When Does Credit Stop Being Finance? Debt Collection before the CJEU

 

Cătălin Gabriel Stănescu, Reader in Law at Newcastle University and member of the European Commission’s Financial Services Users Group.

Photo credit: BNP Paribas HQ, by Boubloub, via Wikimedia Commons

A pending reference from the Greek Council of State brings an old question back before the Court of Justice: when credit is transferred to a purchaser and its administration and recovery are entrusted to a separate company, how should EU financial law classify the activities of that company? In Trapeza tis Ellados II, C-210/26, the Court is asked whether an undertaking engaged exclusively in credit servicing, without assuming credit risk, qualifies as a financial institution under the Regulation (EU) No 575/2013 on prudential requirements for credit institutions (Capital Requirements Regulation, CRR).

In a 2021 article,  I asked whether informal debt collection qualified as a financial service under EU law and what that qualification would mean from a legal perspective (pp 535-538). The answer was less straightforward than the question suggested. Treating collection as a financial service might activate additional protection for those subjected to it, but it might also make that protection depend on the origin of the debt. The difficulty itself was older: my 2015 analysis of MKG-Kraftfahrzeuge-Factoring had already shown how factoring, debt purchasing and collection could overlap without becoming identical.

Two recent CJEU judgments make the pending reference particularly interesting. In Kosmiro, the Court describes debt collection, for VAT purposes, as financial transactions designed to obtain payment (para 50). In IFIS NPL INVESTING, it excludes the purchase or management of non-performing loans from the economic activities covered by a particular definition of financial institution (para 37). The reference in Trapeza II concerns a servicer whose activities include administering and collecting debts, negotiating with borrowers and restructuring repayment obligations. The question is not simply whether these judgments can be reconciled, but how far a classification reached under one EU regulatory regime can be carried over into another. I argue that regulatory differentiation does not require conceptual separation. Lending, assignment, servicing and collection may attract different legal classifications while remaining connected stages in the life of the same credit relationship.

1.     The question in Trapeza II

The reference in Trapeza II was lodged by the Greek Council of State on 13 March 2026 and is currently pending. It arises from sanctions imposed by the Bank of Greece on the former managing director of a credit-servicing company (summary of the request, paras 6-8). The company was authorised and supervised by the Bank but operated exclusively as a servicer. Its activities included legal and accounting monitoring, recovery, negotiations with debtors, settlements and debt restructuring (paras 14-15). It assumed no credit risk (para 29).

The question is whether such a company is a “financial institution” under Article 4(1)(26) of CRR. The definition applicable at the material time covered an undertaking, other than an institution or pure industrial holding company, whose principal activity was acquiring holdings or pursuing one or more activities listed in points 2-12 and 15 of Annex I to Directive 2013/36/EU on access to the activity of credit institutions and the prudential supervision of credit institutions (Capital Requirements Directive, CRD). The dispute turns on two possible Annex I classifications: point 2, ‘lending’, which expressly includes factoring with or without recourse, and point 11, ‘portfolio management and advice’.

The Bank of Greece characterises the work as administrative factoring and, where securitised loans are concerned, portfolio management (summary of the request, para 10). The referring court provisionally rejects both propositions. That conclusion forms part of a broader line of reasoning. Drawing on Lineas – Concessões de Transportes and Others, joined Cases C-207/22, C-267/22 and C-290/22, judgment of 26 October 2023, the referring court starts from the proposition that an undertaking which does not carry out activities in the financial sector cannot qualify as a financial institution (para 23). It then reads the Annex I activities as, in principle, activities linked to the banking or financial sector. It invokes Directive (EU) 2021/2167 on credit servicers and credit purchasers as confirmation, in its view, that credit servicers are not financial institutions within the meaning of the CRR and do not simultaneously carry out activities in the financial sector (para 27). It ultimately concludes that companies performing only the specified servicing activities, without assuming credit risk, ‘do not carry out activities in the banking or financial services sectors’ (para 29). The classification is of immediate importance because it determines whether the sanctions applicable to financial institutions could be used against the applicant. The reference nevertheless raises a broader issue: how EU financial law classifies an undertaking that performs functions arising from a credit relationship while neither originating the credit nor assuming the associated credit risk.

There is also a temporal limit. Regulation (EU) 2024/1623 (CRR III) amended that definition with effect from 1 January 2025. The current Article 4(1)(26) also expressly includes an “ancillary services undertaking” as an alternative basis for classification as a financial institution. Article 4(1)(18), as amended, defines such an undertaking by reference to a principal activity consisting, inter alia, of a direct extension of banking or other activities ancillary to banking. That express route to financial-institution status was absent from the version of Article 4(1)(26) applicable to the dispute. The judgment will therefore interpret a superseded definition. Its reasoning on factoring, portfolio management and credit servicing may nevertheless remain relevant under current law, even though the amended definition now provides an additional route to financial-institution status.

2.     Factoring as debt collection

Kosmiro, C-232/24, judgment of 23 October 2025, concerned two factoring models. Under trade factoring, the factor purchased invoices and assumed the debtors’ default risk. Under invoice factoring, the debts remained with the client, which retained the default risk while obtaining finance secured by those debts. In both arrangements, the factor administered and collected the debts (paras 11-13). The VAT question was whether the relevant fees constituted consideration for exempt credit or taxable debt collection under Article 135(1)(b) and (d) of Directive 2006/112/EC on the common system of value added tax (VAT Directive).

The Court began from the debt-collection exception to the exemption for transactions concerning debts. Exemptions are interpreted strictly, while the exception, which restores taxation, is interpreted broadly (paras 46, 49). Debt collection refers to “financial transactions designed to obtain payment of a pecuniary debt” (para 50). That concept encompasses both factoring models (paras 51-53).

That conclusion builds on an established line of VAT-related activities case law. In MKG-Kraftfahrzeuge-Factoring, C-305/01, judgment of 26 June 2003, the Court held that factoring is merely a variant of the more general concept of debt collection, whatever the manner in which it is carried out (paras 76-77). In AXA UK, C-175/09, judgment of 28 October 2010, it added that debt collection need not concern an overdue debt or involve coercion (paras 31-35). The resulting VAT concept is therefore not confined to coercive recovery of debts already in default.

In Kosmiro, the Court held that, in principle, factoring constitutes a single economic transaction whose main purpose is to transfer responsibility for recovery and collection from the client to the factor and splitting that transaction would be artificial (para 60). In trade factoring, the funds paid for the purchased debts were not a loan to the client and there was consequently no credit relationship between factor and client (para 61). In invoice factoring, by contrast, financing was provided, but not independently of the debt-collection service, of which it was the corollary (paras 62-63). The Court distinguished arrangements whose essential purpose was independently the provision of financing (paras 64-66). Importantly, it rejected the argument that all financing arrangements must receive the same VAT treatment: the legislative distinction between debt collection and the granting of credit permits “financing arrangements subject to VAT and financing arrangements exempt from VAT” to coexist (para 67). The relevant invoice-factoring fees therefore constituted consideration for a single and indivisible debt-collection service subject to VAT (paras 68-69). 

There are firm limits. The Court classifies the transactions for VAT purposes but does not define the institutional perimeter of prudential supervision. It establishes that factoring falls within debt collection, not that every collector performs factoring. Nor does the debtor purchase the service examined in Kosmiro: the factor supplies it to the creditor-client. Finally, purchasing debts need not supply a remunerated service at all. In GFKL Financial Services, C-93/10, judgment of 27 October 2011, a price reflecting the actual economic value of defaulted claims was not consideration for a service to the seller (paras 22-26).

Kosmiro consequently provides no trans-substantive definition of financial services. Its narrower significance is nevertheless important. EU VAT law describes debt collection as consisting of financial transactions designed to obtain payment of a pecuniary debt, and Kosmiro demonstrates that financing and collection can form part of the same economic arrangement without becoming identical legal services. What it undermines is therefore any assumption that the functions of financing and collection must always be legally or economically separate.

3.     The classification puzzle

The difficulty is that the Court has recently used a different regulatory definition to reach a conclusion that appears to leave little room for treating the purchase or management of non-performing loans as a financial activity. In IFIS NPL INVESTING, C-65/25, judgment of 11 June 2026, the Court stated that the economic activities covered by Directive (EU) 2015/849 on the prevention of the use of the financial system for the purposes of money laundering or terrorist financing (Fourth Anti-Money Laundering Directive, AMLD) do not include the purchase or management of non-performing loans (para 37). The conclusion is relevant for Trapeza II because the Greek undertaking did not purchase the claims it serviced, but its activities included their administration, collection, negotiation and restructuring. The question is whether the exclusion of NPL “management” in IFIS also encompasses those distinct servicing functions.

IFIS arose from Italian legislation governing bulk assignments of non-performing loans during the period before the expiry of the deadline for transposing Directive 2021/2167. The referring court asked, among other things, whether entities carrying out such assignments had to be subject to anti-money-laundering supervision.

In this instance, the definition before the Court was Article 3(2)(a) of AMLD, not Article 4(1)(26) CRR. It covered an undertaking other than a credit institution which carried out one or more activities listed in points 2-12, 14 and 15 of CRD Annex I. That statutory route nevertheless matters for Trapeza II because it incorporates the two Annex I activities at issue there: point 2, which includes factoring, and point 11, portfolio management and advice. Yet the Court does not analyse either Annex I activity separately. Its unqualified exclusion of NPL purchase and management consequently raises the question whether either activity can bring an undertaking within the corresponding CRR definition.

That passage is difficult to confine to its facts. Unlike the historical CRR definition, the AMLD did not require an Annex I activity to constitute the undertaking’s principal activity. If the purchase or management of NPLs does not fall within the relevant Annex I activities at all, making that activity the undertaking’s principal business cannot by itself bring it within the CRR definition. IFIS therefore points strongly towards exclusion if the Greek servicer’s work is characterised simply as the management of NPLs.

IFIS bears directly on Trapeza II, but does not resolve it. The dispute in IFIS arose from bulk assignments of NPL portfolios, whereas the Greek undertaking in Trapeza II neither acquired the claims it serviced nor assumed their credit risk. Paragraph 37 of the IFIS judgment extends the Court’s conclusion beyond the purchase of NPLs to their “management”, but does not examine what that activity encompasses. In particular, the Court did not consider whether the administration, collection, negotiation and restructuring performed by a credit servicer could constitute factoring under point 2 of Annex I to the CRD, or whether the management of a portfolio of credit claims could fall within “portfolio management and advice” under point 11 of that Annex. Instead, the reasoning moves directly from the scope of the AMLD to the exclusion of NPL purchase and management. The temporal analysis in IFIS does not qualify that conclusion. Paragraphs 31–36 of the judgment explain why Directive 2021/2167 did not apply to the transfers at issue and could not be applied retroactively; paragraph 37 states a separate conclusion concerning the scope of the AMLD. That conclusion must therefore be confronted on its own terms.

The two definitions also perform different functions. The AMLD identifies obliged entities for the purposes of anti-money-laundering supervision, whereas the CRR definition operates within a system of prudential regulation and financial-sector supervision. In Lineas, the Court interpreted the historical CRR definition in that regulatory context. It concluded that an undertaking whose sole activity consisted in acquiring holdings in companies which did not carry out activities in the financial sector was not a “financial institution”. Its reasoning emphasised the distinction drawn by the CRR between holdings in financial-sector entities and holdings outside that sector, and the inconsistency that would result from subjecting the latter to rules designed for the former (paras 69-81).

Those different regulatory purposes matter, but they do not by themselves resolve Trapeza II. Both definitions incorporate points 2 and 11 of the same CRD Annex I. If those common terms produce different classifications in the AMLD and CRR contexts, the difference must be explained through the wording, structure and purpose of the respective regimes. Regulatory purpose alone cannot determine what counts as factoring or portfolio management.

Trapeza II thus offers the Court an opportunity to provide the analysis compressed into paragraph 37 of IFIS. It could explain whether servicing without credit risk can constitute factoring under point 2; what “portfolio management and advice” means in point 11 when the portfolio consists of credit claims; and whether the servicer’s combination of collection, negotiation and restructuring affects either classification. If the result remains exclusion, the reasoning should identify why those activities fall outside the particular definition rather than turn that exclusion into a general proposition that servicing and collection are external to finance.

Consumer law reveals the relevance of that distinction. In Gelvora, C-357/16, judgment of 20 July 2017, the Court held that the concept of commercial practices directly connected with the sale of a product extends beyond the conclusion of the contract to its performance, to include measures taken to obtain payment (paras 19-21). The claims at issue originated in consumer credit, and their recovery could fall within the Unfair Commercial Practices Directive even after assignment to a debt collection agency (paras 22-25). Excluding credit-repayment transactions following assignment could undermine consumer protection by encouraging traders to separate the recovery phase in order to escape the Directive (para 28). The Directive could also apply where unilateral collection measures were pursued in parallel with formal enforcement (paras 29-31).

Gelvora does not classify debt collection as a financial service. Paragraph 25 appears to distinguish the two activities: the collection agency does not provide consumer credit “as such”. Its significance lies instead in the continuity recognised for the particular purposes of consumer protection. Assignment changes the creditor and recovery changes the activity, but neither necessarily breaks the legal connection between collection and the credit from which the payment obligation arose.

INKO, C-127/15, judgment of 8 December 2016, provides a complementary, activity-specific example. Under Directive 2008/48/EC on credit agreements for consumers, a debt collection agency acting on behalf of a lender to conclude a rescheduling agreement for unpaid credit must be categorised as a “credit intermediary” within the meaning of Article 3(f) (paras 42-44). The classification follows not from debt collection alone but from the additional function performed by the agency within the existing credit relationship: concluding the rescheduling agreement on the creditor’s behalf. The Court further held that acting as a credit intermediary only in an ancillary capacity may remove particular pre-contractual information obligations without altering that classification (paras 47-53).

4.     Continuity without regulatory identity

The case law does not produce a single answer to whether collection is a “financial service”. And, perhaps, it should not. Lending creates credit; purchasing transfers a claim; servicing manages rights and obligations arising from it; collection seeks payment; and enforcement invokes legally compulsory procedures. Those functions may overlap, but they are not identical and may justify different authorisation, conduct, prudential and tax rules.

The difficulty begins when a conclusion designed for one of those purposes is detached from it. Under Directive 2021/2167, a credit purchaser does not require authorisation to operate because it acquires existing non-performing credit rather than creating new credit and raises negligible systemic risk (Recital 40). IFIS excludes the purchase or management of NPLs from the economic activities relevant to the AMLD, while Trapeza II asks whether particular servicing functions fall within the historical CRR definition. A particular acquisition of debts may not constitute a remunerated service for VAT purposes. None of those situations establishes, without another legal step, that collection has become external to the financial relationship from which the payment obligation arose.

Directive 2021/2167 embodies this distinction. A credit servicer manages and enforces rights and obligations related to a creditor’s rights under a non-performing credit agreement, or to the agreement itself, on behalf of a credit purchaser. Credit-servicing activities include collecting or recovering payments, specified renegotiation, complaints administration and informing borrowers of changes in interest rates or charges or of payments due (Article 3(8) and (9)(a)-(d)). Rather than simply designating servicers as credit institutions or financial institutions, the Directive subjects them to a bespoke authorisation and supervisory regime (Articles 4-14 and 21-23).

At the same time, the regulatory object remains connected to the underlying credit relationship. Article 10 of Directive 2021/2167 regulates the conduct of purchasers and servicers towards borrowers, requiring good faith, fairness and professionalism and prohibiting harassment, coercion and undue influence. It also requires specified information to be communicated after transfer and always before the first debt collection. The limited prudential concerns associated with credit purchasers do not, however, displace the conduct obligations applicable to them and their servicers. Yet Union and national consumer-protection rules continue to apply and borrowers retain the rights arising from the initial credit agreement. Regulatory identity changes; legal continuity remains.

This also separates different forms of risk. Whether an undertaking assumes the economic loss resulting from non-payment may be highly relevant to prudential classification. It says much less about the conduct risks created by controlling communications with borrowers, repayment negotiations and collection. A servicer’s absence of credit risk may therefore matter to one regulatory classification without establishing that its activities are unrelated to the underlying financial relationship. Undertakings might pose little or no systemic risk, but that does not mean no risk to consumer-debtors. Directive 2021/2167 responds to such conduct risks through authorisation and conduct obligations tailored to servicing rather than lending.

It should be emphasized that the Directive is selective evidence, not a complete solution. Its framework concerns non-performing credit agreements issued by credit institutions established in the Union and contains significant exclusions (Articles 1-2). It does not create a common regime for the collection of rent, utility debts or other non-credit claims. That limitation leaves intact a problem I identified in my earlier work: if protection against abusive collection depends on the regulatory identity of the debt or collector, comparable collection practices may attract different legal responses for reasons having little to do with the risks those practices create.

The answer is not to declare every collector a financial institution, but to resist allowing institutional definitions to determine questions they were not designed to answer. Prudential classification should follow prudential purposes. Consumer-facing collection should remain subject to effective conduct controls because of the activity and its connection to the underlying obligation. Where gaps remain outside the sectoral regime, they require direct regulation rather than conceptual disappearance.

Credit does not stop being finance at a single identifiable moment. Default, assignment and outsourcing redistribute its functions and risks. They may change the applicable service, institution and regulatory regime without erasing the relationship between the claim being collected and the transaction that created it. IFIS requires purpose-specific exclusions to be taken seriously. Kosmiro, Gelvora and INKO require equal caution before turning such an exclusion into a general separation between finance and collection. Trapeza II is an opportunity to articulate both propositions and to clarify a boundary that EU law has so far defined only imperfectly.

Two Regimes of Recovery: Greece’s Exit from Macroeconomic Surveillance and the Social Findings That Do Not Close a File

 Joanna Demopoulou*

 

Photo credit: Mstyslav Chernov, via Wikimedia Commons

 

* Joanna Demopoulou holds a PhD in International Affairs and is a former Executive in Residence at the Geneva Centre for Security Policy (GCSP). Her research examines the legal and institutional architecture of public authority and governance, focusing on how law structures, distributes, and constrains public authority across European and international institutions.

 

 

On live television in late September, Greece’s deputy foreign minister, Haris Theocharis, told a fellow panelist that his party represents “noikokyrides”, householders who manage their affairs responsibly, while opposition parties represent the “kakomoiris”, the unfortunate. He later called the phrase a mistake. The remark is domestic party rhetoric, not the language of any EU institution, and this post does not treat the Union as its author. It is a symptom of an institutional environment that makes it easy, more than a decade after Greece entered its first adjustment programme, to recode continuing hardship as a private failing. That environment is the subject here.

 

An asymmetry, not just two numbers

 

The familiar observation is that macroeconomic recovery and social recovery are different things, and that a country can score well on one while scoring badly on the other. That is true but not new, and by itself it is a complaint about incomplete data, not a constitutional argument. The sharper claim is about institutional force: what each kind of finding is allowed to do. Some of the Union’s indicators can close a file. Others can only describe what remains in it.

 

Under the Macroeconomic Imbalance Procedure, a Member State is placed into one of several formal categories, and movement between them triggers or lifts surveillance, and in the worst case, correction. A finding of “imbalances” carries institutional consequences; it is a status, not a description. A finding of “no imbalances” removes that status and the surveillance that comes with it. Continuing hardship, measured by Eurostat’s indicator of poverty and social exclusion or by the Commission’s own Social Convergence Framework, is real institutional knowledge, produced and published by the same Union. It has, as far as I have found, no comparable status that opens or closes. It accumulates as description.

 

One answer is that this is simply a difference of subject matter: debt and deficits are quantities a procedure can act on directly, while poverty has many causes and no single lever. That is a reason to design a different kind of instrument, not a reason for the Union to have no closing instrument at all. The Social Convergence Framework already collects comparable indicators and already names “overall challenges” for specific countries. What it does not produce, on the evidence available, is a formal status with an identifiable condition for exit. The asymmetry is better described as a feature of institutional design than as a fact of nature about the difference between fiscal and social policy; nothing here requires assuming that anyone chose it deliberately.

 

Closing the macroeconomic file

 

The 2026 European Semester Spring Package (COM(2026) 200), adopted on 3 June, found that Greece, the Netherlands and Sweden “are assessed as no longer experiencing imbalances”. For Greece this followed a finding of “imbalances” in the 2025 cycle. Alongside the Spring Package, a separate Post-Programme Surveillance assessment (C(2026) 4000), whose stated objective is to confirm that a country “maintains its capacity to service its debt”, concluded of each of Ireland, Greece, Cyprus and Portugal in turn that it “retains the capacity to service its debt”. Together these findings close the imbalance status that had attached to Greece and record that, on the Union’s own assessment, it retains the capacity to service its debt.

 

The same spring, Eurostat reported that 27.5% of people in Greece were at risk of poverty or social exclusion in 2025, the second-highest share in the Union after Bulgaria, against an EU average of 20.9%. The Commission’s Social Convergence Framework, published weeks earlier, named Greece among six Member States facing “overall challenges to upward social convergence”, alongside Bulgaria, Italy, Latvia, Lithuania and Romania. Both are Union findings, produced through the same European Semester as the imbalance procedure, by the same Commission, in the same year. Neither carries a formal status with an identifiable condition for exit. There is no announced moment at which a Member State is declared to have exited social hardship the way it exits an imbalance. No scoreboard entry reads “no longer experiencing social exclusion.”

 

A pattern with a history

 

The pattern predates this recovery. It was present at the start of the adjustment programmes, and admitted along the way. The European Parliament’s 2014 resolution on the employment and social aspects of the Troika found that Greece’s programmes rested on “a mistaken assumption regarding the economic multiplier effect, resulting in failure to act in time to protect the most vulnerable against poverty”, and it regretted that “proper impact assessments were not made of their distributional impact on different groups of society”. Parliament called on the Commission to take social indicators into account in any renegotiation of the programmes.

 

The Commission’s own ex-post evaluation, published nine years later in 2023, concluded that the Greek programmes “did not initially perform as expected, but that their results improved over time and their main objectives have been achieved”. The objectives referred to are fiscal and financial: the evaluation’s own executive summary names “fiscal consolidation to ensure debt sustainability” and Greece’s gradual return to sovereign markets as the programmes’ main objectives. The 2023 evaluation neither revisits the 2014 finding about households left unprotected against poverty nor contradicts it. That kind of finding is simply not what a programme evaluation is built to reopen. The two documents, nine years apart, describe the same programmes from within two different regimes of assessment, and only one of those regimes produces a verdict of success.

 

Why this is a constitutional question

 

It would be possible to read all of this as a policy complaint dressed in legal language: the Union should care more about poverty, full stop. The claim here is different. Article 9 TFEU and Article 3(3) TEU are not external moral commentary on economic governance; they are commitments inside the same legal order that produces the Semester, the imbalance procedure and the surveillance reports. This post does not argue that the Treaties require a social counterpart to the imbalance procedure, or that the absence of one is unlawful in itself. The question is why one part of that legal order can close a file while the other, under the same Treaties, can only describe.

 

What the constitutional texts actually commit to

 

The Treaties are not silent on this asymmetry, even if they do not resolve it. Article 9 TFEU requires the Union, in defining and implementing its policies, to take into account “a high level of employment, the guarantee of adequate social protection, the fight against social exclusion”. Article 3(3) TEU sets out objectives of the Union rather than a requirement of that kind: “a highly competitive social market economy, aiming at full employment and social progress”, and the combating of “social exclusion and discrimination”. Neither provision creates a directly enforceable right, and neither did the work in this recovery that the imbalance procedure did.

The Charter of Fundamental Rights binds Member States only when they implement Union law. Article 1 on human dignity and Article 34 on social security therefore cannot be read as a freestanding review of a minister’s remark, or of Greek welfare policy generally. But the Union’s institutions were not outside the Charter during the adjustment era. In Ledra Advertising, the Court of Justice held that the Charter is addressed to the Commission even when it acts within the framework of the European Stability Mechanism, and that the Commission, under Article 17(1) TEU and Article 13(3) and (4) of the ESM Treaty, must ensure that a memorandum of understanding is consistent with the fundamental rights the Charter guarantees. That reasoning concerned Cyprus, but the same provision of the ESM Treaty governed Greece’s third programme, financed through the ESM from 2015. The Commission negotiating that programme’s conditionality was, on the Court’s own logic, not acting in a Charter-free space. The constitutional commitments were therefore legally live while the programme took its social shape. What they did not generate, then or since, was a closing mechanism comparable to the one that, a decade later, closed the macroeconomic file. A Member State’s exit from the imbalance procedure is announced by name, in a document with legal consequences. I have found no act that records how the requirements listed in Article 9 TFEU, which the Union must take into account in defining and implementing its policies, were weighed in reaching that finding.

 

From institutional story to personal blame

 

None of this explains a minister’s choice of words; it describes the space in which that choice becomes intelligible. When the dominant institutional narrative is that a country has recovered, continuing difficulty is harder to read as an unfinished collective problem and easier to read as an individual one. If the country succeeded, the reasoning runs, whoever is still struggling did something wrong, or belongs to the group the recovery was never going to reach. “Noikokyris” and “kakomoiris” give that reasoning a name; they do not invent it. Pressed on air by an opposition MP, Tonia Antoniou, Theocharis said his party represents both the “noikokyrides” and people who “get by with difficulty”, which kept the division intact even as he tried to soften it.

 

“Recovered” is not a loose figure of speech picked up by commentators. It is the closest available paraphrase of what the Semester’s own category change says about Greece, a status built on indicators designed to answer exactly that question. Against that background, any remaining hardship needs a separate explanation, and “noikokyris/kakomoiris” supplies one.

 

This is a diagnosis about institutional design, not an accusation against any single actor. Those who speak of recovery are not wrong about what they measure: the debt ratio has fallen, and Greece’s return to market financing is real. The claim is narrower. An architecture that lets one kind of finding close a file and confines another to permanent description makes it easier for political language to treat the second as somebody’s personal failing. Nothing here shows that the architecture causes that language; it shows what it leaves open for that language to do.

 

Several Union reports on poverty already exist, so another one is not the remedy. The task, instead, is to ask, as an open constitutional question rather than a ready design, what it would take for the Union’s own commitments in Article 9 TFEU and Article 3(3) TEU to carry institutional weight of a kind comparable to the imbalance procedure’s, and what would have to be true before a Member State’s file could be said to reflect both. This post does not answer that question; it argues that the gap in weight, not merely in numbers, is where the answer has to start. Until that changes, the Union will keep producing two kinds of knowledge about the same country, and the difference between them lies less in what the Union knows than in what each kind is institutionally permitted to do: one can end a crisis, and the other can only keep describing it.

How to balance freedom of expression and child protection in the digital age? A comparative legal analysis

Mohammed Abbas Mohsin, PhD in Public Law, Diplomatic advisor 

Photo credit: San Jose Public Library, via Wikimedia Commons

  Abstract

    This blog post presents a comparative legal analysis of the regulatory and judicial frameworks governing the balance between the fundamental right to freedom of expression and the state’s pressing interest in protecting children in the digital age. By examining recent landmark decisions handed down by the Supreme Court of the United States, the Court of Justice of the European Union, the French Council of State, the Italian Regional Administrative Court, and the Supreme Court of England and Wales, this paper analyses how the various judicial authorities in those countries reconcile the requirements for age verification on digital platforms with constitutional safeguards and human rights. This post explores the shift in levels of judicial scrutiny, the extent to which procedural requirements under internal market rules (such as the country-of-origin principle) apply, the unique technical and operational challenges faced by collaborative open-source platforms, and the emerging international consensus that, whilst child protection is of the utmost importance, the regulatory measures employed must remain carefully tailored and procedurally compliant to avoid disproportionate restrictions on adults’ freedom of expression and privacy.

    1. Introduction

    The rapid shift from physical to digital media has led to a radical and evident change in minors’ exposure to sexually explicit content, rendering the traditional restrictions imposed on high-street shops ineffective and unable to keep pace with this problem. Governments around the world have responded by enacting comprehensive age-verification laws targeting online commercial platforms. However, these laws inevitably conflict with the fundamental rights of adult users – who have a constitutional right to access age-restricted content intended for adults – and with digital platforms that rely on the free flow of information.

The key legal conflict lies in determining where the state’s regulatory authority to protect the innocence of minors ends, and where the unlawful restriction of adults’ rights begins.

  This comparative legal analysis examines how various democratic legal systems have addressed this tension. Specifically, it reviews the standard of scrutiny applied to digital age verification laws, the impact of federalism and cross-border economic frameworks, and the unique challenges faced by non-profit cooperative structures in this regard.

 

 

2. The US Constitutional Model: Scrutiny and Incidental Burdens

   The American approach is characterised by the Supreme Court’s strict interpretations of the First Amendment to the Constitution.

  In the case of ‘Freedom of Expression Coalition v. Paxton 2005 ’, the US Supreme Court considered an apparent constitutional challenge to Texas House Bill 1181 (H.B. 1181), which requires commercial entities that distribute sexually explicit content harmful to minors to implement reasonable age verification measures .

   The central constitutional debate in the Paxton case centred on the appropriate level of judicial scrutiny to be applied by the Supreme Court.

   Traditionally, regulations concerning content protected by freedom of expression are subject to strict scrutiny, the most stringent standard in constitutional law, which requires the government to demonstrate that the law is the least restrictive means of achieving a compelling .

   However, the majority opinion, delivered by Justice Thomas, held that Bill No. 1181 was subject only to intermediate scrutiny. in its ruling, the Supreme Court held that the state possesses the traditional authority to prevent minors from accessing speech which it deems obscene from its own perspective; consequently, adults are not entitled, under the First Amendment to the US Constitution, to access such speech without first providing proof of their age.

    Based on the principle established by the Supreme Court in United States v O’Brien, 1968 the majority described the age verification requirement as an ‘incidental burden’ on adults’ right to protected freedom of expression, rather than an outright prohibition

   Under the intermediate standard of judicial review (the intermediate constitutional scrutiny standard), the law was upheld because it advances a substantial government interest unrelated to the suppression of freedom of expression, and does not impose a greater burden on freedom of expression than is necessary.

    The Supreme Court noted that the requirement for age verification is a traditional, widespread and legitimate means of enforcing age restrictions, similar to the purchase of alcohol or firearms

  In a sharp dissenting opinion, Justice Elena Kagan—joined by Justices Sonia Sotomayor and Ketanji Brown Jackson—criticised this analysis, describing it as “entirely circular”. Rather than asking the correct constitutional question —“Does this law impose an unjustified burden on protected adult rights?”—

   the majority defined the constitutional right of adults from the outset as “a right subject to age verification”, and then logically concluded that the law did not violate that right; this is a clear circular logical fallacy that assumes the validity of the conclusion to be proven in order to evade the application of the strict scrutiny standard.

       The dissenting opinion (written by Justice Kagan) held that Texas House Bill 1181 imposes unjustified restrictions on the constitutional rights of adults by virtue of its communicative content—as a direct, content-based regulation—and must therefore be subject to strict scrutiny, in accordance with decades of case law, including the cases of Sable,  Playboy,  Reno and Ashcroft II.

  Justice Kagan emphasised that requiring adults to provide sensitive personal information in order to access controversial content imposes an unacceptable ‘chilling effect’, particularly on the internet, where users fear for their reputation, The dissenting opinion was written and drafted by Justice Elena Kagan and joined by Justices Sonia Sotomayor and Kay Tang Brown Jackson. It held that the law must be subject to strict scrutiny, with Justice Kagan and her colleagues arguing that the Texas law (H.B. 1181) for age verification constitutes a direct content-based restriction on adults’ constitutionally protected freedom of expression, thereby requiring it to be subject to the highest standard of judicial scrutiny in order to protect fundamental rights. Consequently, the dissenting opinion affirmed that if there is a less restrictive alternative (such as user-side content filtering) that could protect children to the same extent, the state must adopt it rather than unnecessarily burdening adult expression .

  

3. The Organisational Structure of the European Union: Market Integration and the Country of Origin Principle

   Unlike the dualistic framework based on the principle of separation between freedom of expression and the state’s freedom in the US Supreme Court’s case law, the European Union’s approach is closely intertwined with the rules of the internal market.

  Under Directive 2000/31/EC (the E-Commerce Directive), the ‘country of origin’ principle ensures that information society services are regulated exclusively by the Member State in which the provider is established, thereby preventing the targeted Member States from imposing additional regulatory burdens.

    In its landmark judgment delivered by the Grand Chamber on 16 June 2026, in the case of WebGroup Czech Republic and Others (joined cases C-188/24 and C-190/24), the Court of Justice of the European Union clarified the limits of Member States’ authority to regulate the protection of minors in the digital sphere.

    The Court of Justice of the European Union held that Articles 2(h) and 3 of Directive 2000/31/EC preclude any Member State from imposing a general and abstract obligation under criminal law to prevent minors from accessing pornographic content from service providers established in other Member States.

     These sweeping criminal obligations undermine the country-of-origin principle by going beyond the procedural and substantive safeguards set out in the Directive.

   These sweeping criminal obligations undermine the country-of-origin principle by going beyond the procedural and substantive safeguards set out in the Directive.

  However, the Court of Justice of the European Union has ruled that, provided there is strict compliance with Article 3(4) of Directive 2000/31/EC, Member States are not precluded from adopting measures requiring specific service providers to establish age-verification systems, provided that those providers have not implemented ,appropriate preventive measures under Article 28b of the Audiovisual Media Services Directive .

     These individual exceptions must be interpreted in the light of Article 1 (human dignity) and Article 24 (rights of the child) of the Charter of Fundamental Rights of the European Union .

   In doing so, the Court of Justice of the European Union established a regulatory framework: general, unilateral restrictions on targeted countries are prohibited, but age verification orders directed at specific platforms may be issued provided that the requirements for procedural consultation and notification are met.

  

4. Transposition of laws and litigation between Member States: France and Italy

   The practical contradiction between protection policies in the target countries and the European Union’s Internal Market legislation is clearly illustrated by recent administrative proceedings in Italy and France.

 4.1 The French Council of State (Conseil d’État)

   France has transposed the requirements for the protection of minors through amendments to Article 10 of Law No. 2004-575 on strengthening confidence in the digital economy, empowering the French national regulatory authority (Arcom) to impose age verification standards.

    In the case of Webgroup Czech Republic and NKL Associates (10 June 2025), the Council of State considered a priority constitutional question (QPC) challenging Article 10-2 of this Act.

   The applicants argued that the Act violates the principle of the personal nature of penalties because it penalises publishers for failures in the age verification system caused by third-party ‘trusted identity’ providers who operate the ‘double anonymisation’ protocol .

   For its part, the Council of State refused to refer the request for approval of age verification to the Constitutional Council, considering that the obligation to ensure age verification in accordance with the law remains a personal responsibility incumbent upon platform operators, and that the exclusion of third-party service providers from the direct scope of Arcom’s enforcement does not raise a fundamental constitutional issue.

    Shortly afterwards, in the case of Hami Media Ltd (xHamster) v Minister of Culture (15 July 2025), the Council of State quashed a court order that had suspended the ministerial decree

subjecting the xHamster website (hosted in Cyprus) to French age verification legislation.

   The Cypriot provider claimed that French law would cause severe economic losses as a result of user migration and a breach of the privacy rights of adult users.

    Under Article L. 521-1 of the Code of Administrative Justice, the suspension of a licence requires proof of an ‘emergency situation’ that outweighs the public interest. The Council of State held that the commercial and privacy-related objections did not satisfy this condition when weighed against the overriding public interest in protecting children from explicit pornographic material .

  

  4.2 The Italian Regional Administrative Court (TAR Lazio)

  In Italy, the Caivano Decree (Article 13-bis of Legislative Decree 123/2023) a general obligation for age verification on pornography platforms operating in Italy, delegating the technical rules to the Italian Communications Authority (AGCOM) .

    The Italian Communications Authority (AGCOM) issued Circular No. 96/25/CONS and subsequent lists of platforms subject to the obligation, thereby extending the age verification requirements to include EU service providers based abroad, including Aylo Freesites Ltd (registered in Cyprus) .

   As a result of this  Aylo challenged the proceedings on several grounds, including breaches of the data minimisation principle set out in the General Data Protection Regulation (GDPR) and the country of origin rule.

   The Lazio Administrative Court of Appeal dismissed the complaints relating to the General Data Protection Regulation and the substantive complaints, concluding that the Communications and Information Technology Authority’s ‘technologically neutral’ standards, adopted following consultation with the Data Protection Authority, respect the principle of data minimisation by prohibiting the retention of identity documents.

   However, the court annulled the automatic extension of the requirements to foreign platforms established in the European Union, as the Italian Communications Authority (AGCOM) had completely disregarded the mandatory procedural requirements laid down in Article 3(4)(b) of Directive 2000/31/EC .

   Under this rule, before any Member State can restrict an information society service from another Member State, it must first formally request the home Member State (Cyprus) to take precautionary measures and notify the European Commission. 

  As the Italian Communications Authority (AGCOM) had not followed this procedure, the Italian court partially annulled the measure, emphasising that host states may not unilaterally impose restrictions on freedom of expression across the borders of the European Union without due regard for the rule of law.

 

5. The United Kingdom and the dynamics of collaborative platforms:

The Online Safety Act 2023

    The ongoing debate in the United Kingdom highlights how broad legislative definitions of digital technologies can inadvertently threaten non-profit digital infrastructures that serve the public interest.

   Under the UK’s Online Safety Act 2023 (OSA), ‘Category 1’ online services are subject to extensive legal obligations, including a duty to ‘empower users’ under Section 15 to allow users to filter out content from unverified accounts .

     In the case of Wikimedia Foundation and Others v Secretary of State for Science, Innovation and Technology (11 August 2025), the host of the Wikipedia website challenged the threshold regulations set out in the Online Safety Act. The Wikimedia Foundation argued that Wikipedia’s collaborative editing model – in which anonymous and attributed edits are dynamically merged – is functionally incompatible with an identity verification mechanism.

    Isolating edits from unverified users would render the encyclopaedia unusable, which would constitute a breach of Article 10 (freedom of expression) of the European Convention on Human Rights.

  The High Court, presided over by Judge Johnson, dismissed the European Convention on Human Rights claims as premature.

   Applying the strict ‘victim status’ criteria reaffirmed by the European Court of Human Rights in the case of ‘Swiss Anti-Espionage Association v. Switzerland’, the court ruled that, as the Office of Communications (Ofcom) had not finalised the Category 1 listing or classification of Wikipedia, the applicants had failed to establish a ‘risk of a future breach’ with sufficient concreteness.

     As regards the substance of the challenges relating to the interpretation and rationality of the law, the court upheld the Secretary of State’s regulations, ruling that the Secretary of State had acted rationally in relying on high-quality research carried out by Ofcom, and that the legislature was entitled to set objective and broad criteria rather than ad hoc decisions.

 

6. Comparative Analysis and Judicial Interpretation

    Comparative analysis and judicial interpretation reveal differing legal classifications of the same legal issue and the technical burden involved.

   In the United States, age verification is considered an incidental, content-neutral burden and is assessed under moderate judicial constitutional scrutiny, thereby reducing the constitutional obstacles facing the states .

   While in the European Union, whilst the substantive right to the protection of minors is highly valued, the legal system focuses on the country-of-origin rule and the strict two-tier procedural approach under Article 3(4) of the E-Commerce Directive . 

  Administrative cases in France and Italy show that platforms established in EU Member States (such as Cyprus) are protected from abuses by the host state, unless those states fully comply with the requirements for notification and consultation.

   Finally, the UK’s experience serves as a warning that a failure by legislative bodies to narrow the scope of definitions may ensnare non-commercial collaborative platforms serving the public interest, such as Wikipedia, thereby threatening to fragment the open internet under the guise of safety.

 

7. Conclusion

Striking a balance between freedom of expression and child protection in the digital age is not merely a technical challenge, but a matter central to the constitutional framework.

 Recent judicial consensus agrees on the urgent need to protect minors. However, the legal approaches taken vary considerably: the United States applies a lenient standard of scrutiny with regard to ‘incidental’ restrictions, whilst the European Union imposes a strict procedural framework to preserve the single digital market; administrative courts exercise close oversight of regulatory compliance.

  As digital platforms continue to evolve, the challenge remains to implement verification systems that are secure, privacy-preserving and procedurally sound, protecting children without fragmenting the global flow of knowledge or restricting adults’ freedom of expression online.