Why Do Settlements Matter?
In the practice of mass litigation, collective settlements play a major role. Empirical evidence shows that the majority of collective disputes are not adjudicated on the merits, but instead are resolved through amicable settlements. High-value settlements are increasingly observed in digital-rights litigation, as illustrated by the notable USD 1.5 billion Anthropic AI piracy settlement, in which the company allegedly downloaded books from piracy websites to train its AI models (U.S. District Court of the Northern District of California, 17 October 2025, preliminary approval).
Despite this reality, the Representative Actions Directive (RAD) as the EU’s recent answer to collective harm and focal point of this blog post, provides only rudimentary procedural safeguards to ensure fair and balanced outcomes for represented consumers. By contrast, the U.S. class action system has developed a sophisticated regime of judicial settlement review, offering valuable insights for the European context.
The Law and Economics of Digital Rights Litigation and Collective Settlements
If damages arise from an infringement relating to digital rights infringements (e.g. personal data, copyright infringements due to AI training as in the Anthropic settlement), individual claims are typically small, leading to scattered damages and prohibitive litigation cost.
Another economic problem when enforcing privacy rights is to measure the value of personal data (valuation dilemma), particularly when non-material damages are to be compensated.
Those peculiarities of digital rights litigation can be remedied best by collective action, but in many cases even better by collective settlements. Compared to lengthy and costly litigation, whose outcome can feel like rolling the dice, such amicable solutions are more economical and expeditious, while also making the outcome of the dispute more predictable for the litigants (as they can negotiate the settlement terms and decide whether or not to accept a settlement offer).
On the flip side, complex legal and factual cases must not become subject to a “bazaar-like” trade, detached from justice. Therefore, robust procedural safeguards are essential.
Compensation for Digital Rights Violations—Persisting Unclarity
Digital rights violations—whether arising from data protection infringements, the misuse of AI systems, or breaches of the DMA or DSA—call for collective redress. The resulting harm is typically dispersed, diffuse, and often intangible; it implicates fundamental rights and reveals a structural enforcement gap. In the GDPR for example, its enforcement largely hinges on public sector oversight, but it also establishes a civil law basis for compensating both material and non-material damages resulting from infringements of the Regulation (art. 82). This aligns with the ex-post, individualist, harm-based fundamental rights logic underlying European data protection law. Despite the unified civil law basis for data protection violations in the GDPR—and some guidance provided by the Österreichische Post-CJEU judgment of 4 May 2023—the determination of the amount of damages is partly still governed by the national laws of the respective Member States, leading to an unfavorably fragmented legal landscape (in particular due to their cross-border nature).
In the current and highly debated context of AI regulation, liability questions have also come to the forefront. Unlike the GDPR, the EU Artificial Intelligence Act (2024) does not establish an autonomous civil law basis for AI-related liability, which can be explained by the regulatory concept of the AI Act. It follows a risk-based, ex-ante compliance model. Nevertheless, recent disputes—such as the copyright litigation involving Anthropic—illustrate that liability scenarios are not merely hypothetical but practically inevitable. As AI systems operate at scale and affect large groups of right holders or users simultaneously, the structural case for collective litigation remains compelling. Nevertheless, discussions on dedicated civil liability regimes for AI have stalled in the EU and it is unlikely that this political discussion will restart soon.
Collective settlements can offer pragmatic, approximate responses to these persistent uncertainties, but legislators and courts must develop more precise and principled solutions in the long run. As litigants dispose mandatory fundamental rights and data protection entitlements, when concluding collective settlements, fairness reviews are indispensable to ascertain that settlements are in line with EU law.
Comparative Law Perspective on Settlement Procedures
The U.S. class action is a global forerunner in mass litigation, and mass settlements in particular. The Federal Rules of Civil Procedure (Rule 23) provide for a strict settlement court control, complemented by procedural tools to enrich the court’s decision-making basis (e.g. fairness hearing, group members’ right of complaint).
The Netherlands followed this example of amicable mass settlements in 2005, with a consensual group settlement procedure (Wet collectieve afwikkeling massaschade, WCAM). In 2016 the WCAM was amended by an adversary collective redress procedure (Wet afwikkeling massaschade in collectieve actie, WAMCA), which was subsequently adapted to comply with the requirements of the RAD. Despite the WAMCA’s adversarial nature, it remains true to the WCAM’s underlying idea that mass claims are often most effectively resolved through settlement, and judges accordingly encourage the parties to reach an agreement.
The RAD aims to promote collective court settlements and regulates them in art. 11. It is mandatory for the settlement to be reviewed and confirmed by the court (art. 11 para. 2). It must be examined whether the settlement “contradicts mandatory provisions of national law or contains conditions that are not enforceable”. The Member States can extend this review standard by a fairness test.
The German implementation of the RAD’s core element is the Consumer Rights Enforcement Act (hereinafter VDuG). Peculiarities of the VDuG are its multiphased structure and the financing cap, according to which litigation financiers may not be promised a profit share exceeding 10 percent (§ 4 para. 2 no. 3 VDuG).
Under the Austrian Verbandsklagen-Richtlinie-Umsetzungs-Novelle (VRUN), the provisions on collective settlements (§ 631 Austrian Code of Civil Procedure) are rudimentary and even the fairness test was waived, as the Austrian legislator assumes that the qualified entity and the trader will “regularly arrive at a fair result on their own”.
Court Review of Digital Rights Settlements
What distinguishes a fairness test from a mere review of lawfulness is its sensitivity to the specific characteristics of collective redress. In this context, market forces among the negotiating parties cannot be reliably expected to produce a fair bargain, as collective settlements are a textbook case of information asymmetries and agency dilemmas. Consequently, where a fairness test is omitted—as under the Austrian VRUN—crucial dimensions of judicial settlement control fall by the wayside.
Under the fairness standard, it is essential to examine the substantive value of a claim. An option is to compare the settlement offer with the hypothetical result of a court judgement. For calculating the “fair” settlement value, the net expected value method may be applied where the settlement offer is multiplied by the probability of success in a performance action, and the hypothetical litigation costs are deducted.
Yet, U.S. courts have also employed benchmarking—comparing proposed settlements to those in similar class actions—as in the notable Apple-Siri privacy litigation (U.S. District Court, Northern District of California, 14 October 2025). However, previous judgments should not be relied upon uncritically; courts ought to use supplementary methods to determine whether the settlement offer is substantively fair.
In Germany—under the earlier KapMuG regime—courts sometimes gave settlement approvals short shrift, as demonstrated by the Constantin Medien case (Constantin Medien AG, Oberlandesgericht München [Higher Regional Court of Munich], Sept. 18, 2014, 5 Kap 2/09).
Out-of-Court and Injunctive Settlements
In European collective actions one frequently encounters out-of-court settlements. The notable Dieselgate cases in both Germany and Austria, for instance, were concluded outside judicial oversight. Similarly, the Österreichische Post case was resolved through an out-of-court agreement.
Such out-of-court settlements lack the watchful scrutiny of the courts. Standalone settlement procedures, such as the WCAM, safeguard fairness not only from a substantive perspective (with the fairness test), but also from a procedural point of view (e.g. class member objections, fairness hearing, qualified entities independence, transparency of distribution). Such settlement procedures are advantageous, as they encourage – or even nudge – parties to resolve disputes through judicially supervised settlements.
Quantification of Damages & Non-Material Damages
A common obstacle in privacy litigation are non-material damages, such as loss of control over personal data, or emotional distress. There is case law emerging from individual redress cases, e.g. the German Federal Supreme Court ruled in 2024 in the Facebook data leak/scraping litigation, that non material damage for loss of control over one’s own personal data “in the range of 100 euros” was appropriate. The recent CJEU judgment Quirin Privatbank AG of 4 September 2025 further clarified that negative feelings may constitute non-material damage, but data subjects bear the burden of proving that such feelings were actually experienced. When assessing the fairness of collective settlements, this case law should be taken into account, besides the above-mentioned fairness review methods.
In-Kind and Cy Près-Settlements
A particularly distinctive feature of U.S. data privacy class actions are cy-près settlements. In such cases, either residual or unclaimed funds—or in some instances, the entire compensation—are allocated to charitable organizations whose work serves the objectives of the litigation. This can be well illustrated with the Google Buzz privacy litigation (United States District Court Northern District Of California, San Jose Division, 3 September 2010), where Google agreed to pay USD 8.5 million to data protection organizations, as distributing compensation to millions of affected users would have resulted in payouts of only a few cents per person. Despite their practical advantages, cy-près settlements are not beyond reproach, particularly as they might undermine the compensatory function of tort law.
An extraordinary “creative” in-kind settlement arrangement can be found in the Clearview AI settlement, where billions of facial images from the web were scraped and sold information without consent (U.S. District Court for the Northern District of Illinois, Eastern Division, 21 June 2024). The victims were offered an equity stake, or alternatively a cash payment equal to 17% of its revenue since the settlement approval date. Turning harmed data subjects into shareholders of the very company that infringed their rights—or inviting them to participate in its profits—raises ethical concerns and thus represents a paradigmatic case demonstrating the necessity of robust judicial oversight of collective settlements.
Tentative Conclusions
Collective court settlements lower transaction costs, mitigate the valuation dilemma, and facilitate rapid compensation. Yet, without adequate judicial fairness review, settlements risk devolving into rough justice, trading accuracy for expedience.
U.S. class action settlements have developed effective—and occasionally innovative—mechanisms to govern digital rights violations, and the Netherlands have absorbed many valuable transatlantic lessons. By contrast, the Austrian and German implementations remain cautious. Limited review standards, and a lack of procedural incentives for court settlements, leave significant gaps in consumer protection, transparency and democratic legitimacy.
Ultimately—if efficiency is reconciled with fairness—collective settlements hold the potential to make digital markets more accountable, rights more enforceable, and European private law more responsive to the realities of the digital age.
This blog post builds on our lecture presented at the Collective Redress and Digital Fairness Conference at the University of Amsterdam (10-11 December 2025).
(Photo: Max Harlynking)
