A Case Study of Judicial-Legislative Interactions via the Lens of the DSA’s Host Liability Rules

Table of Contents: 1. Introduction. – 2. Analytical framework and methodology. – 3. Case Study: the development of intermediary liability for host service providers. – 3.1. Justification of case selection. – 3.2. Host liability under the ECD. – 3.3. The Court’s jurisprudence on host liability. – 3.4. Host liability under the Digital Services Act. – 4. Has the Court exercised influence? – 5. Conclusions.

Abstract: This paper explores the relationship between the Court of Justice of the European Union and the EU Legislature. Taking intermediary liability rules for hosts under Article 14 of the e-Commerce Directive (ECD) and its subsequent revision under the Digital Services Act (DSA) as a case study, the paper explores this relationship. The analysis focuses on the distinction between ‘active’ and ‘neutral’ roles for hosts developed by the Court of Justice in its case law. The paper shows how the Court’s rulings shaped the application of host liability rules under the ECD, subsequently impacted their revision during the DSA negotiations, and how the Legislature addressed the uncertainty created by its case law on voluntary investigations via the introduction of a ‘Good Samaritan’ clause. Methodologically the paper draws on legal scholarship and political science literature, employing a mixed methods approach which combines doctrinal analysis, process tracing, and a law attainment typology to explore these dynamics. The paper concludes that while the Court’s jurisprudence was important during negotiations, it also highlights the challenges it posed as both a form of legal regulation, and by extension as a tool of integration. These issues necessitating legislative intervention. The paper’s findings lend weight to the increasing body of literature in legal scholarship which argues that there has been a shift in European integration dynamics, with positive integration playing an increasingly important role.

Keywords: Court of Justice – EU legislature – host liability – Digital Services Act – judicial-legislative interactions – European integration dynamics.

1.   Introduction

The relationship between the Court of Justice and the Union Legislature has long been the subject of scholarly debate. At the heart of this debate lies the tension between law and politics and, in the European context, theories of European integration that explore the dynamics between negative and positive integration, ‘Integration through Law’, and more recently narratives of the Union’s over-constitutionalisation. This paper seeks to consider these dynamics by exploring how the Court impacts legislative negotiations, how political institutions engage with case law, and what these dynamics might tell us about broader institutional relationships and integration as a process.

To do this, the paper explores the interactions between the Court of Justice of the European Union and the Union Legislature during the drafting of legislation.^[1] It takes the development of intermediary liability for host service providers under Article 14 of the e-Commerce Directive (ECD)^[2] as a case study, assessing how the provision was interpreted by the Court and the subsequent impact of its jurisprudence on the drafting of the Digital Services Act (DSA).^[3] The legal analysis focuses on the distinction between ‘active’ and ‘neutral’ roles developed in case law to determine when hosts may be attributed with ‘knowledge’, and therefore when they may be deprived of the protection offered by the ECD’s liability exemptions.^[4] This jurisprudence created legal uncertainty regarding voluntary own investigations conducted by hosting providers, leading to the introduction of a ‘Good Samaritan’ clause in response to the ambiguity created by the Court. That provision partially draws upon the Court’s jurisprudence to resolve legal uncertainty.

The paper argues that while the Court’s influence on the Legislature was wide (in that there is extensive consideration of its jurisprudence and its ideas are drawn on both implicitly and explicitly), its influence was limited. The Legislature, in seeking to determine how to revise the ECD’s liability rules turned to the Court for guidance, reinforcing the findings of Weatherill who has argued that case law has become a body of drafting guidelines for the Union Legislature.^[5] Yet, the case study also finds the influence of the Court was varied. Different institutions, and different actors within those institutions, drawing different conclusions as to the meaning of case law or deployed it differently.

The paper opens by outlining its analytical framework and methodology. The analytical framework explores the synergies between some of the dominant theoretical frameworks for understanding the relationship between the Court and Legislature in European integration. These frameworks are then linked to broader perspectives of the Court as a ‘powerful’ institution in European integration. The paper shifting its focus towards the concept of ‘influence’ which is deployed as a tool to trace the wider array of ways in which judicial impact may occur.

 To distinguish between degrees of influence on legislation the paper pays particular attention to the locus of the incorporation of the Court’s jurisprudence. Emphasising the differing purposes and functions of recitals and articles, and the implications of this for the operation of legislative frameworks, the paper explores how the asymmetrical incorporation of jurisprudence into these provisions impacts a framework’s application, and by extension the influence of case law on law-making. This is combined with three methods for tracing influence in legislative negotiations. Traditional doctrinal research,^[6] process tracing,^[7] and a law attainment typology which typifies influence into four categories; ‘codification’, ‘modification’, ‘non-adoption’, or ‘override’.^[8] With these tools in hand the paper turns to the case study, justifying the case selection, assessing the rules under ECD and how they were interpreted by the CJEU.^[9] The paper then evaluates the impact of the Court’s jurisprudence on the formulation of the DSA host liability rules and the offshoot effect of that jurisprudence for the insertion of the Good Samaritan clause under Article 7. The study places a particular focus on the uneven incorporation of the Court’s jurisprudence into recitals and articles to better understand the nature of the influence exercised.

In the broader scholarly context, the case study contributes to the literature exploring the relationship between the Union’s legislative institutions and its judicial branch, as well as reflecting on broader trends in the dynamics of European integration.^[10]

2.   Analytical framework and methodology

The idea that the Court of Justice is a powerful institution is particularly prominent in the political sciences. This attribution of power is clearly present in Scharpf’s asymmetry thesis. He posited that while the Union is adept at integrating negatively, its ability to integrate positively is limited. Negative integration is easily pursued owing to the impact of European law and the ability of the Commission and the Court to expansively interpret the Treaties.^[11] Over time extending the Union’s reach across almost all areas of policy without recourse to political decision-making. Politics unable to respond to these negatively driven integrationist tendencies due to the high institutional barriers to legislating in areas driven by negative integration. The resulting asymmetrical preference for negative integration acting to the detriment of the democratic accountability and the legitimacy of European integration.^[12]

Dieter Grimm later argued that the Court, through its rulings read in conjunction with the principles of primacy,^[13] direct effect,^[14] equivalence,^[15] and effectiveness,^[16] has over-constitutionalised the Union’s legal order.^[17] The Court’s reach extending to almost every area of law and policy, national and supranational alike. These factors, taken in conjunction with its position as the final arbitrator of all issues of Union law,^[18] and its non-majoritarian institutional nature,^[19] resulting in a situation where it de facto directs policy choices at the expense of democratic decision-making by reducing political choices to legal questions; minimising the answers to the application of substantive and procedural legal rules. Grimm noted that this would lend support to Scharpf’s asymmetry thesis. While acknowledging that the Union had effectively pursued positive integration in fields such as environmental protection, it had failed in areas such as social policy.^[20] Scharpf later built on Grimm’s work, drawing a link between his thesis and over-constitutionalisation. He forwarded that the concurrent factors of (1) modern capitalism’s encompassment of all aspects of our societies coupled with (2) the construction of the treaties around economic freedoms allows (3) the Court (and Commission) to intervene in all areas of national governmental powers. Resulting in the prioritisation of economic rights over individual and social rights due to the high barriers to reaching political agreements.^[21]

Both Grimm’s over-constitutionalisation narrative, and Scharpf’s asymmetry thesis, sit alongside the ‘Integration through Law’ framework which is prominent in legal scholarship.^[22] This framework views the law not just as a technical system of rules, but as a normative legitimising force. Law serving as a fundamental mechanism for regulating and fostering economic, social, and political integration via the constitutionalisation of the Union’s legal order. However, it too highlights the difficulties encountered by politics in driving integration forward when Member States seek to protect their own interests.^[23] Union law is not just as an agent for integration, but also an object of integration. Union law is both a product and creator of the European polity.^[24]

While the integration through law narrative lends support to the asymmetry thesis it also differs from it. The latter highlights the imbalance between the economic and the social by emphasising the inability of politics to respond to negative integration. While the former emphasises the central role of politics through the law it creates, this reflects broader trends in legal scholarship that highlight the steady growth of secondary law instruments today compared to previous decades. This growth in secondary legislation is viewed by some scholars as shifting us toward positive integration and rebalancing the historical asymmetry towards politics and the Legislature.^[25] In legal scholarship this idea is often articulated via the ideas of ‘judicial deference’ or ‘legislative priority’ rules.

Legal research increasingly emphasises the importance the Court places on legislation.^[26] Van den Brink laments the systematic neglect of the study of the impact of the Union Legislature on Union law, arguing that shifting our focus away from the Court towards political institutions reshapes our understanding of their relationship.^[27] He notes that integration through law has become synonymous with ‘judge-made law’ which does lend support to the asymmetry thesis. However, legislation – in his opinion – is understudied and undervalued in exploring how Union law is made and operates.^[28] His earlier research shows how in citizenship law, where the Court was historically activist, this trend reversed as secondary law increased and primary law is increasingly framed by politics.^[29] Dawson notes a growth in the Legislature’s role, power and legislative output. This is beneficial not just as a means of meeting new challenges via positive integration, but as a tool to protect the legitimacy of the Court as a non-majoritarian institution. By increasingly interpreting the treaties via secondary law, and with the assistance of political institutions, the Court can ‘share the blame’ for difficult decisions.^[30]

Weatherill has been more critical of the degree of deference the Court has shown to the Legislature, particularly the widespread use of Article 114 as a legal basis.^[31] Whereas de Witte argues the wide discretion applied regarding Article 114 is coherent with the function and character of the competence to legislate for the internal market from a legal perspective. It allows politics to adjudicate if, and how, to incorporate non-market concerns into internal market legislation, rather than the Court.^[32] Enabling the mainstreaming of non-economic (social) considerations into the Union’s legal order, which became more prominent under Lisbon.^[33] Maduro observed in his seminal work that even in areas heavily influenced by primary law, the Court paid significant attention to the legislative and political preferences of political actors.^[34] Bogoeski, examining the revision of the Posted Workers Directive, argues that while the Legislature did ‘not reverse or radically transform the Laval paradigm but rather operates within its framework’ the revision ‘substantially adjusted the Laval Quartet’s operational framework’.^[35] Equally Zglinski argues that it is fundamentally a mistake to assume that the Legislature merely takes its cues from the Court. Highlighting how secondary law departed from Cassis de Dijon by establishing minimum alcohol requirements and legislating away one of the Court’s most infamous decisions.^[36]

In summary, there is considerable debate as to the relationship between the Court and the Legislature. These differences to some degree grow from the different theoretical frames, and the different empirical methods, used to reach and validate findings. To consider these issues the paper explores interactions between the Court of Justice and Legislature in practice on political decision-making processes. It does so by considering the Court’s impact on the development of the Digital Services Act. The paper shifts its methodology away from notions of ‘power’ and toward that of ‘influence’.^[37] The justification for this approach is that preoccupations with notions of ‘judicial power’ may lead us to neglect the broader ways in which politics responds to court-driven law, and vice versa.^[38] By considering the Court’s influence rather than its power, this paper seeks to capture a wider array of avenues via which the Court may impact the creation of legislative acts, and political responses to case law.^[39] Influence – from a political science perspective – characterises an actor’s ability to shape a decision in line with their preferences.^[40] Like power, it is a relational concept, implying an actor may cause another actor(s) to change their course of action.^[41] However, influence, unlike power is devoid of the threat of sanctions, a key characteristic of power.^[42] In seeking to trace influence the paper is not attempting prove causality as such claims can be misleading. In particular as the level of documentation needed to account for the impact of individuals – if it exists – is not publicly available.^[43]

How then can we track influence? My proposed answer is rather simplistic; look for doctrinal and interpretative similarities between case law and legislative acts. In addition, try to identify references to the Court by legislative institutions during negotiations. Where we see a mirroring of language between texts it is fair to presume there is a possible link between case law and the legislation adopted. Yet, the mere presence of linguistic similarities does not indicate definitive influence; it could be coincidence. To address this the paper uses inductive reasoning underpinned by the combination of three methods – doctrinal research, process tracing, and a law attainment typology – to identify links between case law and legislation.

To better assess the nature of the Court’s influence the paper pays special attention to the distinctions drawn between articles and recitals of legislation.^[44] Article 296 TFEU specifies that Union acts ‘shall state the reasons for which they are based…’.^[45] This was expanded upon by the institutions via two instruments. In 1993 the Council adopted a Resolution on the drafting of legislation, which was built upon in 1998 by an international agreement between the Parliament, Council, and Commission establishing common guidelines for drafting legislation.^[46] These instruments have been expanded upon via a joint practice guide – also issued by the Parliament, Council, and Commission – for persons authoring Union legislation. It states that recitals are a mandatory requirement of Union legislation, representing the locus where the obligation imposed by Article 296 TFEU is fulfilled.^[47] Klimas and Vaičiukaitė note that legally, recitals must be ‘sufficient as a matter of law; that is, they must sufficiently describe the purpose of the act’ for it to be operative, otherwise, the measure is invalid.^[48]

However, recitals do not have the same legal status as articles. Articles are the binding provisions of a legislative measure,^[49] also called the operative provisions or enacting terms.^[50] Recitals are subordinate to articles and may only affect their scope where they suffer from ambiguity or inconsistency.^[51] Where both clear but there exists an inconsistency between them, the article’s provisions take precedence.^[52] Where recitals are clear, but an article suffers from ambiguity, the recitals will be used to interpret the provision.^[53] The Court often deploys recitals to address uncertainty in articles. This has recently been restated by the Court in Kočner v Europol.^[54] Connecting this back to the question of influence, that the Court’s jurisprudence is not incorporated directly into articles does not necessarily reflect a lack of, or decreased, influence. Recitals may act as controls on articles where ambiguity exists, opening a (intentional?) doorway for judicial influence. There are even examples of the Court’s jurisprudence being directly and explicitly codified in recitals, most notably Directive 2006/54/EC on the implementation of the principle of equal opportunities and equal treatment of men and women in matters of employment (though this is undoubtedly unusual).^[55]

Building on this methodology, the paper combines three methods to track the Court’s influence. First, doctrinal research is used to identify similarities between jurisprudence and legislation.^[56] It requires the use, and understanding, of specific words and phrases with unique, multi-layered, legal meanings, entangled and interwoven with other legal rules and concepts, creating a web of meaning(s) not always accessible to non-lawyers.^[57] Doctrinal approaches to legal research are therefore quantitative and qualitative, requiring engagement with both the trends in the use of rules over time (quantitative) and detailed engagement with minute details of the law on a case-by-case basis (qualitative).^[58] A doctrinal analysis is used to compare case law with the final DSA text. However, this approach reveals only part of the Court’s impact given it only allows us to compare A (the ECD), with B (the CJEU’s case law), and C (the DSA) from a textual perspective. Events occurring between B and C remain unobserved. To address this an adapted form of process tracing is used to examine what happens between a ruling and the legislative proposal, with a particular emphasis on the impact of jurisprudence on legislative negotiations.^[59] This enables inferences that can be used to map the impact of the CJEU’s jurisprudence.^[60] The research is conducted backwards to review the hypothesised cause (the Court’s case law) which is traced through the mechanisms (e.g. the EU Legislature)^[61] that act as conduits via which case law – and the Court’s influence – may travel into a legislative act as recorded in documents produced during legislative negotiations.^[62] I draw on an array of documents from legislative negotiations, beginning with the Commission’s original proposal,^[63] intervening documents produced during legislative negotiations,^[64] and the final text adopted, to account for the input of a wide array of actors between (A) and (C).^[65] There are of course limitations to this approach. Examining only negotiation documents excludes other avenues via which interactions might occur, such as spontaneous meetings of committee members at the coffee bar, conversations via email, and the Commission’s role as a repeat player before the Court.^[66] Moreover, this approach also has the downside of not examining the Commission and Member State relationships with Union institutions and the impacts of these interactions on litigation strategies before the Court, prior to the issuance of a ruling at point (A). This by extension leaves it unknown how much of the Court’s case law is of its own genius.^[67] The findings herein nonetheless provide valuable insights by utilising official records of these negotiations to gain at least a partial view of these interactions at a very detailed level of examination.

Finally, how can we measure influence? Martinsen forwards a four-category typology for measuring judicial influence on legislative outputs, adopted by this paper.^[68] ‘Codification’ represents the incorporation of jurisprudence into a legislative output, the strongest form of influence.^[69] ‘Modification’ denotes the adjustment of jurisprudence via the legislative process so its impact is reduced or augmented, a weaker – but still notable – form of influence.^[70] ‘Non-adoption’ denotes situations where no political response can be reached, resulting in legal uncertainty.^[71] ‘Overrides’ constitute the weakest form of influence, characterising the overturning of the Court’s jurisprudence.^[72] Her typology therefore attempts to identify if any of ‘…the rules, principals, and interpretations generated by the court are attained in legislative acts…’.^[73] Martinsen’s typology is augmented via the use of the distinctions between articles and recitals (outlined previously) to better understand the Court’s influence on lawmaking and the relationship between the Court and Legislature during negotiations. Allowing us to draw broader findings regarding the relationship between law and politics, and positive and negative integration.

3.   Case Study: the development of intermediary liability for host service providers

An information society service was defined, under the ECD, as ‘any service provided for remuneration, at a distance, by electronic means and at the individual request of a recipient of services’.^[74] The ECD’s exemptions for intermediary liability were used to identify when such service providers can be held liable for information posted on their platforms. An intermediary service provider, a subcategory of information society service provider, was defined, inter alia, as a provider that either; (A) transmitted information within a communication network, (a ‘mere conduit’), (B) facilitated the automatic, intermediate and temporary storage of information for more efficient onward transmission (a ‘catching service’), or (C) stored information provided by service recipients (a ‘hosting service’).^[75] This paper focuses on the latter providers and hosting services and considers how the Court of Justice’s case law impacted the text of the DSA.

The Commission in its proposal for the ECD noted that while electronic commerce represented a unique opportunity for economic growth, this potential could not be realised ‘unless the many legal obstacles which remain to the online provision of services’ were eliminated.^[76] It linked the proposal to its 1997 ‘Communication on electronic commerce’,^[77] and highlighted the development of a heterogeneous regulatory environment within the Union – owning from both legislative and jurisprudential divergence between Member States – as creating significant barriers to the freedom of services and by extension fragmenting the internal market.^[78] The removal of such obstacles would ensure the free circulation of online services.^[79] Highlighting some of the pressures driving positive integration in this area.

3.1.  Justification of case selection

I argue that the DSA, and these specific provisions, constitute a strong test case for exploring judicial influence on law-making. A significant degree of legal integration has occurred here.^[80] The ECD harmonised liability rules over two decades ago, establishing it as a well-embedded area of Union-level regulatory control. Furthermore, it has been the subject of considerable litigation. Of the forty-four rulings issued by the Court of Justice wherein the ECD formed part of the operative grounds of its decisions between 2002 and 2023, sixteen specifically raised issues regarding Article 14 on hosting.^[81] Building on this point, when legal integration occurs, we may assume that politicians will engage.^[82] The movement of the regulatory fora from the domestic to the Union level is likely to result in competing and complimentary political interests between Member States in an area of regulation of significant economic impact.^[83] The dossier was furthermore subject to considerable scrutiny during its negotiations.^[84] This area of regulation links to wider political and societal debates as to the governance of our online world, in particular how and when we should hold online service providers accountable for the impact of their services on individuals, businesses, and society.^[85] Host provider’s activities, through their storage and making accessible information provided by recipients to third parties, bring into contact with the Union’s internal market freedoms and fundamental rights guarantees.^[86] The Commission observed that while online intermediary platforms have brought benefits to individuals and contributed to the growth of the internal market, there is increasing concern regarding issues such as the trade of illegal goods, services, and online content they facilitate. Moreover the exploitation of the underlying algorithmic systems supporting their services to amply the spread of disinformation has significant implications for the protection of fundamental rights.^[87]

The case law of the Court was fundamental to shaping our legal understanding of the ECD’s host liability provisions, filling in gaps in the legislative framework and providing increased clarity (but also uncertainty) as to its application. These factors increase the likelihood of the dossier’s politicisation and scrutiny given legislation provides a means to manage these conflicts, providing ‘more predictability for any potential dispute on clash of rights’ than case law.^[88]

3.2.  Host liability under the ECD

The ECD was key to the Commission’s legislative agenda during a period of increased concern regarding divergences between the regulatory approaches of Member States.^[89] It grew from the Commission’s ‘European Initiative in Electronic Commerce’^[90] which sought to ensure access to the global market place, establish a favourable legal framework, and promote a favourable business environment.^[91]

It's liability regime was a central component of this agenda as it would help create legal certainty for businesses. Historically there were no Union-level rules regulating this issue and the Court issued no rulings prior to legislative intervention. Under the ECD liability was governed by Articles 12–14. Article 14, our focus, is concerned with liability rules for ‘hosting’ providers. Article 14(1) specified that where an online provider conducts activities which involve storing the information of a service user, they are not liable for the information stored, subject to two conditions; (A) they do not have ‘actual knowledge’ of either illegal activity or information, and with regard to claims for damages they are not aware of the facts or circumstances from which such illegality is apparent.^[92] (The majority of analysis will relate to this provision). (B) Upon gaining such knowledge or awareness the provider acted ‘expeditiously’ to remove or disable access to the information in question.^[93] Article 14(2) specifies that 14(1) does not apply where the service recipient is ‘acting under the authority or control’ of the provider. Hosts were therefore providers who store information supplied by users under the ECD. E.g., YouTube, iCloud, Facebook, or Instagram. Relatedly Article 15 of the ECD prohibits Member States from imposing general obligations to monitor or seek facts relating to illegal activity, meaning specific and voluntary monitoring requirements are permissible.^[94]

3.3.  The Court’s jurisprudence on host liability

Google v Louis Vuitton is a central case in this area.^[95] The Court was asked how to attribute knowledge to a provider per Article 14. Louis Vuitton took issue with Google’s advertising system, AdWords, arguing it infringed its trademarks by allowing third-party advertisers to use similar words and iconography stored as search terms within the AdWords system. The Court relied on the idea of an internet referencing provider playing an ‘active role’ with regard to its knowledge or control over the stored data per Article 14(1)(a).^[96] It considered Recital 42 ECD which outlined that the liability exemptions covered only activities of the provider which were ‘of a mere technical, automatic and passive nature,’ wherein information is transmitted or temporarily stored, for the sole purpose of making its transmission more efficient. The Court reasoned this implies the provider possesses neither knowledge nor control over the information.^[97] Recital 46 stated that ‘to benefit from the limitation of liability, the provider of an information society service, consisting of the storage of information, upon obtaining actual knowledge or awareness of illegal activities, has to act expeditiously to remove or disable access to the information concerned’.^[98]

The Court reasoned that the exemptions were intended to apply where the provider ‘has neither knowledge of nor control over the information’ per Article 14(1), with Recital 42 specifying that the activity is of a ‘mere technical, automatic and passive nature’.^[99] To establish liability it needed to be established if a provider acted in a ‘neutral’ manner; whether its conduct was merely ‘technical, automatic and passive, pointing to a lack of knowledge’ or control.^[100] This test of neutrality was new, appearing nowhere in the ECD, the Court equated the technical, automatic and passive conditions to mean a party must act neutrally. It concluded that Google – with the aid of software – processes data entered by advertisers, and the resulting advertisements were displayed under conditions controlled by Google who determined the order in which results were displayed based on the amount of money paid by the advertisers.^[101] The Court clarified that such payments, or that fact Google set the terms and conditions of payments, could not have the de facto effect of depriving them of the liability exemptions.^[102] Additionally, that there were similarities between the keywords selected, and the search terms of internet users, wasn’t itself sufficient to attribute knowledge, or control, to a provider.^[103] However, Google’s role in drafting the commercial messages which accompany advertising links, or the selection of keywords, was a factor which could be considered to attribute knowledge or control.^[104] If Google had not played an active role, – it acted neutrally for the purpose of Article 14(1)(a) – it could not be held liable for the data unless it failed to act expeditiously to remove it per Article 14(1)(b). This was ultimately a matter for the referring court to determine as the arbitrator of facts.

Similarly to Google v Louis Vuitton, it was argued in L’Oréal v eBay that eBay was liable for trademark infringements, in particular through the displaying of trademarks on its website and in sponsored advertising links displayed by search engines.^[105] The Court observed that eBay processed data, entered by its ‘customer-sellers’, provided assistance to sellers to enhance their sales offers, helped set up online shops, promoted and increased sales, including by advertising products sold on its platform, by service users, using search engine operators to trigger the display of advertisements on the behalf of its customer-sellers.^[106] It specified that where a service provider gave assistance entailing the optimisation of the presentation of offers for sale or promoting said offers, ‘it must be considered not to have taken a neutral position’ but rather to ‘have played an active role’.^[107] However if a provider merely stored sale offers, set the terms of service, was remunerated, and provided general information to customers it ‘cannot have the effect of denying it the exemptions’.^[108] The Court cited Google v Louis Vuitton, pointing to the origin of the neutral concept.^[109] This ruling also saw the Court underline that general monitoring obligations of ‘all the data’ of customers to prevent future infringements were incompatible with Article 15 ECD.^[110]

These rulings, while creating clarity regarding Article 14’s scope and application, simultaneously created ambiguity vis-à-vis situations where hosts utilised automated technological measures to detect, without human intervention, illegal activities or information.^[111] Could this lead to the attribution of actual knowledge? These issues were highlighted by the Commission during negotiation on the DSA,^[112] and as was anticipated^[113] the Court clarified some of these issues in YouTube v Cyando which was issued six months after the DSA proposal was tabled.^[114] The Court reiterated that when assessing if a host may benefit from Article 14, it is necessary to determine if its role is neutral.^[115] It clarified that the fact that a provider voluntarily implements technological measures to detect, in uploaded videos, content which infringes copyright does not, per se, mean it plays an active role.^[116] It outlined a provider may not be deprived of Article 14(1)(a)’s exemptions solely on the grounds it is generally aware or possesses ‘abstract knowledge’, that its platform is used to share information which illegal.^[117] Actual knowledge must be ‘specifically established or readily identifiable’,^[118] and ‘the fact that the operator … automatically indexes content uploaded … that that platform has a search function and that it recommends videos based on users’ profiles or preferences is not a sufficient ground for the conclusion that that operator has ‘specific’ knowledge of illegal activities carried out on that platform or of illegal information stored on it.^[119]

3.4.  Host liability under the Digital Services Act

The Commission proposed the DSA in December 2020.^[120] The proposal took the form of a regulation and was framed as ‘building on key principles set out in the e-Commerce Directive’.^[121] It described the ECD’s liability rules as a ‘foundation of the digital economy’,^[122] citing stakeholder evaluations^[123] as strongly supporting their maintenance.^[124] The European Parliament, adopted two resolutions before the DSA’s formal tabling.^[125] In the first it called for a ‘comprehensive revision of the E-Commerce Directive’^[126] which would not ‘create any derogations or new exemptions to the current liability regime’^[127] specifying that it should take into account the case law of the Court concerning ‘active’ and ‘passive’ roles.^[128] In the second it underlined its preference for the ‘preservation of the current framework on the limited liability …’.^[129] The Commission interpreted these as a call for the ‘ambitious reform of the existing e-Commerce legal framework while maintaining the core principles of its liability regime’.^[130]

The proposal noted that the liability rules had been interpreted by the Court which provided ‘valuable clarifications and guidance’, but argued these clarifications should be included in a regulation to ensure effective harmonisation and to avoid legal fragmentation.^[131] (Implying the Court caused the Commission to change its course of action from maintaining the status quo). The Commission stated that the ECD’s liability framework had been maintained via the DSA proposal ‘as interpreted by the Court of Justice…’.^[132] The proposal draws on several legal studies regarding the ECD,^[133] in addition to its legal analysis which utilised a ‘rich collection of case law’.^[134] The proposal highlighted the uncertainty created by case law, in particular its disincentivising impact on voluntary own-investigations or using automated measures to detect illegal activity.^[135] This uncertainty is also highlighted in the literature.^[136] This points towards the Court’s impact on the drafting process, and the Commission’s detailed consideration and instrumentalisation of case law to justify and insulate its policy positions.^[137]

The staff working documents observe ‘clear evidence of legal fragmentation and differentiated application of the existing rules by Member States, and ultimately by national courts’.^[138] Specifically the ‘liability regime for online intermediates continues to establish the key regulatory pillar’ for the online world,^[139] however ‘conflicting interpretations in national courts (sometimes even between domestic courts) have introduced significant levels of uncertainty’.^[140] The existing rules were described as ‘outdated’.^[141] Furthermore, while Recital 42 ECD was interpreted to mean host providers must act in a ‘passive and neutral role’,^[142] national courts subsequently applied this case law in contradictory fashions. Some equated the active role with ‘appropriation’ of content such that a ‘reasonably informed user’ would conclude the host is the author.^[143] Thus highlighting the weakness of case law as a form of legal regulation, as well as broader structural issues with the Union’s legal system (in particular the co-opting of national judicial infrastructures to establish a decentralised pan-European judicial network).

The Commission emphasised the importance of avoiding scenarios where automatic, algorithmic ordering, displaying, tagging, or indexing, of content leads to the attribution of an ‘active role’, given these activities are necessary for modern online services.^[144] Case law might disincentivise smaller players leading them to avoid taking proactive measures to detect illegal activities which might lead to their attribution with an ‘active role’.^[145] The Commission’s proposal highlighted the need to clarify the rules and eliminate disincentives for voluntary own-investigations undertaken to ensure user safety.^[146] This concern was highlighted by stakeholders: ‘intermediaries, national authorities, research institutes and civil society organisations consider that the current liability regime creates disincentives to act and call for the removal of disincentives for voluntary measures’. Equally, some user and trade associations, and creative industry representatives, expressed concerns ‘that such clarifications could weaken the responsibilities of intermediaries, absent positive obligations’.^[147]

Thus, the Commission’s proposal shows considerable engagement with the relevant case law, highlighting its limitations as a form of legal regulation,^[148] and of negative integration by emphasising the importance of intervention by political institutions.^[149] It is important to recall that the DSA proposal was tabled before the ruling in YouTube v Cyando which clarified some of the above issues.

The original proposal referred to the ‘Court of Justice’ on eight occasions,^[150] and to ‘case law’ twice.^[151] The proposal maintained the ECD liability regime, replicating Articles 12–14 as Articles 3–5 in the DSA. The recitals restate the Court’s above jurisprudence. Recital 18 addressed the application of liability rules, outlining they should only apply where a service provider is providing services ‘neutrally, by a merely technical and automatic processing of the information’ and should not apply where a provider plays an ‘active role of such a kind as to give it knowledge of or control over, that information’,^[152] mirroring the Court’s ruling in L’Oréal v eBay.^[153] Indicating significant influence, which is particularly evident when we consider that Recital 18 concerns all intermediary service providers, not just hosts.^[154] Indeed this language is retained in the final regulation.^[155] However it is important to note that other amendments, which sought to widen the scope of this recital, ultimately failed.^[156] Recital 19 also illustrates influence, that provision indicating the different categories of intermediary service providers – ‘mere conduits’, ‘catching’, and ‘hosting’ services – should be considered having regard to the way those distinctions have been interpreted by the Court.^[157] Similarly, Recital 20 specified that providers who deliberately collaborate with a service recipient to undertake illegal activities do not provide their services ‘neutrally’,^[158] mirroring both L’Oréal v eBay and Google v Louis Vuitton.^[159]

Articles 3 ­–5 replicated the ECD’s Article’s 12 – 14 with relatively minor textual changes. Article 5 governed host liability and Article 5(1) was almost identical to Article 14(1) ECD, however instead of imposing a positive obligation on Member States to ensure hosts are not liable where they fall within Article 14, it creates a direct entitlement for providers reflecting the change from that of a directive to a regulation.^[160] One notable change was the addition of Article 5(3), concerning platform liability under consumer protection law for the trade of goods or services online, however, this provision is not of immediate concern to the present analysis.^[161] A notable change from the ECD’s liability framework is the introduction of a ‘Good Samaritan’ clause. This clause can be connected to the Court’s rulings that created uncertainty regarding voluntary investigations where a platform takes measures to prevent illegal activities. Article 6, titled ‘voluntary own-initiative investigations and legal compliance’,^[162] outlined that providers would not be deemed ineligible for the exceptions in Articles 3–5 solely on the basis that they conducted a voluntary investigation, or other activities ‘aimed at detecting, identifying and removing, or disabling access to, illegal content’.^[163] Nor for taking the ‘necessary’ steps to comply with Union law.^[164] This provision should be considered in light of Recital 25 of the proposal which specified it should be clarified, to not ‘discourage activities aimed at detecting, identifying and acting against illegal content’ and that providers who undertake such actions are not necessarily deprived of the exceptions provide the activities are conducted ‘in good faith and in a diligent manner’.^[165] Good faith including actions taken to ‘comply with Union law’ and that such activities ‘should not be taken into account when determining’ if they can rely on an exemption from liability, particularly where they provide a ‘service neutrally’ and ‘can therefore fall within the scope of the relevant provision’.^[166] Relatedly, we see the replication and simplification of Article 15 ECD as Article 7 DSA which specifies that no general obligations to either monitor information transmitted or stored by intermediary providers nor to actively seek facts relating to illegal activity shall be imposed.^[167] This provision must of course be considered in light of Recital 28 of the dossier which specified the prohibition does not affect monitoring in specific cases.^[168] The Commission observed during the DSA’s tabling that cases were pending before the Court addressing the issue of voluntary own investigations. Equally, the protection of ‘Good Samaritans’ has existed in the United States since the 1990s.^[169]

An amendment to Recital 18 proposed by Parliament sought to specify the mere ranking, ordering, and use of recommender systems, does not, itself, amount to an exercise of control.^[170] This amendment from January 2022 appears to incorporate the CJEU’s decision in YouTube v Cyando, issued in June 2021, which resolved some of the uncertainty emerging from Google v Louis Vuitton and L’Oréal v eBay, pointing towards the Court's ongoing influence. A draft text dated the 15^th of June, circulated by Fabrice Dubreuil, Chairman of COREPER, following trialogue, sees the removal of the jurisprudence from YouTube v Cyando.^[171] This might indicate pushback against the Court. Equally, it may indicate a choice to insulate adjacent legal areas from the Court’s influence by limiting the incorporation of its specifically to hosting providers rather than extending it to all intermediary service providers.

3.4.1. Host Liability

The First Reading stage before the Council is particularly interesting, in one progress report Member States indicated their support for the existing ECD liability framework.^[172] Several delegations demonstrate detailed engagement with host liability issues, such as the Netherlands, who queried when the ‘actual knowledge of illegal content’ requirement would be fulfilled under Article 5 DSA,^[173] and Denmark, who while critical of Article 5(3)’s failure to address importers responsibilities vis-à-vis online marketplaces and the sale of goods from third countries, indicated general support of the DSA’s liability provisions.^[174]

An examination of internal working documents from Council negotiations, released following freedom of information requests, show far more detailed consideration of the CJEU’s rulings, and their impact, on the formulation of the host liability provisions. Following a COREPER^[175] meeting in November 2021, a compromise text on the DSA was circulated to Member States.^[176] An amendment to Recital 22^[177] was proposed which specifies that actual knowledge, for host service providers, cannot be obtained solely on the basis that a provider is generally aware, or possesses abstract knowledge, that their service is used to share illegal content.^[178] Furthermore, it specifies that a provider’s automatic indexing of content, offering a search function, and recommendation of content to users based on their profiles or preferences, are not sufficient grounds to reach the conclusion a provider has specific knowledge regarding illegal activities/content.^[179] This amendment mirrors, almost verbatim, YouTube v Cyando.^[180] However this language was subsequently removed bringing the Court’s influence into question. Yet, the language is retained in Recital 22, which specifically addresses liability for hosting services in the proposed DSA text from the 15^th of June, 2022.^[181] This was approved by the Council on the 4^th of October 2022.^[182] Demonstrating the continuing impact of the Court. While the inclusion of the jurisprudence from Youtube v Cyando would have extended the impact of case law to mere conduits and catching service providers, the maintenance of the Court’s test in Recital 22 upholds the current position and embedded it.

Government queries on the dossier point towards the importance placed on the Court’s rulings. For instance, a question submitted by an unspecified Member State regarding Article 2(f) of the draft expressed concern regarding search engines and the applicability of the DSA’s requirements to such providers.^[183] The response noted that Google v Louis Vuitton did not exclude referencing services from being considered hosting services, but simultaneously remarked that every case brings a unique set of facts and circumstances which may mean such a provider falls outside the scope of the liability exemption, this ultimately being an issue for the Court.^[184] Indicating an intention to avoid creating a rigid framework, favouring a flexible approach that would facilitate the Court’s continued interpretation, development, and adaptation of the rules to new situations.

France, in query forty-two, submitted that Articles 5(2) – (3) and Recital 18, taken collectively, codify and incorporate case law. The French authorities point out that Article 5(2) excluded the exemption from liability where the recipient of the service acts under the control or authority of the service provider.^[185] Furthermore, Recital 18 applies the same ‘terms of case law regarding the ‘active role’ played by the marketplace in the transaction’.^[186] However, they argued that these terms require legislative clarification, in particular the criteria to be used to determine when the role of the third-party sellers, or service providers, can be considered weak or de facto null and void.^[187] In a detailed response the Commission stresses that the status of an intermediary as ‘neutral’ needs to be ‘assessed on a case-by-case basis’.^[188] It highlighted that there was still uncertainty as to when ‘a host has played an active role of such a kind as to lead to knowledge or control…’^[189] It goes to lengths to clarify that automated activities, such as tagging and providing search functionalities, are now ‘absolutely necessary to navigate among an endless amount of content and should not be considered as ‘smoking gun’ for such an ‘active role’…’.^[190] It concluded by noting that the Court might clarify these questions in pending cases,^[191] which it subsequently did in Youtube v Cyando, and for which the Parliament appears to have tabled an (unsuccessful) amendment.^[192] Regardless of the success, this again highlights the impact of the Court’s on negotiations.^[193]

Finally, in query fifty-two, concern is expressed regarding the non-codification of case law. While the query notes that the Court’s jurisprudence forms part of several recitals, it laments that is not directly incorporated into the articles.^[194] The Commission responded rather strongly stating the ‘DSA does not ‘codify’ existing case law…but reflects such case law in the recitals, making also clear that it is subject to further interpretation by the Court’.^[195] It again stressed that more cases are pending that might provide further clarifications on how to interpret the liability exemptions.^[196] There are a few other amendments proposed to Articles 3 ­­– 5, most involving changes of a minor nature.^[197]

3.4.2. The Good Samaritan Clause

Article 6 – which allows providers to retain their safe harbour liability exemption where they take steps to detect, identify, remove, or disable access to illegal content – is discussed in Council negotiations in February 2021. The Commission was asked to ‘elaborate on which obligations, if any, are imposed on platforms that obtain ‘actual knowledge of illegal activity or illegal content’ because of ‘voluntary own-initiative investigations?’. In particular if it had considered clarifying the provision to make clear that ‘the good samaritan provision applies to the distinction between active and passive intermediaries?’ which could be achieved by ‘codifying the general requirements for neutral/passive intermediary services in line with the case law’.^[198] The Commission stated the DSA did not codify case law in its enacting terms, rather its recitals reflect case law to make it clear that the law is ‘subject to further interpretation by the Court’. The Commission noted that there were then two cases pending in which it might ‘give further guidance on how to interpret the requirement of ‘neutrality’ in relation to the conditional liability exemption for hosting services’.^[199]

Concerning Article 6, the Commission specifically states that such activities and measures taken by providers ‘should not be taken into account’ when determining exemptions from liability, in particular if the provider provided a service ‘neutrally’ without this rule ‘implying the provider can necessarily rely thereon’. This indicates that the Commission was conscious of the Court’s existing jurisprudence and its likely impact on the interpretation of the provision, in addition to the likelihood that the Court would provide future clarifications. The Commission wanted to keep the door open for the Court to influence the direction of the law moving forward and envisaged the Court elaborating on its active and neutral jurisprudence. The notion of neutrality explicitly included in the proposal’s original text via Recital 25. Given the brevity of Article 6 as tabled, it is very likely that Recital 25 was intended to control that article’s application and interpretation where uncertainty arose.

Article 6 was amended by the Parliament to include measures to comply with national legal obligations, which is retained in the final regulation.^[200] Amendment 138 introduced an extensive revision specifying that all such voluntary measures must be ‘effective and specific’ with appropriate safeguards, including human oversight, documentation, and requirements to ensure such measures are ‘non-discriminatory, proportionate, transparent and do not lead to over-removal of content’.^[201] However this amendment was unsuccessful.^[202] In the Council, Germany sought to have the words ‘diligent manner’ added, expressing concern that the existing wording was ‘too vague’ and would create ‘legal uncertainty’. The Austria similarly suggested the wording ‘in good faith and in a diligent manner’ be introduced.^[203] The Austrian delegation highlighted its proposed wording on the Commission’s text for Recital 25.^[204] Trilogue documents reveal that Austria’s proposal was the only amendment proposed by the Council to Article 6.^[205] Interestingly Spain links Article 6 with its comments concerning Article 5 and its view that the Court’s jurisprudence from both L’Oreal and eBay should be included in to increase legal certainty.^[206] This proposal was unsuccessful.^[207]

4.   Has the Court exercised influence?

Applying Martinsen’s typology to the case study what can we determine regarding the nature of the Court’s influence? ^[208] The Commission resists the characterisation that it is codifying case law, this resistance in part seems to stem from a motivation to allow the judiciary to continue to develop the rules within the confines of the relevant legislative frameworks. The Commission stated that decisions on attributing liability need to be assessed on a ‘case-by-case basis’ by the Court for example.^[209]

The DSA does incorporate the case law on active and neutral roles within its text via the recitals, but not in the articles.^[210] Under Martinsen’s law attainment typology, incorporation is considered codification, the strongest form of influence.^[211] Therefore the Court has exercised influence over the legislative process.^[212] Augmenting Martinsen’s typology with the legal distinctions between articles and recitals we can further reflect on the type of influence. Recitals are not independently operative; therefore, the Court has exercised a weaker form of influence versus had the case law been replicated with the articles. However, Article 5(a) DSA proposal, which became Article 6(a) in the final act, is somewhat unclear and ambiguous, owing to the wording that host providers are exempt from liability on the condition that they do ‘not have actual knowledge of illegal activity or illegal content…’.^[213] The consequence of this is that the DSA’s recitals, which now refer to the Court’s active and neutral roles, will likely be used to resolve any ambiguity regarding the meaning of ‘actual knowledge’. Furthermore, the incorporation of the Court’s test for liability in Youtube v Cyando via Recital 22 concerning host providers’ use of automatic indexing, search, and recommender systems will equally be used to resolve ambiguity regarding the meaning of ‘actual knowledge’. Therefore, while not part of the operative provisions, the jurisprudence may control the interpretation of Article 6. This is particularly notable as the non-codification of those notions in the operative provisions of the directive arguably makes them less concrete, and therefore open to further interpretation by the Court. This indeed would reflect the Commission’s stated intention that the recitals merely reflect the Court’s case law, ‘making also clear that it is subject to further interpretation by the Court’. Thereby embedding a dynamic approach to interpretation that allows the law to evolve.

Martinsen’s typology seeks to ascertain if any of the rules, principles, and interpretations generated by the Court are attained in legislative acts.^[214] This categorically is the case regarding the host liability rules under the proposed regulation, and the Court of Justice has exercised significant influence in that it has caused another actor to change their course of action. In this instance it has changed the course of action of the legislative institutions, particularly the Commission, but also the Parliament and Council, from a position of maintaining the status quo for host liability to amending the legislation to reflect its augmentation, via case law, of the ECD’s liability rules for host providers.

Equally, the Court has had a significant impact, via its host liability jurisprudence, on Article 7 (previously Article 6 in the Commission’s original dossier). The Court’s jurisprudence in Google v Louis Vuitton and L’Oréal v eBay created legal uncertainty with regard to the liability of intermediary service providers for voluntary own investigations. In part to address this uncertainty Article 7 is created. That provision neither overrides the Court’s jurisprudence, nor embeds it. Rather it clarifies that intermediary services are not to be made liable for the purposes of Articles 4, 5, and 6 where they diligently and in good faith, conduct actions to detect, identify, remove, or disable access to illegal content, or they take measures to comply with legal obligations, in light of the uncertainty that the Court’s rulings created before YouTube v Cyando.^[215] Indeed there seems to be a degree of alignment between the Court’s reasoning in that judgment and this provision. The Court specifying in Youtube v Cyando that where providers implement voluntary technological measures to detect content infringing copyright, this does not necessarily mean it played an active role unless actual knowledge can be attributed to them.^[216] Similar language can be seen in Recital 26 of the final regulation, in addition to the reference to the Court’s notion of acting neutrally with regard to voluntary actions, indicating that the Court’s jurisprudence influenced the drafting of the provisions.^[217] Equally, the introduction of the provision can be seen to take inspiration from Section 230(c) of the Communications Decency Act in the United States which I have pointed toward already. Article 7 can be understood as a response to the uncertainty created by the Court, its case law perceived as disincentivising voluntary attempts to address such activities and content. In this way, the Court has exercised context-shaping influence given that the institutions have shaped ‘their conduct, strategies, and behaviour to fit what the Court has already ruled or is likely to articulate in the future’.^[218]

5.   Conclusions

This paper has sought to examine the relationship between the Court of Justice and the Union Legislature with the aim of reflecting on the dynamics of positive and negative integration, and the interactions between law and politics in the contemporary period. It attempted to address these questions by bridging political science literature and legal scholarship via an inquiry that was empirically grounded, conceptually embedded, doctrinally informed, and theoretically framed.

Most clearly it considered how the Court’s interpretations impact legislative processes and equally how political institutions engage with, or contest, such decisions. The findings highlight a clear identification, by political institutions, of the limitations of case law as a form of legal regulation. The Commission’s analysis emphasises the necessity of re-regulation to counteract the legal uncertainty created by the Court. The decentralised application of its rulings by national judicial systems – a key infrastructure via which negative integration is pursued – resulting in unequal enforcement, legal fragmentation, and conflicting domestic applications of its jurisprudence. This by extension creates barriers in the internal market, necessitating re-regulation and the pursuit of further positive integration to address the undesirable effects of case law. Equally, the Legislature moves beyond the Court’s jurisprudence – given its negative impact on legal certainty for voluntary own investigations – via the creation of Article 7 DSA.

Therefore, while case law matters in the legislative process, how, why, and to what extent it matters is open to debate and interpretation. In this regard, the findings herein substantiate the view of Weatherill that case law plays an important role as a body of drafting guidelines for the Union Legislature.^[219] By extension the paper lends weight to the claims of Scharpf and Grimm as to the effects of case law on political decision-making. However, it also highlights that political actors do not engage with case law as neutral interpreters. Rather they actively politicise rulings, drawing differing conclusions as to the meaning of the law which they leverage to suit their own preferences and policy objectives. Their fidelity to the Court’s intended meaning (if it is possible to conclusively identify such a thing) is questionable in some respects. Suggesting that while case law may constitute a body of drafting guidelines, the meaning of those guidelines is shaped as much by politics as by law. This conclusion supports the integration through law narrative which emphasises the central role of politics via the law it creates, highlighting the institutional capacity of the Union to politically direct integration through law.

More broadly the paper sought to examine what these dynamics can tell us about the relationship between these institutions. The paper provides grounds to question the established theoretical frames for understanding the relationships and dynamics between the Court and Legislature. In the early periods of the Union’s development the Treaties – and the Court’s interpretation of them – dominated the integration process. Against this backdrop Scharpf developed his asymmetry thesis, the legal community and its integration through law theory, both later influencing Grimm’s over-constitutionaliation narrative. However, legal scholars are increasingly re-examining the accuracy of these theories in the contemporary period, emphasising the growth of secondary legislation as creating a shift in institutional dynamics.^[220]

The findings of this paper lend empirical support to the increasing claims in the literature which suggest that the ability of these theories to sufficiently capture the current dynamics between the Court and Legislature, and the driving forces behind European integration more broadly, need to be revisited. Scharpf and Grimm place at the core of their argumentation a strong belief that politics lacks the institutional capacity to direct the course and pace of European integration given the power of negative integration. While a key concern of their work is the perceived imbalance between the economic and social, this case study provides evidence that the universal extension of those claims to all areas of Union law is problematic. The case study demonstrates that not all aspects of economic activities come within the scope of the Treaty framework in the same way. While the economic activities of host service providers have profound impacts on our societies, both primary law and case law were completely silent as to liability before legislative intervention. This is unsurprising given the technical and commercial nature of the issues. Yet, the fragmented approach to liability before the ECD’s establishment created barriers to the freedom of movement of services, an obstacle to securing the internal market, and by extension impediment to European integration in this area. However, the removal of these barriers was driven by politics.

This finding, coupled with the increasing evidence of judicial deference toward political decision-making processes highlighted by the literature, suggests a shift in the relationship between primary and secondary law. De Búrca in 2003 argued that the Court’s rulings in the area of fundamental rights and citizenship law – in particular its decision in UPA – indicate that the Court has entered a period of judicial deference, acting to the detriment of its previous attitude regarding the protection of individual rights characteristic of its earlier case law.^[221] Zglinski’s research furthermore highlights the Court’s increasingly deferential attitude toward national decision-making as underpinned by domestic legal frameworks.^[222] Recently Gerbaudo’s work illustrated how the Court’s rulings in the area of asylum and refugee law point toward a prioritisation of the application of secondary legislation over the right to an effective remedy enshrined in Article 47 of the Charter and the principle of non-refoulment in International law.^[223]

Whether this latter development is good or bad is beyond the scope of this paper. However, the findings underline a need to reengage with these theoretical frameworks. The symmetry of influence between the Court and the Legislature appears to have evolved, shaped by the growth of secondary law and the rise of the Union Legislature within the Union’s institutional framework. Arguably creating a shift in the relationship between law and politics, the Court and Legislature, and negative versus positive integration more broadly.

-------------------
European Papers, Vol. 10, 2025, No 1, pp. 259-291
ISSN 2499-8249 - doi: 10.15166/2499-8249/832

* Assistant Professor of European Law, Maastricht University; Affiliate Research Fellow, Jacques Delors Centre (Berlin), ronan.riordan@maastrichtuniversity.nl. The author would like to acknowledge the support provided by the Hertie School, Berlin, which facilitated much of the work herein.