The Burden of Proof in European Court of Human Rights Proceedings Involving EU Law after the EU’s Accession to the ECHR

Christine Bicknell[*]

1. Introduction

As is well known, the Treaty of Lisbon created an obligation for the European Union (EU) to accede to the European Convention on Human Rights (ECHR).[1] This vision, turned project, now spans decades, yet it has not come to fruition.[2] With the draft accession agreement published in March 2023[3] (DAA 2023) it is hopeful that EU accession will indeed occur, thereby closing any remaining gaps in rights protection that emanate from the EU. For although all EU member states are also states parties to the ECHR, it is inescapable that those member states have obligations to comply both with EU law and with the ECHR. Very occasionally it has seemed impossible for member states to satisfy both, and currently the member states, but not the EU, can be found to have violated the ECHR. This will change with EU accession.

However, the EU’s accession to the ECHR is an especially complicated area, akin to a three-dimensional puzzle with several vital and interlocking elements.[4] They all intersect but delicately: when the DAA 2013, an assumedly solid thing, was picked up by the Court of Justice of the European Union (CJEU) in Opinion 2/13, it fell apart.[5] Negotiators since 2020 have had the unenviable task of accommodating certain core (rigid) features of EU law (eg. the need to respect EU autonomy, mutual trust), and fixing and reshaping further aspects according to those requirements, (eg. joint responsibility and the co-respondent mechanism). Other features still may or may not be used, useful or relevant post-accession (eg. Bosphorus presumption of equivalent protection[6]), and the minds of the CJEU judges remain a crucial unknown factor.[7] Ultimately it will be seen whether, and if so, how these pieces all fit back together into a coherent (albeit slightly altered) shape that is this time satisfactory to the CJEU.

In this article, detailed consideration is given to one very specific procedural aspect of the European Court of Human Rights’ (ECtHR) work post-EU accession: the burden of proof in proceedings involving EU law. This is considered with reference to arrangements under the DAA 2023, which provides inter alia for a co-respondent mechanism enabling the EU and one or more member state(s) to become co-respondents in a case.[8] Where invoked, the co-respondents would hold joint responsibility for any ECHR breach,[9] whether originating with EU law, the actions of the member state(s), or both. Notably the co-respondent mechanism not only covers situations where EU law may be the problematic point within a case, but encompasses situations in which the EU and a member state might be working together in joint operations: as for example the alleged direct involvement of the EU border force (Frontex) in migrant pushbacks.[10]

Examining the DAA 2023 with this focus on burden of proof is unique in part because the relevance of burden of proof may not be immediately obvious.[11] In fact, burden of proof is an important procedural dimension that, when considered in detail within the context, enables meaningful scrutiny of the meeting point between proceedings of different types such as could arise, and the EU’s prerequisites. More precisely, when we examine how the burden of proof might operate in ECtHR proceedings post-EU accession, certain cracks and tensions are made visible. In this way, and indeed the significance of the article, is that it offers a procedure-oriented academic stress-test of the puzzle’s draft solution, DAA 2023. The outcome is a series of demonstrable challenges and tensions which remain. In particular there are differences in approach to proof between the two systems. There are aspects within EU laws which are vulnerable to challenge for ECHR compliance and which, it is suggested, the EU should look to pre-accession. Under the co-respondent mechanism proving a case means proving a cause, making joint-responsibility something of a legal fiction at the end point of the process. Additionally, within proceedings involving co-respondents it would be naïve to assume that member states and the EU will provide a united front as co-respondents. The challenges are not insurmountable, but they are present and identified within this proof-oriented undertaking.

The article is structured as follows. Sections 2 and 3 indicate the present pre-accession context, with respectively a general overview of burden of proof at the ECtHR, and EU law in cases before the ECtHR. Section 4 then explains the “rules of the game”, features which must be accommodated for within reconstruction of the puzzle, but selected as relevant to the burden of proof: autonomy of the EU legal system; mutual recognition and mutual trust; and relevant arrangements and contours of the DAA 2023 which meet the requirements. Section 5 pulls together these core aspects, and by identifying issues and the types of relevant cases which might occur, the burden of proof in cases involving EU law post-accession is discussed in detail. Section 6 concludes.

2. Burden of Proof within the ECHR System: A General Overview

The ECtHR’s evidentiary system includes the interlinking points of burden of proof, standard of proof, the types of evidence the ECtHR will consider,[12] and the relative weight it will give to that evidence when deciding a case. It is an area which is relatively underexplored in the academic literature, but has nevertheless received criticism for its lack of clarity and coherence.[13] In her editorial of a recent special issue on the topic Dembour highlights ‘the Court’s “evidentiary system” is highly complicated, often confusing, surprisingly technical, constantly evolving, commonly inconsistent, and repeatedly stacked against the applicant.’[14] In a similar vein, Gunn points out the ECtHR ‘has not articulated with sufficient clarity the rules of evidence that apply to its proceedings, thereby permitting inconsistent and ad hoc evaluations’.[15]

Whilst the focus in this article is the burden of proof specifically, it is important to be alive to these issues and the wider evidentiary dimension. Burden of proof is not an isolated concept. In the simplest terms, the burden of proof indicates who must prove the case before a court. It is slightly more complicated than that however, because the question is not only who must do the proving, but what must they prove; how (i.e. acceptable / admissible evidence); and relatedly, to what degree of certainty must they prove whatever it is they must prove (i.e. standard of proof). Following on from this, there is a question of how a court will use the evidence, including the relative persuasive or evidential weight that will be given to specific evidence.

In the context of the ECtHR, it perhaps helps our understanding of the burden of proof (who must prove what) to work backwards from its standard of proof and how the court has approached it.[16] In summary of certain findings in this author’s previous research, the ECtHR insists it has a single standard of proof – beyond reasonable doubt – but this has been variously disputed by claims that in reality the ECtHR applies differential standards.[17] Ultimately the criticism can be attributed to a difference in framing, - between the ECtHR and its critics, - that hinges on the question of what is being “proven”,[18] a point which is directly relevant to the burden of proof. It is argued that the standard of proof can be conceived of in either broad or narrow terms. Commentators identifying differential standards of proof within the ECtHR’s jurisprudence[19] are invariably looking at a broader range of cases than the ECtHR itself does when it refers to its beyond reasonable doubt standard.[20] For proof narrowly conceived, we are looking at proof either of specific facts or a factual account of circumstances. Broadly conceived, proof relates not only to facts but equally to evidence which tips the balance of persuasion, for example that a particular interpretation is to be favoured, that an appropriate balance between competing interests has been struck, or that a threshold has been met.[21] To distinguish the two, let us call this latter type of proof “persuasive evidence”. This may not immediately strike one as proof in a strict sense, since the mind is generally drawn to facts.

Ultimately however, the ECtHR is looking not for proof of individual facts, but proof that the ECHR right in question has been breached.[22] It is the case (i.e. the rights violation) and not all individual facts which must be proven, and it is this burden of proving with which the present article is concerned. To do otherwise and focus only on proof of fact would remove a considerable number of rights claim types from consideration, explained below.

2.1. What

On the question of what must be evidenced to prove a violation, three types of case can be identified: (i) those which hinge on factual evidence, for example involving disappearances and/or allegations of torture (ii) those which hinge on persuasive evidence informing judicial interpretation, for example the proportionality of restrictions placed on qualified rights, and (iii) proceedings in which both types of proof and evidence are key,[23] for example that a person has suffered harm at the hands of the state and that this meets the threshold of severity to be considered inhuman.

Relating this to post-accession proceedings involving EU law, if it were claimed that a particular point of EU law is not ECHR compliant, this would hinge on persuasive evidence, perhaps even argument, not on proof of fact. It is where specific acts or omissions have occurred that proof of fact is important. Accordingly, taking forwards the wider of the two possible conceptions of proof, which includes persuasive evidence, ensures all types of potential proceedings are considered and is thus the most robust approach for this discussion.

It should also be noted, the admissibility requirement to exhaust domestic remedies and the subsidiarity principle mean the ECtHR should not often have to consider proof of fact. In some circumstances however, such as when the ECtHR has waived this admissibility requirement, or does not believe the domestic courts have fully settled the facts, it does.[24]

2.2. Who

To whom the burden of proving falls is not an entirely fixed point. According to Dembour the ECtHR regards the role of ‘producing evidence’, - by which she appears to mean proof of facts, - as being shared by all involved in proceedings, including itself. Conversely, it is generally for the applicant in a proceedings to ‘persuade the Court of the validity of [their] allegations’.[25] Gunn however, questions that the ECtHR is itself clear whether it is an inquisitorial or an adversarial court, a point which has considerable implications as to who carries the burden of proof.[26] He nevertheless notes the ECtHR’s acknowledgement it is adversarial which confirms the main onus of proving falls to the applicant.

For completeness, though it is unlikely to feature prominently, if ever, in the context of ECHR proceedings involving EU law, the ECtHR will occasionally make rebuttable inferences against the respondent state.[27] This relates invariably to proof of fact and happens generally when the respondent state is not a cooperative partner, such as in disappearance cases, of which state denial is a core feature. The rights issue need not necessarily be such an egregious breach as this. Accordingly, it should be held in mind that the central and driving features behind the ECtHR making rebuttable inferences are that the respondent state seemingly has control of vital information and is being uncooperative, even obstructive in the process. The consequence where inferences are made in this way is that the burden of proof is shifted to the respondent state, which must disprove the presumption against it.[28]

In post-accession proceedings involving EU law, the burden of proof will mostly, if not always, require the applicant to provide either proof of fact or persuasive evidence that the relevant Convention right has indeed been breached. This is explored in detail below in Section 5. First however, it is relevant to look at EU law in pre-accession proceedings before the ECtHR, and following that, the “rules of the game” within which post-accession burden of proof must operate.

3. Pre-Accession: EU Law in Cases Before the ECtHR

As things stand, pre-accession, the EU cannot have proceedings brought against it at the ECtHR.[29] However, because all 27 EU member states are also states parties to the ECHR, they have obligations to satisfy both the ECHR and EU law. Accordingly, situations can arise when the ECtHR is seized of a case against a member state with the rights issue arising out of its implementation of EU law. In such circumstances, the origin of the problem may be with the relevant EU law. Alternatively however, it may not be, if the member state had discretion in the detail of its implementing but did so in a non-ECHR compliant manner.[30] Either way, unless the Bosphorus presumption applies (see below) the member state will be held accountable for any ECHR breach that occurs.[31] Crucially and necessarily the EU cannot be held directly accountable as a party in proceedings, only indirectly as a case may require the ECtHR to lend scrutiny to relevant EU laws. In a similar vein, without having been involved, EU member states cannot be held accountable for the actions of EU institutions which are potentially not ECHR-compliant.[32]

In the existing ECtHR jurisprudence there are two core case types involving EU law: (i) complaints about acts of the EU and its institutions; and (ii) complaints about EU member states implementing EU law. This latter however, has a sub-group whose importance is so considerable it is appropriate to distinguish and discuss it separately from more straight-forward implementation of EU law that does not involve third EU member states. This category of cases relates to complaints involving mandatory cooperation between EU member states arising out of their obligations in EU law:[33] the realm of mutual trust and mutual recognition. Within this discussion proceedings involving one implementing member state only shall be referred to as (ii,a), and those involving mutual trust will appear as (ii,b).

The EU’s accountability currently before the ECtHR – type (i) - was briefly explained above. For both (ii,a) and (ii,b) case types the Bosphorus presumption has been a central feature, and questions of evidence and proof have particular relevance to both. Not least because it remains possible Bosphorus will have continued application in ECtHR case law post-EU accession,[34] the two sub-groups are discussed here in turn.

3.1. Implementing EU Law: Presumption of Equivalent Protection

In 2005, in the key case of Bosphorus v Ireland,[35] the ECtHR offered an approach to balancing up some specific features of the international legal environment that it has followed (more or less) in its case law since.[36] The case is well known and represents a classic (ii,a) type of case involving implementation of EU law. Ireland had an obligation and no choice but to follow direct effect Community law by impounding an aircraft. A challenge followed before the ECtHR was that Article 1 of Protocol No.1 ECHR (protection of property) had been breached.

In answer, the rebuttable Bosphorus presumption that the EU law offers “equivalent protection” of Convention rights[37] was given voice by the ECtHR.[38] It applies when the member state in a case before the ECtHR has acted in compliance with EU law and did ‘no more than implement legal obligations flowing from its membership of the organisation’.[39] Importantly, ‘equivalent’ means ‘comparable’ and not identical,[40] and where applied, the member state will not be found to have breached the ECHR.[41] Rebuttal can occur if ‘in the circumstances … the protection of Convention rights was manifestly deficient.’[42]

The continued relevance of Bosphorus post EU-accession has been mooted. Franklin and Tzevelekos demonstrate its possible continuation, such as in instances where the co-respondent mechanism (explained below) is not accepted.[43] Accordingly, it must be noted that the Bosphorus principle bears relevance to proof: it is a presumption in favour of the respondent state with a limited requirement of evidence, albeit the respondent state and not the applicant may need to demonstrate the presumption applies, for example, that the state did not have discretion when applying EU law.

The presumption invites some considerable scepticism for the ‘potential lack of judicial scrutiny and accountability … [it] may entail. … problematic, particularly with respect to access to justice and effectiveness in the protection of human rights.’[44] It should relatedly be questioned how far the application of presumptions can really be stretched whilst still meeting the ECtHR’s beyond reasonable doubt standard of proof. Whilst certainly that standard applies to the proof of a violation, not its refutation, if presumption is followed too easily, this may potentially place an elevated burden on the applicant to demonstrate ‘manifestly deficient’ rights protection. Aligned to this, the malleability of the term ‘comparable’ means applying the presumption when protection is comparable, not identical, may further increase that burden.[45] It is the applicant who in usual circumstances carries the burden of proof in any event, though crucially the point to be proven is different in these circumstances.[46]

3.2. Cooperation Between EU Member States: Equivalent Protection and Mutual Recognition

The principle of equivalent protection has been applied subsequently in this (ii, b) sub-category of cases involving implementation of EU law.[47] The 2016 case Avotiņš v Latvia[48] is notable because it is here the core EU principles of mutual recognition and mutual trust which underpins it were considered. Alongside the autonomy of the EU’s legal system, these concepts are a fixed and essential feature within the Area of Freedom, Security and Justice.[49] Their nature makes them deeply relevant to issues of proof. The two principles – essentially trust between member states and following on, recognition of their respective legal systems - are inherently connected. At their root, they are enabling concepts towards streamlined and efficient processes in areas where EU member states must cooperate.[50] Relevant to the ECHR, mutual trust means acceptance by EU member states that ‘all EU member states are fully compliant with fundamental rights norms.’[51] Mutual recognition which follows on from it creates ‘automaticity in inter-state cooperation’ and requires ‘uncritical acceptance of presumed mutual trust between – and in – the legal systems of EU member states.’[52] This translates into a requirement, enshrined and expressed variously within relevant EU laws,[53] that member states accept all other member states are themselves compliant with EU law, including the EU Charter of Fundamental Rights (CFR).[54]

Areas in which mutual trust and mutual recognition are applicable include: recognising civil law judgments of other member states, and willingness to execute them as appropriate;[55] executing European Arrest Warrants;[56] and removing asylum-seekers to another member state.[57] Where the principles apply, member states (usually)[58] should not question for example, the accommodation of procedural fairness within another member state’s legal system, or that detention conditions within other member states (eg. prisons, and immigration detention) meet acceptable, humane standards.

Avotiņš was indeed a procedural fairness challenge that Latvia, by recognising the decision of the Cypriot courts, had contravened Article 6 EHCR (fair trial). Applying the Bosphorus presumption, the ECtHR found no breach.[59] It did however, include a qualifying point: mutual recognition must not be applied ‘automatically and mechanically to the detriment of fundamental rights’.[60] This had potential to create tension between the ECtHR and the CJEU not least because of the latter’s insistence that member states may not ‘save in exceptional circumstances’ check another has in fact observed fundamental rights.[61] The friction did not ultimately materialize, partly because the CJEU has itself accepted mutual recognition has limits.[62] Indeed, it has been variously observed that since Opinion 2/13, there has been increased synergy between the ECtHR and the CJEU in relevant areas.[63]

This ‘premise of jurisprudential convergence has been used as the foundation of the [DAA 2023] negotiators’ approach’ to mutual trust which by one account has been not to lend it much attention.[64] It is at least fair to say that within DAA 2023 its presence in two sentences is relatively low key. Whilst Di Franco and de Carvalho suggest (not uncontroversially) that mutual trust was not given prominence in negotiations,[65] a more vital point for the present discussion is that they also challenge the extent of convergence between the two courts. Instead of full convergence Di Franco and de Carvalho indicate two different types of CJEU cases ‘depending on the degree of automaticity given to mutual trust’:[66] cases where the CJEU requires quasi-automatic application of mutual trust; and cases where the CJEU allows more leeway for states potentially to depart from mutual trust. In this way they demonstrate a ‘general tendency for the CJEU to afford differing scope to allowed exceptions to mutual trust, largely influenced by the cooperation scheme at play and the nature of the fundamental right or public interest used as the possible basis for an exception to mutual trust.’[67] Conceivably therefore, tensions may still arise between the two courts post-accession and necessarily, this is returned to in the sections below.

Moreover, where the presumption has been extended into the contexts where mutual trust and mutual recognition apply, the delicate balance of relationships (and not only between the courts) is exposed. Depending on the continuation of Bosphorus/Avotiņš post-accession the impact of this could be considerable. These are contexts in which EU law dictates specific relationships between its member states. Accordingly, this is not only a dynamic within which the ECtHR may be looking for particular proof from the applicant: for example, rebuttal of the presumption by demonstrating ‘manifestly deficient’ protection. The two (or more) member states may quite inevitably have a role in provision of proof and conceivably may end up laying causal blame at the EU’s or each others’ door. These points are picked up again below but evidently, post-accession the ECtHR will be treading a difficult tightrope not to encroach on a further, seemingly fixed, rule of the game: EU autonomy, including in allocating responsibility. With this in mind, it is time to turn to the rules of the game.

4. Rules of the Game: the DAA 2023 and the Vital (Rigid?) Contours of the Puzzle

Whereas it was hoped the DAA 2013 had created the means for the EU to accede to the ECHR, the CJEU set out reasons this was not possible under that agreement. At root, there is a divergence in priorities between the two international organizations. One account suggests the Council of Europe ‘naturally tends towards preserving the effectiveness of the ECHR in relation to all its parties; the EU, however, seems more concerned about safeguarding its unique position.’[68] Accordingly, for accession to be possible ‘the ECHR system must recognise the EU and member states as mutually coordinated actors which often carry out the same tasks but at different levels.’[69] The DAA 2023 has aimed to achieve this, but to continue the metaphor of fitting together the pieces of a three-dimensional puzzle, certain fixed elements must be accommodated for.

The most significant of these is the autonomy of the EU legal system, including that the CJEU only can allocate competencies. In addition, mutual trust and mutual recognition, with which the ECtHR has already had to grapple (above), are a central feature of EU law and relevant potentially to many cases that could arise before the ECtHR post-accession. These are essentially, the “rules of the game”, the rigid or semi-rigid structures to be accommodated. In this section the approach to both in DAA 2023 is elucidated. When questions pertaining to proof are then added to this complicated model, in Section 5, it provides a procedure-oriented test, and hopefully some insight, into how the whole might work in practice.

4.1. Autonomy of the EU Legal System

Without question, autonomy of the EU’s legal system was the driving factor behind Opinion 2/13 in which the CJEU indicated reasons the DAA 2013 was not compatible with EU law.[70] Vitally, the CJEU has been clear that this autonomy must not be undermined by any international agreements made by the EU.[71] As Lock summarizes the position: ‘International agreements must not a) give an international court the power to interpret EU law in a binding manner …b) an international court may not decide on the allocation of competences between the EU and its Member States … c) the agreement must not alter the essential character of the powers of the EU’s institutions’.[72] A core means of meeting this is the co-respondent mechanism (discussed below) though there are further points of accommodation within DAA 2023.[73] These include that ‘insofar as a dispute between them concerns the interpretation or application of European Union law’ the EU and its member states cannot bring inter-party proceedings against each other.[74] Furthermore, for states which have ratified ECHR Protocol 16, Advisory Opinions cannot be requested on issues pertaining to EU law if that law ‘as interpreted by the CJEU’ requires the member state to seek a preliminary ruling from the CJEU.[75]

4.2. Mutual Recognition and Mutual Trust under DAA 2023

Mutual recognition and mutual trust are themselves elements of the more encompassing notion of EU autonomy. Respecting the principles is a requirement and core feature of certain aspects of EU law. They allow ‘an area without internal borders to be created and maintained within the EU’[76] and they certainly speak to competencies: for example, state A must not question state B’s compliance with fundamental rights, and must execute a judgment made by a court in state B.[77]

Although Di Franco and de Caravalho suggest otherwise (above), Gragl indicates that mutual trust was ‘among the most difficult issues for the negotiators’.[78] Lock further highlights that accommodating mutual trust within the DAA 2023 was complicated for at least two reasons: the concept is not fixed, but ‘nuanced and developing’; and according to Opinion 2/13, the DAA should not require member states to ensure each others’ compliance with fundamental rights.[79] The solution provided in DAA 2023 is to include a new provision, Article 6, which reads:

‘Accession of the European Union to the Convention shall not affect the application of the principle of mutual trust within the European Union. In this context, the protection of human rights guaranteed by the Convention shall be ensured.’

This of course ‘contributes little’,[80] the ECtHR will continue to scrutinize mutual trust at the risk of interfering with EU autonomy. Elucidating some of the drafting history however, Di Franco and de Caravalho make an important point. The original re-draft (post-Opinion 2/13) formulation of Art.6 DAA had included reference to the Avotiņš requirement: that mutual trust should not be implemented automatically or mechanically.[81] Interestingly however, this disappeared under pressure from the EU negotiators. With reference to both the provision and the explanatory report, Gragl interprets Article 6 of DAA 2023 as meaning – via the first sentence – that EU member states must observe and uphold their obligations under mutual trust as an aspect of their EU membership. At the same time, the ECtHR will not consider whether or not mutual trust applies in a particular case, because to do so would ‘amount to an interpretation of Union law.’[82] This latter point is not stated expressly in the explanatory report but in view of the wider context of Opinion 2/13, it is a reasonable summation. Gragl further highlights, it seems correctly, that the second sentence ‘requires the Union member states to equally ensure the protection of human rights as guaranteed by the Convention when applying the principle of mutual trust.’[83] The ECtHR will nevertheless ‘continue to verify that the principle of mutual trust is not applied automatically and mechanically to the detriment of human rights’.[84]

The second sentence of Article 6 is supported by the following observation within the explanatory report:

‘The Court [ECtHR] has noted the increased convergence between its own case law and the case law of the CJEU with regard to the limits to the operation of mutual recognition-mechanisms in light of a real and individual risk of a violation of Article 3 of the Convention.’

Though noting the ECtHR’s accommodation of mutual trust within Avotiņš, the explanatory report does not speak to convergence between the two courts on anything wider than Article 3 ECHR. At this point, the observations of Di Franco and de Caravalho become important because it seems there are two options. On one hand one can accept the ‘the premise of full jurisprudential convergence between the two courts’ in which case it means accepting that the CJEU and ECtHR both recognize the non-automaticity of mutual trust and apply the same standard in this regard.[85] Alternatively, as Di Franco and de Caravalho do, one could question the extent of judicial convergence. In which case

‘the rewriting of art. 6 turns into a largely formal recognition of EU mutual trust standards, with the ECHR standard of non-automaticity relegated to the Explanatory Report. While recognising the need to protect ECHR rights, this last formulation does not provide any concrete legal consequences nor guidance as to how to navigate potential tensions between EU mutual trust-based schemes and ECHR obligations.’[86]

It is these points particularly with which we must grapple in Section 5 as relevant to proof: how are those tensions to be navigated? And specifically where and how does proof fit within this? Before doing so however, a final piece of the puzzle, a reshaped version of the co-respondent mechanism, as an accommodation of autonomy, must be explained.

4.3. Co-respondent Mechanism and the “Veil” of Joint Responsibility

The co-respondent mechanism existed in DAA 2013 but is in a changed form in DAA 2023. In short, Article 3 DAA 2023 makes it possible for the EU and member states to become co-respondents in a case before the ECtHR. This can happen if proceedings are brought against a member state and the EU joins as a co-respondent,[87] or conversely, in proceedings against the EU with the member state joining.[88] Where proceedings are brought against both, the parties’ status can also be shifted to become co-respondents.[89]

The scope of cases to which the co-respondent mechanism would apply is still not entirely certain. In particular it is unclear whether it is limited to cases where the member state did not have discretion in its implementation of EU law, or captures a wider body of content.[90] The key points which can be stated with greater certainty are that, assuming all requirements are satisfied, the co-respondent mechanism can occur either by the EU or a member state accepting an invitation from the ECtHR, or on its own initiative.[91] Ultimately, the EU has control of the process since the mechanism can only be activated following its ‘reasoned assessment’.[92] Co-respondents are parties in the proceedings and as such, have the same procedural rights. If the EU is not a co-respondent, the relevant member state would still be responsible for any ECHR violation including as flowing from its EU membership.[93] However, if a violation is found when the EU is a co-respondent in the case, the co-respondents will have joint responsibility for it.[94]

Joint responsibility avoids the obvious issue – a key objection of the CJEU in Opinion 2/13 – that the ECtHR could end up allocating responsibility. This is however, a framework set out in a text isolated and abstracted from the reality of what might occur within specific proceedings, fleshed out and enlivened with the detail of facts. Joint responsibility of course avoids the allocation of responsibility in law insofar as the outcome is concerned. Within case specifics however, and necessarily, the detail may lead fairly unavoidably to 'attribution of conduct’.[95] Writing in 2016, long before the DAA 2023 was concluded, Tacik made some significant observations about this. Whilst joint responsibility ‘would produce a complete organisational veil which shields individual parties from attribution (and execution) of responsibility separately’ it is problematic.[96]

‘[T]his kind of responsibility would be extended both to the assessment of premises of finding a violation and to execution of judgments. As far as the former is concerned, the ECtHR would not be entitled to first establishing [sic] violations of each of the parties and, subsequently, to produce a “sum of individual responsibilities”. What the Strasbourg Court would be entitled to do would be determining the act and its relation to the Convention. Attribution of conduct, however, would have to be supplemented by the request of both (or more) parties to trigger the CRM [co-respondent mechanism].’[97]

Under the DAA 2023 this appears to remain the case. When the practical point of proving the case comes into play, it is impossible not to look at the detail of what precise violation(s) are being claimed and identification of cause at this point – in the first instance by the applicant – is inescapable. The origin of the rights issue must be ascertained if an appropriate remedy is to be provided. Moreover, the dynamic itself is difficult and it can be questioned how far the EU or its member states can or would refute their own responsibility, at least in relation to the case facts. Observers of the ECtHR jurisprudence will inevitably reach their own conclusions about cause when reading judgments. As Tacik aptly notes, the co-respondent mechanism ‘blurs the distinction between attribution of conduct and of responsibility.’[98] It not only allows co-respondents to be jointly responsible but in doing so ‘breaks the chain between conduct, obligation and responsibility. It focuses on the result that violates the Convention, but abstracts from causal links that relate it to acts and omissions of the EU and its Member State(s).’[99] It is submitted and argued further below that via joint-responsibility, a legal fiction is produced.

5. Burden of Proof in Post-Accession Proceedings Involving EU Law

The above discussion necessarily explains burden of proof and the context in which, in proceedings involving EU law, it would operate post-accession. The burden of proof refers to both the who and the what questions of proving a case. It can be narrowly conceived as referring to proof of fact, or more broadly as including persuasive evidence of a particular position, for example proving an ECHR violation by demonstrating restrictions on a qualified right are disproportionate. This latter, persuasive dimension is far more likely to feature in cases involving EU law and so the wider interpretation is applied in the discussion. The burden of proof generally falls to the applicant, with the exception of cases in which the respondent state is proving uncooperative.

As noted in Section 3, once the EU accedes to the ECtHR, there are two core case types which may occur, with sub-categories within the second. These are: proceedings (i) against the EU because of the conduct of its institutions, including their underpinning in law; (ii,a) against a member state for its implementation of EU law, but without the involvement of a third EU member state; and (ii,b) against member states arising out of their EU mandated cooperation with each other. In all three scenarios the co-respondent mechanism may apply, subject to the case meeting the relevant criteria and the EU providing a reasoned assessment that it should.[100] If the co-respondent mechanism is not invoked, it is possible the Bosphorus presumption would still be used. The presumption in favour of a member state carries the conceivable consequence of elevating the burden of proof on the applicant in making a rebuttal (discussed below).

What is abundantly clear is the delicate balance of relationships within the context: between EU and ECtHR; between the EU and its member states, as well as the member states with each other; and between all of these and the applicant(s) in proceedings. In what follows, the types of case noted above are considered in turn with specific reference to burden of proof in the context. As this is done, two fixed points of EU law are tested: autonomy as regards allocation of responsibility, and the limits of mutual trust. On the first, though the co-respondent mechanism accords joint responsibility for any breach, proof is a practical dimension in which causative attribution is unavoidable. Based on the proof presented the wrong will be evident, belying the legal fiction. On the second, the relegation to the explanatory report of the current rule that mutual trust must not be automatically and mechanically applied, and the contested point of convergence between the two courts must be borne in mind.

5.1. Complaints about Conduct of the EU and its Institutions

In pre-accession case law, the two key examples of rights issues arising out of the actions of EU institutions are Connolly[101] and Andreasen,[102] both of which essentially involved employment rights. Post-accession it is conceivable the EU would strengthen its internal processes, knowing the risk of proceedings before the ECtHR otherwise. The burden of proof to demonstrate wrongs on the part of EU institutions would, fairly uncontroversially, fall to the applicant.

There is, however, a point with some greater complexity. In much of the literature on EU accession to the ECHR it appears as a given that the EU’s wrongs, insofar as ECHR compliance is concerned, would fall almost exclusively with its law: law with which member states must then comply.[103] The EU’s Border and Coast Guard Agency (Frontex) however, nuances this. For example, in 2020 NGOs and the media reported Frontex’s direct involvement in pushback operations alongside member states.[104] Frontex is now subject to European Parliamentary scrutiny,[105] has distanced itself from the matter,[106] and its regulations set out clearly Frontex must ‘comply with EU and international law’ including the law against refoulement and collective expulsions.[107] There have however been several (as yet unsuccessful) challenges before the EU courts seeking damages for Frontex’s actions.[108]

Accordingly, and though this article is concerned primarily with ECtHR proceedings involving EU law, it is clear that not only EU laws but also the actions of EU institutions may potentially raise issues with ECHR compliance. Looking towards laws which may underpin such actions, some of the EU’s immigration and asylum policies and agreements are noteworthy. For example, regarding ongoing talks about a status agreement on operational activities by Frontex in Senegal, the Rapporteur to the EU Committee on Civil Liberties, Justice and Home Affairs flagged some major concerns. In particular their report noted the risk of pushbacks and foresaw the agreement

‘providing for the exercise of executive powers by Frontex in the country would entail a high risk for Frontex of becoming complicit in direct and indirect violations of fundamental rights or international protection obligations; believes that these risks are of a serious nature and are likely to persist’.[109]

The report proposed Parliament should make a recommendation to the Commission based on its concerns, but this was rejected.[110] Conceivably an agreement such as this could lead to challenge before the ECtHR, with EU-level detail such as this offered as evidence.

Proof depends heavily on the context, including the respondent(s)’ cooperation in the process. Moreover, and perhaps because of its very high standard of proof, the ECtHR triangulates evidence from a variety of sources to reach its conclusions.[111] For example, in Ilias and Ahmed v Hungary[112] on the allegation of inhuman treatment, the ECtHR took into account the facts as indicated by the applicants and reports of the European Committee for the Prevention of Torture (CPT), a Special Representative of the Secretary General of the Council of Europe, and a psychiatrist.[113] In this way proof was resolved, the Article 3 ECHR threshold of severity was not met.

In pushbacks, not all the information might be available, meaning considerable difficulty for an applicant to meet the burden of proof.[114] Accordingly, Baranowska argues the burden of proof in pushback operations should be approached in the same way as it is in disappearances: where the applicant brings prima facie evidence of the alleged violation(s) the burden should be reversed.[115] Where the underlying (EU) law may be the cause, this is far more accessible to the ECtHR to make its evaluation in the usual manner. The question of the EU legal order’s autonomy is however inescapable and Frontex operations may at some point come under ECtHR scrutiny.

5.2. EU Law and its Implementation by Member States

Remaining with asylum and immigration, there is a demonstrated causal relationship between the Common European Asylum System and systemic human rights violations.[116] At one point there was a considerable difference between the two European courts regarding the need to demonstrate ‘systemic deficiencies’ in a member state before the presumption of safety could be rebutted.[117] This has been addressed,[118] but other points of variance remain. Notably, the EU-Türkiye statement and action plan, which outsources refugee protection to Türkiye has been raised with both the EU General Court[119] and the CJEU,[120] to consider its compatibility with EU fundamental rights. The former court rejected this as non-justiciable because the agreement was authored by member states and not by EU institutions, the latter found the matter inadmissible.[121] Unless this and other similar agreements[122] are rigorously considered by the EU pre-accession, proceedings may arise at the ECtHR and assuming the involvement of member states, the co-respondent mechanism may be invoked.

For the applicant this means demonstrating issues with the underpinning law, which the ECtHR will see for itself. It means also demonstrating, as relevant, any factual points. Where the co-respondent mechanism applies, both the EU and member state(s) have equal rights as respondents. There is implicit in the DAA 2023 an assumption of trust between the two. However, at the point of proving cause, - whether law, its implementation, or both, - it is not inconceivable that in response the member state and the EU might each respectively attempt to pin blame on the other. For example, they may argue oppositely the member state did or did not have freedom in the manner of its implementing EU law. Such arguments would have no consequence in the outcome in the sense there is joint responsibility for any violation. Wrongs from the ECHR perspective must nevertheless be identified within the case, without which the case cannot be found proven, and appropriate remedies cannot be indicated. Frictions may therefore result not only in the relationship between the two courts, but also between the EU and its member state(s), exposed in part by the attempt of proving and in reply, disproving allegations.

An area where this would potentially become an issue in the future is with regulation and restrictions on the use of artificial intelligence. Two concrete and current examples of when this may occur relate to mutual trust under respectively the Dublin III Regulation and the European Arrest Warrant scheme.

5.3. Member States’ Cooperation: The Mutual Trust Dimension

When member states applying EU requirements of mutual trust are brought before the ECtHR, there is an additional dimension that the member states may contest against each other. Potentially, and depending on the matter, the EU, which would have been informed of an application,[123] could resolve the issue before a case would get that far. It is nevertheless worth noting as an example that, in the determination of which state should assess individual asylum claims, member states have applied Dublin III unevenly, with no common standards.[124] The applicant would face the task of demonstrating unacceptable treatment amounting to an ECHR violation, and the member states and (potentially) the EU would end up putting forwards their own respective sides. It would be strange, naïve even, to expect a united front always across the respondent parties. One implication of this would be inadvertently to place pressure on EU cohesion. At the same time within proceedings, allocation of causal responsibility could be presented not only from the side of the applicant, but would be suggested conceivably also by responding member states and the EU. If allocation of responsibility is the sacred and untouchable EU prerequisite that necessitates joint-responsibility for any ECHR violations, it is within proceedings themselves, as cause and fault are considered and offered proof, that the joint-responsibility fiction would stand to be exposed.

Under the European Arrest Warrant (EAW) scheme[125] mutual recognition – based on mutual trust – is required between member states to secure the transfer of a crime suspect from one jurisdiction to another.[126] This is an area in which, via a series of cases, the CJEU brought its approach to mutual trust more in line with expectations under the ECHR.[127] Accepting that detention conditions in prison and remand centres in certain member states fall far short of the standard required of Article 3 ECHR and corresponding Article 4 CFR (the prohibition of torture and inhuman or degrading treatment or punishment), in 2016 a two stage test was developed in Aranyosi.[128] This is used to determine whether the state asked to surrender a person (“executing state”) can indeed transfer a person to the state issuing the arrest warrant (“issuing state”) and applies when the executing state has evidence of a “real risk” of inhuman treatment to people detained in the issuing state. In such circumstances the executing state must first ascertain the risk to the general prison population, following on, the specific risk to the individual must be assessed.

The series of cases related to this underpin the reference in the DAA 2023 explanatory report to confluence between the two European courts.[129] Indeed, the explanatory report indicates the cases establishing the two stage test with its ‘exceptional circumstances’ when one member state is entitled not to assume another complies with ‘fundamental rights’.[130] The Avotiņš qualifier is also noted. However, apart from Avotiņš, all of the case examples at this point in the explanatory report relate directly to EAWs and Article 3 ECHR/Article 4 CFR.

Judicial confluence between the two European courts has been challenged (see above). Part of this lies in Di Franco and de Cahvalho’s identification that the CJEU takes a variable approach to mutual trust, apparently expecting in some contexts ‘quasi-absolute application of mutual trust’[131] but in others, more ‘leeway’ to depart from mutual trust.[132] Di Franco and de Cahvalho find it problematic that the Avotiņš qualifier is buried in the explanatory report because there is no guidance on either how to ‘navigate’ tensions between mutual trust and ECHR requirements or the legal consequences.

Drilling deeper into the detail however, even within the Article 3 ECHR dimension, there may still be some variance between the two courts. Aranyosi and cases which follow it provide an ECHR accommodation, in that mutual trust cannot always and blindly be observed. It is argued however, that in this context even the CJEU’s approach nevertheless represents a drive to the minimum standard: the risk ‘specifically and precisely’ to the individual, requires the executing state to consider only conditions in facilities where, according to the issuing state, the person ‘might be held’;[133] and without specific indications a particular detention centre presents a risk, diplomatic assurances from the issuing state must be relied on.[134] Moreover, in their reasoning in ML the CJEU balanced the prohibition of inhuman treatment against mutual recognition and efficacy considerations.[135] Needless to say, proportionality has no place when absolute rights are involved.

The key question for this paper is what this means for the burden of proof. The applicant may need to prove their case overall, but fundamentally, they are not relied upon to provide all relevant evidence and indeed, they do not carry the burden of proof in all matters. In light of the above it is not inconceivable an applicant would seek to argue against equivalence between EU human rights protections, including through its jurisprudence, and those under the ECHR. ‘Equivalent’ is already an unhelpfully vague term, but such argument could be raised to prove or demonstrate the requirement for a higher level of protection than that provided, and hence, an ECHR breach. The consequences of such argument could be considerable, and perhaps the best protection against such an eventuality would be for the EU to undertake a wholesale review of its laws and approaches in the jurisprudence for their ECHR compliance.

When factual proof is required, time influences what must be proven. In the non-refoulement context before a transfer has occurred, persuasive evidence of risk (albeit supported by factual evidence) is required. Crucially, in both contexts of asylum and EAW, the burden is on the executing state from which a person would be returned or extradited, to evaluate the risk to them.[136] The state needs to prove that an appropriately robust risk assessment was made and that the evidence underpinning it was itself robust. In the asylum context, the Council of Europe has indicated the burden of proof ‘should not switch to the applicant to prove that a country is unsafe’ and they must have opportunity to rebut any presumptions of safety.[137] Moreover, since the ECtHR triangulates evidence from a range of sources it is relevant to highlight that under the EAW scheme, evidence from the issuing state that risk is absent would not be sufficient on its own to satisfy the ECtHR. If a person has been transferred already to another member state, it becomes more likely that proof of harm or suffering would be needed. Whilst the applicant may carry the burden of proof, this is not a given, since the burden can be reversed if appropriate.

6. Conclusions

The intention of this article has been to examine how the burden of proof in ECtHR proceedings involving EU law after the EU’s accession to the ECHR. This was done against the expectations set out in DAA 2023. It is an especially complex area, and the EU’s accession can be likened to a three-dimensional puzzle within which a series of delicate relationships are held in the balance between different interlocutors and particular rules must be observed. Moreover, the ECtHR’s system of evidence is complicated. Although the burden of proof falls generally to the applicant, the reality is that the ECtHR triangulates and takes into account all relevant evidence available to it, and where necessary it will reverse the burden of proof. There is a distinction between proving fact or providing persuasive evidence in support of a particular interpretation, balance or threshold. A broad understanding of proof, which includes persuasive evidence has been used in this work, but where the facts are before it the ECtHR can draw its own conclusions.

What is no doubt apparent is that considering burden of proof in this context is not really an evaluation of proof in isolation or for its own sake. Considering burden of proof in this context offers an academic procedure-oriented stress-test of relevant areas of the puzzle as it has been reconstructed by in the DAA 2023. Differences between the two systems, and potential challenges and tensions have been made visible. As far as the applicant’s role in proving is concerned, as the rights issue dictates, they may take aim at the wider EU system, specific EU law, and/or application of that law. The EU, notably via Frontex, has agency and conceivably factual proof may be brought, potentially even presumed, against it. The EU should therefore be attentive to aspects within its laws and legal agreements, and subsequent protections at EU level, which might be vulnerable to challenge for non-compliance with the ECHR.

The co-respondent mechanism creates joint responsibility for co-respondents where a breach is established, but the reality is that proving a case means demonstrating cause. Joint responsibility is thus a legal fiction relevant only to the stated outcome of a case. Allocating wrongs within judgments is logically inevitable as it is necessary. Moreover, it is not a given that co-respondents would provide a united front in any proceedings as they refute any allegations. This may produce tensions between member states and the EU, and within proceedings may also see these parties themselves allocating causal responsibility for wrongs. In addition, differences in approach to proof between the two courts have been revealed in this discussion. In sum, the context is both complicated and delicate and the challenges are considerable. Nevertheless, with a mindset on both sides oriented to cooperation and full and meaningful protection of rights, of which the DAA 2023 itself offers some reassurance, there is both vision and hope. With the right commitment, none of the points raised herein should be insurmountable.

[*] Senior Lecturer in Public International Law, Law School, University of Exeter. I would like to thank Vassilis Tzevelekos for his helpful comments when this work was in draft, also Vassilis Pergantis and Paris Arvanitakis for enabling an earlier presentation of this work. All errors are my own.

[1] Consolidated Version of the Treaty on the European Union [2012] OJ C326/13, Article 6, together with the Treaty of Lisbon Amending the Treaty on the European Union and the Treaty Establishing the European Community [2007] OJ C307/1, Protocol No.8.

[2] Tobias Lock, ‘Implications of the Revised Draft EU Accession Agreement for the ECHR’ (2025) 6 European Convention on Human Rights Law Review, 65-101, 67-68.

[3] Steering Committee for Human Rights, CDDH Ad Hoc Negotiation Group (46+1) on the Accession of the European Union to the European Convention on Human Rights, 46+1 (2023) 35 FINAL, 30 March 2023. For discussion see: Christos Giakoumopoulos and David Milner, ‘Accession of the European Union to the European Convention on Human Rights: A View From Inside the Council of Europe’ (2025) 6 European Convention on Human Rights Law Review,15-38.

[4] Below.

[5] Opinion 2/13 Accession of the European Union to the European Convention for the Protection of Human Rights and Fundamental Freedoms [2014] ECLI:EU:C:2014:254. For discussion of objections raised in Opinion 2/13 matched against the DAA 2023 see: Paul Gragl, ‘The New Draft Agreement on the EU Accession to the ECHR: Overcoming Luxembourg’s Threshold’ (2025) 6 European Convention on Human Rights Law Review, 39-64.

[6] Discussed below. See Demi-Lee Franklin and Vassilis P Tzevelekos, ‘The 2023 Draft Agreement on the EU Accession to the ECHR: possible “gaps” and “cracks’ in the co-respondent mechanism and the implications for the Bosphorus doctrine’ (2024) 9 European Papers, 745-768.

[7] It is generally accepted that the CJEU will be called on to give an opinion on the DAA 2023 in light of EU law, Lock (n 2).

[8] DAA 2023, Article 3(5).

[9] DAA 2023, Article 3(8)

[10] Discussed below, Section 5.

[11] It is generally understood the applicant carries the burden of proof in ECtHR proceedings which would mean nothing changes. Crucial nuance is explained in Section 2.

[12] The term “admissible” has been purposely avoided here. There is no clear rule that particular types of evidence would be admissible or not: circumstantial evidence is very much taken into account when the ECtHR deems it relevant to do so.

[13] Marie-Bénédicte Dembour, ‘The Evidentiary System of the European Court of Human Rights in Critical Perspective’ (2023) 4 European Convention on Human Rights Law Review, 363-374; T. Jeremy Gunn, ‘Limitations Clauses, Evidence, and the Burden of Proof in the European Court of Human Rights’, (2020) 15 Religion and Human Rights, 192-206.

[14] Dembour, ibid 373-4

[15] Gunn (n 13), 192.

[16] For more detailed consideration of the point see: Christine Bicknell, ‘Uncertain Certainty?: Making Sense of the European Court of Human Rights’ Standard of Proof, (2019) 8 International Human Rights Law Review, 155-187.

[17] In particular and cited ibid. Matthew Smith, “The adjudicatory fact-finding tools of the European Court of Human Rights’ (2009) European Human Rights Law Review 206-228; Mónika Ambrus, ‘The European Court of Human Rights and Standards of Proof: An Evidentiary Approach towards the Margin of Appreciation’ in Lucasz Gruszczynski and Wouter Werner (eds), Deference in International Courts and Tribunals (OUP 2014), Chapter 13.

[18] Bicknell (n 16), 160.

[19] Above (n 17).

[20] See Joseph Finnerty, ‘When is a State’s Hidden Agenda’ Proven? The Role of the Merabishvili’s Three-Legged Evidentiary Test in the Article 18 Strasbourg Case Law’ (2023) 4 European Convention on Human Rights Law Review, 447-472, 450-451.

[21] To this a temporal dimension can be added. Proof can look into the past where it must be proven something happened, or more speculatively to the future, such as the necessary evaluation risk in non-refoulement cases. Ibid, 166.

[22] Bicknell (n 16), 186. Ireland v the United Kingdom Application no 5310/71 (18 January 1978), [161].

[23] Bicknell ibid; Finnerty (n 20).

[24] Dembour (n 13) 363-4; Bicknell (n 16).

[25] Dembour ibid 364. Italics added.

[26] Gunn (n 13) 202-4.

[27] Bicknell (n 19).

[28] Ibid, 177.

[29] A straight-forward application of general international law since the EU has not acceded to the ECHR. See also Confédération Française Démocratique du Travail v the European Communities, [ECHR] Commission decision, (10 July 1978).

[30] ECtHR examples when this occurred: Cantioni v France Application no. 17862/91, (11 November 1996); O’Sullivan McCarthy Mussel Development Ltd. v Ireland, Application no. 44460/16, (7 June 2018).

[31] Matthews v United Kingdom, Application 24833/94, [GC] (18 February 1999); Cantioni ibid; Michaud v France, Application no.12323/11, (6 December 2012).

[32] Connolly v 15 Member States of the EU, Application no 73274/01, admissibility decision, (9 December 2008); Andreasen v 27 Member States of the EU, Application no 28227/11, admissibility decision, (31 March 2015).

[33] This categorization, numbered as and identified as three distinct types of case, is applied by Rick Lawson, ‘Atlas Shrugged: an analysis of the ECtHR case law involving issues of EU law since Opinion 2/13’ (2024) 9(2) European Papers, 647-671.

[34] Franklin and Tzevelekos (n 6).

[35] Bosphorus Hava Yollari Turizm ve Ticaret Anonim Sirketi v Ireland, [GC] Application no.45036/98, (30 June 2005).

[36] Demi-Lee Franklin and Vassilis Tzevelekos, ‘The ECtHR Bosphorus Doctrine in cases calling for indirect scrutiny of EU law: Judicial Smoke Signals?’ in Vassilis Pergantis (ed), EU Responsibility in the International Legal Order, vol B (Sakkoulas Publications 2023) 35-70.

[37] Bosphorus (n 35), paras. 148–156.

[38] The Commission in 1990 had previously made a presumption that the European Communities offer equivalent protection of ECHR rights, leading to its inadmissibility decision in M & Co. v Federal Republic of Germany, Application no.13258/87, Commission decision, (9 January 1990).

[39] Ibid, para. 156. Where the state has discretion in its implementation of EU law, this can negate the application of the Bosphorus principle.

[40] Ibid, para. 155.

[41] Ibid, para. 156.

[42] Ibid, para. 156.

[43] Franklin and Tzevelekos (n 6).

[44] Franklin and Tzevelekos (n 6), 767.

[45] See Section 5.

[46] Ibid.

[47] For example, Bivalaru and Moldovan v France, Applications nos. 40324/16 and 12623/17, (25 March 2021).

[48] Avotiņš v Latvia [GC], Application no. 17502/07, (23 May 2016).

[49] For discussion relevant to DAA 2023 see Lock (n 2); see also Valsamis Mitsilegas, ‘Mutual Recognition and Mutual Trust in Europe’s Area of Criminal Justice: The Centrality of Fundamental Rights’ in V. Mitsilegas, A. di Martino and L. Mancano (eds), The Court of Justice and European Criminal Law: Leading Cases in a Contextual Analysis (London: Bloomsbury, 2019).

[50] Christine Bicknell, ‘Same tools used differently: the preventive potential of the European arrest warrant against inhuman treatment’ (2020) European Human Rights Law Review, 624-639, 626.

[51] Mitsilegas, (n 49), 125.

[52] Ibid.

[53] Examples below.

[54] Bicknell (n 50).

[55] Regulation (EU) No 1215/2012 of the European Parliament and of the Council of 12 December 2012 (Brussels II).

[56] Council Framework Decision on the European Arrest Warrant and the surrender procedures between Member States of 13 June 2002, and its amending 2009 Council Framework Decision.

[57] Regulation (EU) No 604/2013 of the European Parliament and of the Council of 26 June 2013 (Dublin III).

[58] Slight exceptions are discussed below and follow on from CJEU, Joined Cases C-404//15 and C-659/15 PPU, Pal Aranyosi and Robert Cãldãrau v Generalstaatanwaltschaft Bremen, ECLI:EU:C:2016:198, Grand Chamber (5 April 2016).

[59] Avotiņš has been followed further eg. Romeo Castano v Belgium, Application No. 8351/17 (9 July 2019); Bivolaru and Moldovan (n 47).

[60] Avotiņš (n 48) para. 116.

[61] See Lawson (n 33), 660-1; Avotiņš ibid, para. 114.

[62] Lawson, ibid, citing Aranyosi (n 58).

[63] Lawson ibid. notes this as a particular consequence of the issues with Hungary and Poland. The DAA 2023 explanatory report itself picks this up, but in a limited area (Art.3 ECHR). See discussion below.

[64] Elenora Di Franco and Mateus Correia de Carvalho, ‘Mutual Trust and EU Accession to the ECHR: are we over the Opinion 2/13 hurdle?’ (2023) 8 European Papers, 1221-1233.

[65] Ibid.

[66] Ibid, 1226.

[67] Ibid,1227-8.

[68] Przemyslaw Tacik, ‘Attribution of Responsibility after the EU Accession to the ECHR and the “Co-Respondent Mechanism”’ (2016) 16 Baltic Yearbook on International Law, 29-54, 34.

[69] Ibid, 38.

[70] Among abundant relevant literature see Christina Eckes, ‘EU Accession to the ECHR: Between Autonomy and Adaptation’ (2013) 76(2) Modern Law Review, 254-285; Daniel Halberstan, ‘ “It’s the Autonomy, Stupid!” A Modest Defense of Opinion 2/13 on EU Accession to the ECHR, and the Way Forward’ (2015) 16(1) German Law Journal, 105-146; more recently Lock (n 2).

[71] Opinion 1/91 Agreement on the European Economic Area ECLI:EU:C:2019:341, para 118. See references ibid.

[72] Lock (n 2), 71.

[73] Discussed inter alia Lock ibid.; Gragl (n 5); Lawson (n 33).

[74] DAA 2023, Art. 4(3) and Explanatory Report, paras 82-83.

[75] DAA 2023, Art.5. Eloquently explained in Gragl (n 5),15.

[76] DAA 2023, Explanatory Report, para 87.

[77] This was the situation in Avotiņš (n 48).

[78] Gragl (n 5), 52-54.

[79] Lock (n 2), 89.

[80] Gragl (n 5), 56.

[81] Di Franco and de Carvalho (n 64).

[82] Gragl (n 5), 56-57.

[83] Gragl, ibid. 57.

[84] Ibid.

[85] Di Franco and de Caravalho (n 64), 1225.

[86] Ibid.

[87] DAA 2023, Article 3(2).

[88] DAA 2023, Article 3(3).

[89] DAA 2023, Article 3(4).

[90] Franklin and Tzevelekos (n 6), 755.

[91] DAA 2023, Article 3(5).

[92] Ibid.

[93] DAA 2023, Article 1(4).

[94] DAA 2023, Article 3(8).

[95] Tacik (n 68).

[96] Ibid, 50.

[97] Ibid.

[98] Ibid, 53.

[99] Ibid, 53.

[100] DAA 2023, Article 3(5).

[101] Connolly (n 32).

[102] Andreasen (n 32).

[103] More realistically Tacik notes the EU’s ‘scarce capabilities of using its own institutions and forces’ (n 68), 46.

[104] Lighthouse Reports, Le Monde et al. ‘Frontex Complicit in Pushbacks’, 23 October 2020, [https://www.lighthousereports.com/investigation/frontex-chapter-ii-complicit-in-pushbacks/] accessed 28 August 2025.

[105] Regulation (EU) 2019/1896. See Michaela Del Monte and Katrien Luyten, European Parliamentary Research Service, European Parliament Scrutiny of Frontex, PE 698.816 – November 2023 [https://www.europarl.europa.eu/thinktank/en/document/EPRS_BRI(2021)698816] accessed 28 August 2025.

[106] Ibid, 6.

[107] Ibid.

[108] Del Monte and Luyten (n 105).

[109] European Parliament, Committee on Civil Liberties, Justice and Home Affairs Report, A9-0032/2024, 7 February 2024, [https://www.europarl.europa.eu/doceo/document/A-9-2024-0032_EN.html] para.2, accessed 28 August 2025.

[110] 2023/2086(INI), Report on the ongoing negotiations on a status agreement on operational activities carried out by the European Border and Coast Guard Agency (Frontex) in Senegal, https://oeil.secure.europarl.europa.eu/oeil/en/procedure-file?reference=2023/2086(INI) accessed 28 August 2025.

[111] Bicknell (n 16).

[112] Ilias and Ahmed v. Hungary [GC], Application no. 47287/15, (21 November 2019).

[113] Ibid, paras 186-194.

[114] Two recent pushback cases in which the government denied all allegations exemplify this. A.R.E. v Greece, Application no.15783/21 (1 January 2025); G.R.J. v Greece, Application no. 15067/21 (3 December 2024).

[115] Grażyna Baranowska, ‘Exposing Covert Border Enforcement: Why Failing to Shift the Burden of Proof in Pushback Cases is Wrong’ (2023) 4 European Convention on Human Rights Law Review, 473-494.

[116] Evangelia Tsourdi and Cathryn Costello, “Systemic Violations” in EU Asylum Law: Cover or Catalyst?’ (2023) 24 German Law Journal, 982-994.

[117] Ibid, 985-988.

[118] Ibid.

[119] Case T-192/16, NF v European Couuncil, ECLI:EU:T:2017:128 (28 February 2017); Case T-193/16, NG v European Council, ECLI:EU:T:2017:129 (28 February 2017); Case T-257/16 NM v European Council, ECLI:EU:T:2017:130 (28 February 2017).

[120] CJEU, Joined Cases C-208/17 P to C-210/17 P, NF v European Council, ECLI:EU:C;2018;705 (12 September 2018).

[121] See Tsourdi and Costello (n 116), 993.

[122] For example, law, policy and its implementation arising out of the EU and Mauritania agreement: Joint Declaration establishing a Migration Partnership between the Islamic Republic of Mauritania and the European Union, 7 March 2024.

[123] DAA. 2023, article 3(2).

[124] Rachel Garrett and Nicole Barrett, ‘Dublin III in Practice: Synthesizing a Framework for European Non-Refoulement Cases at the Human Rights Committee’ (2021) Journal of Human Rights Practice, 250-269.

[125] Council Framework Decision on the European Arrest Warrant and the surrender procedures between Member States of 13 June 2002, and its amending 2009 Council Framework Decision.

[126] Bicknell (n 50).

[127] Aranyosi and Cãldãrau (n 58); Case C-220/18 ML v Hanseatisches Oberlandesgericht Bremen, ECLI:EU:C:2018:589, (25 July 2018); Case C-128/18 Dumitru-Tudor Dorobantu v Generalstaatanwaltschaft Hamburg, ECLI:EU:C:2019:857, Grand Chamber (15 October 2019). Discussed ibid.

[128] Ibid.

[129] DAA 2023, Explanatory Report paras 87 and 88 which explain Article 6. Bivolaru and Moldovan (n 47) is cited.

[130] Aranyosi and Cãldãrau (n 58).

[131] Di Franco and de Cahvalho (n 64), 1228.

[132] Ibid. 1229-1230.

[133] ML (n 127), para.77.

[134] ML and Dorobantu (n 127) respectively at paras. 112 and 68.

[135] Bicknell (n 50), 635.

[136] Romeo Castaño (n 59); Bivolaru and Moldovan (n 47).

[137] Council of Europe Parliamentary Assembly, Resolution 1471 (2005) on Accelerated Asylum Procedure in Council of Europe Member States para.8.2.2 cited in Ilias and Ahmed (n 112) at para. 64.