violation the CoA maintained that surrender should be allowed. 89 The recent developments and the adoption of Poland's 'muzzle law': Re-opening the dialogue In December 2019, the Labour and Social Insurance Chamber of the Polish Supreme Court decided that, in light of the ruling by the CJEU in AK, 90 the NCJ in its current composition could no longer be regarded as an independent and impartial tribunal due to the (in)direct influences of legislative and executive powers. Moreover, the Labour Chamber stated that the Disciplinary Chamber of the Supreme Court could not be considered a court within the meaning of the Charter and the Constitution of the Republic of Poland. The Polish authorities retaliated with the adoption of the 'Law on the judiciary' of 20 December 2019 (the so-called 'muzzle law') which entered into force in February 2020. This law prevents the judiciary from questioning the legitimacy of any judges appointed by the President and the constitutionality of central authorities. Moreover, the 'muzzle law' confers new powers to parliament and the executive in the nomination and appointment of judges and facilitates new and stricter disciplinary proceedings. In light of these developments, several procedures were started by a number of European bodies. Most notably, in April 2020 the EC commenced its fourth infringement procedure against Poland after obtaining the issuance of interim measures by the CJEU. 91 In a decision on a Polish EAW issued for the purposes of criminal prosecution in March 2020, the CoA 'automatically' took step 1 and step 2A in line with earlier case law. However, in light of the developments mentioned above, the CoA decided to adjourn the hearing in order to further investigate the situation regarding the rule of law. 92 In the follow-up decision adopted in June 87. CoA 16 January 2020, ECLI: NL: RBAMS:2020:181. 88. Just as before, the CoA noted that these questions might remain relevant for step 2B and added that this might change in light of future developments.

296
New Journal of European Criminal Law 11(3) 2020, the CoA found that recent legislative changes put such pressure on the independence of the judiciary that they could no longer be disregarded for the purposes of a decision on surrender. 93 The CoA based this conclusion, for example, on the Venice Commission Joint Urgent Opinion of 16 January 2020 94 and media reports indicating that the Disciplinary Chamber was still operating after the interim measures of the CJEU. Consequently, the CoA resumed its dialogue with the issuing authorities for the purposes of step 2B and required more information regarding the practices of the Disciplinary Chamber, and the current developments surrounding the 'muzzle law'. Moreover, the Dutch court inquired whether any disciplinary actions had been undertaken against competent judges. In its decision, the CoA has given a stringent deadline for the issuing authority to provide its answers; at present, it remains impossible to foresee how Polish authorities will react. Significantly, in this case, the defence has questioned the validity of the rules on the burden of proof as interpreted by the CoA and argued that the executing authority may refuse to give effect to an EAW even in the absence of specific information provided by the requested person.

Conclusion
This article started from the assumption that the two-tier test designed by the CJEU left significant discretion to executing authorities. Our analysis of the CoA's reasoning has shown that the Dutch court has used its discretion to gradually shift the emphasis from mutual trust to the fundamental rights of requested persons. This approach resulted in the introduction of a 'step 0' and increased use of summary proceedings to deal with repeated claims of ill-treatment. One of the key aspects guiding the CoA's discretion lies in the existence of consolidated findings of risk. Still this is clearly not the only factor, as the CoA resists automaticity and scrutinises the relevance of any information. This attitude indicates a readiness to engage in a dialogue with issuing authorities and resistance to indulge in 'blind trust'. Moreover, the two-tier tests allow a great deal of flexibility to executing authorities, especially in the context of highly politicised cases such as those involving the independence of the judiciary in the issuing State. Yet, the burden of proof placed on the concerned individual poses significant hurdles for the CoA to demonstrate the existence of a real risk for the requested person's right to a fair trial in specific cases. This poses an obstacle to formally postpone and discontinue surrender. Recent decisions however indicate that a new approach might be forthcoming.