1. An expanding repertoire
Regulatory experimentalism is becoming increasingly relevant in the current phase of the European integration process. The relevant European Union (EU) instruments differ in form and function, but they all treat regulation as an iterative practice oriented to promoting innovation through regulatory learning rather than as a set of command-and-control tools aimed at verifying compliance with a number of predefined standards. Sandboxes promise to create supervised environments in which innovative activities may be tested under tailored legal conditions. Living labs move experimentation into real-world settings and bring public authorities, researchers, firms and users into the same process. Testbeds aim at providing controlled technical spaces, while pilot regimes allow rules or administrative arrangements to be tried before they are generalised. Taken together, these devices respond to a common problem: how to regulate innovative technologies whose risks, uses and social consequences cannot be fully known in advance. Unsurprisingly, they are now presented by the Commission as elements of a coherent regulatory strategy.
The breadth of the current repertoire is striking. Sandboxes are envisaged for net-zero technologies,Footnote 1 artificial intelligence (AI) systems,Footnote 2 digital financial servicesFootnote 3 and innovative medicinal products for human use.Footnote 4 EU-funded projects extend the same logic beyond formal sandboxes. Living labs may test technologies in cities, farms, neighbourhoods and other ordinary environments, producing evidence that can later inform policy and legislation. For example, the UP2030 Project, which is part of the Horizon 2020 programme, uses urban planning to explore climate adaptation and mitigation, while the Sunrise Project, co-financed by the EU, relies on a network of living labs engaged in the development of novel agroecological solutions addressing key agronomic issues. As for testbeds, they are used in a variety of different sectors, ranging from transport to innovative health, food production and green technologies – for example, the SHOW project, funded by Horizon 2020, has tested autonomous mobility, including robo-taxis and freight vehicles, across 20 European cities, while a number of projects have been financed to support the establishment of open innovation test beds for the development, testing and upscaling of renewable hydrogen.
2. Continuity and change: developments in search of interpretation
There is, of course, nothing entirely unprecedented about experimentation in the EU legal order. The EU has long relied on non-conventional regulatory approaches. As early as the beginning of the 2000s, Sabel and Zeitlin identified experimentalist governance in the development of European administrative networks and in the open method of coordination.Footnote 5 Héritier and Rhodes examined the effectiveness of these new modes of governance.Footnote 6 Later scholarship turned to experimental legislation, sandboxesFootnote 7 and sector-specific practices – for example, Galán and Svetiev have read the peer-review mechanisms operating within the system coordinated by the European Securities and Markets Authority as a relevant form of experimentalism in the field of financial supervision.Footnote 8
In the last few years, both the rationale underlying experimentation and the level of generality at which it is now presented have changed. During the first decade of the twenty-first century, experimental regulation responded to a range of concerns, from the need for flexible, reflexive, adaptive and risk-based forms of regulation beyond traditional command-and-control approaches to the search for ways to advance the construction of a composite EU administrative system – an arrangement of ‘distributed public governance’ bringing together EU, national and mixed authorities. Recent developments retain the earlier commitment to reflexive, responsive and iterative regulation, yet they amount to a qualitatively distinct mode of regulation, characterised by a specific orientation towards promoting innovative technologies in a balanced and controlled manner. The regulation of innovation is both their central rationale and their distinguishing feature. Moreover, these developments are no longer framed merely as sectoral exceptions or pragmatic responses to individual problems; rather, they belong to a broader search for regulatory techniques that supplement traditional instruments, including command-and-control, meta-regulation, risk-regulation and regulatory science. Sandboxes in particular are spreading across policy fields and are being treated by the Commission as elements of a coherent strategy rather than as unrelated devices.Footnote 9 The significance of the present moment therefore lies in the shift from experimentation as a dispersed practice to experimentalism as an explicit regulatory programme.
The rise of regulatory experimentalism deserves the attention of European legal scholarship. It is appropriate to discuss its underlying rationale, its basic regulatory option and its place within the EU legal order. The position taken in this Editorial is that the Commission has supplied a plausible strategic direction, but not yet an adequate legal architecture. The future of EU regulatory experimentalism will therefore depend less on further rhetorical endorsement than on careful institutional micro-design and harmonious positioning in the EU legal system.
3. Why experiment? Innovation, competitiveness and managed disruption
A. The Commission’s policy strategy
The current experimental turn is integral to the reorientation of the EU political agenda around innovation and competitiveness. The Commission does not treat experimentation as an end in itself; it presents it as an instrument for reducing the distance between technological change and the legal framework within which that change occurs. Innovation is expected to support productivity, industrial renewal, the green and digital transitions and, ultimately, the Union’s capacity to act in a more demanding geopolitical environment. Regulatory experimentalism is thus embedded in a larger policy narrative: Europe must become more innovative, and its legal order must learn to accompany innovation without abandoning public protection.
The Commission’s approach is visible across successive policy documents. The New European Innovation Agenda identified deep-tech innovation as a priority and expressly included experimental regulatory approaches among the framework conditions that could support it.Footnote 10 The Commission’s 2023 Staff Working Document then brought sandboxes, testbeds and living labs together under the heading of regulatory learning.Footnote 11 REPowerEU connected sandboxes to innovation in renewable energy after Russia’s aggression against Ukraine.Footnote 12 The Draghi Report placed the issue within a wider diagnosis of Europe’s competitiveness gap,Footnote 13 and the Competitiveness Compass confirmed the political centrality of innovation in 2025.Footnote 14 Across these texts, regulatory experimentation becomes part of the institutional response to economic, technological and geopolitical pressure.
B. Innovation without deregulation
This orientation should not be confused with ordinary simplification. EU and national simplification programmes usually seek to shorten procedures, remove formalities or rationalise sectoral legislation.Footnote 15 Experimentalism responds to a different need. The aim is not simply to reduce legal burdens, but to create a setting in which regulation may sustain and govern the development of innovative technologies.
The present strategy also differs from earlier forms of European experimentalist governance. The innovations of the 1990s and 2000s sought unconventional ways of implementing public policies in fields where the balancing of public interests was increasingly intertwined with the resolution of strictly scientific questions, as exemplified by EU legislation on chemicals and medicinal products. They also served the objective of advancing the construction of an EU administrative system while avoiding both a direct strengthening of supranational administration and prevalent reliance on the administrations of Member States. Networks, coordination procedures and risk-based administrative processes supplied intermediate solutions between direct and indirect execution. The contemporary strategy responds to another priority: the perceived need to accelerate innovation and reinforce competitiveness. The history of EU experimentalism is therefore one of changing purposes. Non-conventional techniques have been mobilised to solve different institutional problems at different stages of the integration process.
The Commission’s account is most convincing when it rejects the idea that innovation can simply be left alone. New technologies may generate economic and social benefits, but they also displace established practices, redistribute risks and create ethical problems before public authorities fully understand them. The Commission describes innovation as potentially ‘disruptive’: ‘innovations push the boundaries of existing regulatory, economic, and societal arrangements, with sometimes unintended consequences. Innovation developments require the constant attention of policymakers in order to anticipate them and adapt existing legislation and policies.’Footnote 16 In this perspective, regulatory experimentalism is represented as the space in which innovation can be – in principle, at least – facilitated and supported while also taking account of the economic, social and ethical issues raised by the development of the relevant technology.
It is rather easy to agree with this overall perspective. On this view, experimentalism is not a disguised programme of deregulation; rather, it is a method appropriately aimed at combining the promotion of innovation with the protection of the variety of interests that EU law has progressively incorporated into market regulation. Its distinguishing feature is its orientation towards revitalising technological innovation without abandoning the complexity of balancing the interests at stake. That ambition is normatively preferable to a simple opposition between innovation and protection. It nevertheless rests on an empirical assumption that cannot be accepted in the abstract: established techniques such as authorisation, ex ante control and uniform standards may sometimes inhibit useful innovation, but in other settings they may remain indispensable. The value of an experimental instrument must be demonstrated case by case, its normative justification lying in testing technological innovation without degrading the quality of public protection.
The overall design nevertheless raises an obvious question: how exactly is regulatory experimentalism expected to deliver on its promise? How is it meant to promote and govern innovation?
4. Regulatory learning as the organizing idea
A. Bridging the regulator–innovator divide
The Commission’s answer to such question is ‘regulatory learning’. Innovators and regulators are assumed to occupy different positions and pursue different aims in the regulatory field. Innovators possess knowledge of technologies and advanced expertise, but may treat regulation and public interest requirements as external constraints. Regulators are responsible for policy objectives, yet may lack information about how a technology is developing and how existing rules affect it in practice and create obstacles to its development. Experimental instruments are intended to reduce this distance between innovators and regulators by fostering regulatory learning.
The Commission defines regulatory learning as a process bridging the regulator–innovator divide. It is expected to enable competent authorities ‘to gain better knowledge and understanding of the risks and opportunities as well as the need for possible changes to or new interpretations of existing legislation to effectively address new technological developments and enable innovation’.Footnote 17 Its main objective is to introduce ‘a wider evidence-base and agile, participatory and anticipatory elements’ into what the Commission calls the regulatory governance of innovation, with a view to revising the existing legal framework or developing new legislation.
B. A relational account – and its normative ambiguity
This formulation matters. Regulatory learning offers a dynamic and relational account of regulation. It aims both at analysing and governing the multiple interactions that may occur in the regulatory process between three different actors: innovators, regulators and citizens or users. Within this framework, regulation can be represented as an object, a ‘good’ supplied by regulators, demanded by innovators and legitimated with the contribution of citizens.
This regulatory perspective is analytically productive. It allows strategic conduct to be seen more clearly. Innovators aim to develop the technologies on which they are at work and may seek favourable regulatory frameworks to do so. Regulators are called to secure innovation alongside a series of policy objectives that may conflict with it. Citizens and consumers, who are expected to receive the social benefits of innovation and also bear its risks, may attempt to influence the objectives and conditions of an experiment. At the theoretical level, the Commission’s approach carries traces of ‘public choice theory’, particularly the idea that regulation can be analysed as a good around which actors organise their demands and strategies.Footnote 18 This also has some useful practical implications. Experimental settings make the ‘games’ between the involved actors both visible and, at least potentially, manageable. They appropriately open the way to identifying regulatory strategies capable of rationalising their behaviours in the service of controlled innovation.
Yet the same perspective is unavoidably problematic. If regulation is treated principally as a negotiable good, the public purpose of the innovation process may become blurred. This is a major shortcoming of the public choice approach implicitly taken by the Commission. Public-interest theories of regulation – differing one from another but all based on the basic idea that regulation is justified insofar as it is oriented to pursue public interests and goals – would provide a more robust normative foundation.Footnote 19 Experimentalism cannot dispense with that normative anchor. Moreover, the core of the Commission’s regulatory strategy remains underdeveloped: it claims regulatory learning is functionally appropriate and desirable, but it does not really clarify how it should be concretely operationalised.
C. Sandboxes: institutions of learning or negotiation?
The problem is sharpest in regulatory sandboxes. In general terms, a sandbox is a temporary and supervised arrangement, tailored to a specific situation and designed for the controlled development of an innovative technology – usually with a view to possible market deployment. In other words, it provides a flexible but supervised regulatory environment that facilitates innovation while continuing to take account of the many other interests at stake. It permits tailored interpretations, procedural adaptations or limited derogations while leaving the competent authority responsible for the boundaries of the innovation process. This combination of flexibility and supervision explains the appeal of sandboxes.
The European Commission adopts this conventional understanding while emphasising the importance of regulatory learning within sandboxes. Yet its account of regulatory learning is highly simplified. The Commission describes sandbox learning as a two-way process. From the top down, regulators explain the applicable legal framework and invite innovators to identify obstacles and propose regulatory arrangements more in line with the needs of the innovation process. From the bottom up, experience generated during the innovation process itself may reveal the need for new interpretations or legal change. That account is too narrow. A credible learning process must also include those who may be affected without being participants in the experiment. Patients, consumers, workers, local communities and public service providers may possess knowledge that neither the innovator nor the regulator can supply. Their involvement is particularly important where the experiment concerns health, safety, discrimination or access to essential services. Without such participation, a sandbox risks becoming a structured dialogue between the regulator and the innovator – a negotiation – rather than a public framework for mutual learning between all actors involved in the technology-development process.
The example of sandboxes for testing technologies in the field of medicinal products for human use is instructive.Footnote 20 It illustrates both the potential and the incompleteness of the current approach. The legislative proposal presented by the Commission identifies the procedure for establishing a sandbox, beginning with an initiative of the European Medicines Agency and ending with a Commission decision that sets out a plan and duration. It also addresses the relationship between experimentation and marketing authorisation and establishes a liability framework. Those elements are significant: they recognise that the sandbox must be connected to the ordinary legal regime rather than operating outside it. Yet a number of important operational questions are postponed. Eligibility criteria, participants’ rights and obligations, and the practical conditions of the sandbox are left to later Commission implementing acts. The legislature supplies a procedural shell while providing little substantive guidance on how the experiment should reconcile unmet medical needs with technical uncertainty and social and ethical concerns.Footnote 21 In this wider context, it remains to be determined how the objective of regulatory learning – on which, in the Commission’s design, the success of regulatory experimentalism ultimately depends – will be pursued. The proposal speaks of ‘proactive regulatory learning’, enabling regulatory authorities ‘to gain better regulatory knowledge and to find the most effective means of regulating innovations on the basis of real-world evidence, particularly at a very early stage in the development of a medicinal product’. But it does not yet explain how evidence will be selected, how affected interests will be represented, how unsuccessful experiments will be evaluated or how lessons will travel from an individual sandbox to general regulation. Those are not secondary administrative details; they are the institutional structure and dynamic of regulatory learning. They have been left, for the moment, to the Commission and the European Medicines Agency in their capacity as sectoral regulatory authorities – a choice that suggests the process within the sandboxes may easily shift from regulatory learning between all parties involved to negotiation between regulators and industry.
5. Regulatory Experimentalism and the EU Legal Order
A further problem of regulatory experimentalism concerns its position within the EU legal order. The Commission presents the new regulatory techniques as a development consistent with the EU legal system. The picture, however, is less straightforward than it may initially appear.
A. Delegation, sectoral autonomy and control
The legal position of experimental instruments depends partly on where they operate in the policy cycle. Testbeds and living labs generally contribute to the stage of policy formation. A testbed is primarily technical, while a living lab incorporates social actors in the innovation process and operates in real-world contexts. Both may generate knowledge for new legislation or for the adaptation of an existing policy. They make it possible to test technologies that may lead either to the development of new policies or to new initiatives within existing policies. Sandboxes, by contrast, operate in the phases of policy implementation. They may be understood as administrative rule-making instruments of a particular kind, specifically intended to enable well-considered innovation through a framework that is simultaneously flexible, derogatory, participatory and supervised. They translate broad legislative mandates into temporary and individualised regulatory conditions. They contribute to the evolution of the EU ‘administrative state’ into a genuine ‘regulatory state’, increasingly engaged in implementing EU public policies and legislation through administrative rule-making. Precisely for this reason, however, sandboxes are more problematic than other instruments of regulatory experimentalism. They raise the issue of the scope of delegation and the limits on the autonomy of sectoral regulators.
Comparative experience helps to identify the problem. In the United States, the boundaries of the regulatory state have long been defined by non-delegation and judicial deference. In West Virginia v Environmental Protection Agency,Footnote 22 the US Supreme Court held that, under the major questions doctrine, an agency seeking to make decisions of vast economic and political significance must point to ‘clear congressional authorization’ rather than rely on ambiguous statutory provisions. That restrictive approach may reduce the capacity of agencies to respond to novel problems that the legislature could not foresee when it conferred regulatory powers. Disputable as it is, though, it provides a relatively clear criterion for agencies’ regulatory action. EU law has a much less clearly defined legal framework. The relationship between EU legislation and administrative rule-making measures is governed by the principles and rules on the hierarchy and relationship of legal sources laid down in the Treaties, as well as by the case law of the Court of Justice and the Meroni doctrine. Yet the Union has no single doctrine that directly answers how much autonomy a sectoral regulator may exercise when exercising its administrative rule-making tasks.
This is likely to be a problem when EU agencies are called to design a sandbox. EU agencies will be called to make uneasy choices about the substance of their sectoral autonomy and regulatory mandate: what may be adapted, for whom, for how long, on the basis of which evidence and subject to which forms of accountability. This will require, at the same time, careful legal reasoning, institutional creativity, sensitivity to legal change and substantive understanding of technological innovation. It will not be an easy task, and the process might lead to judicial litigation and require intervention by the Court of Justice. The example of sandboxes in the field of medicinal products mentioned above suggests that this is far from unlikely. EU legislation leaves the definition of the operating arrangements and conditions of sandboxes to subsequent Commission implementing acts, thus creating scope for challenges that will probably have to be resolved by the Court of Justice through the identification of a standard governing the regulatory state’s margin of manoeuvre.
B. Experimentalism and the precautionary principle
A second legal issue deals with the precautionary principle. Experimentalism and precaution are not opposites; their relationship depends to a large extent on how the experimental framework is designed. What experimentalism cannot do, however, is treat precaution as legally optional.
Admittedly, the precautionary principle does not prescribe one fixed regulatory response to uncertainty. Richard Stewart’s classical analysis of US environmental decision-making distinguished more enabling and more prohibitory versions of precaution,Footnote 23 and the European debate in the early 2000s likewise revealed a plurality of possible applications.Footnote 24 In the same vein, Craig has appropriately pointed to the need to move away from an abstract search for the one true model of precaution and to analyse the way it has been designed and operationalised in particular sectoral regimes.Footnote 25 The legal meaning of precaution becomes visible in the concrete allocation of burdens of proof, thresholds of evidence, monitoring duties and powers to intervene. It is at that level that one can determine which regulatory model the EU legislature has adopted in each case. It is also possible to recognize how the plural logic of precaution emerges in new fields without the language of precaution being used expressly. The ‘do no significant harm’ requirement, for example, can be understood as an expression of the prohibitory version of precaution in Stewart’s taxonomy.Footnote 26
Flexible and adaptable as it is, the precautionary principle remains inescapable. It is rooted in the Treaties and has been recognised by the Court of Justice as a general principle of EU law. Experimental regulatory techniques may articulate variants of the precautionary principle that are not exclusively restrictive, taking advantage of its plasticity. But they cannot simply disregard it. Nor can they elude it by invoking an autonomous ‘innovation principle’. Such a principle has been proposed in scholarship on European competitivenessFootnote 27 and promoted in Commission-related policy documents,Footnote 28 but it does not presently enjoy any credible legal foundation.
6. From strategic endorsement to legal micro-design
The Commission has so far identified a real and urgent problem in the European integration process – competitiveness – and offered a plausible strategic response: the search for a pro-innovation regulatory approach. This strategic response, however, requires further articulation. The next step is to move to the level of institutional micro-design and engage in the careful and patient design of each experimental regulatory instrument. The mandate, participatory structure and accountability regime are three critical dimensions of any experimental tool. Moreover, the Commission’s regulatory strategy will remain fragile if its legal design is treated as a matter to be left entirely to sectoral authorities. Adelante, therefore – but con juicio. Once again, administrative law proves to be, in Jerry Mashaw’s felicitous formulation, an unceasing and laborious ‘project of institutional design’.Footnote 29
7. In this issue
Curtin and Fia worry about the use of AI in European governance of migration and other fields. More precisely, they worry about the way we cannot know what (if anything at all) exactly is going on, not (just) because of the intrinsic enormous complexity of the technology but because of the seemingly ironclad legal protection of secrecy surrounding it. They unpack the confluence of public secrecy – what they rightly refer to as ‘political secrecy’ – and private secrecy – the patchwork of contractual and commercial legal instruments involved in the public procurement and use of proprietary technologies – and show how these different drivers of secrecy converge and reinforce each other. Be afraid, be very afraid.
A lot has been written about the struggles between the CJEU and the Bundesverfassungsgericht over the ECB’s unconventional monetary policy. But we know very little about the way these cases came to Karlsruhe in the first place, before they made their way to Luxembourg and back. Thierse and Lange make full use of the available material – the original pleadings in particular – to tell the story of how the ‘right to democracy’ was mobilised in an iterative process between the litigants and the constitutional court. This is much-needed solid socio-legal work on legal mobilisation.
Costanzo turns our attention on European citizenship away from the more recent judicial developments and back to its foundations. She argues that EU law has changed the meaning of citizenship to a point where continuing to use an unaltered conception of national citizenship as a yardstick to understand European citizenship is misplaced. She places the core of that transformation of citizenship in the broader depletion of the socio-democratic principles that anchored post-war constitutionalism. Against this background, Costanzo proposes the term ‘European material citizenship’ to express the reordering of social, economic and political relations in a constitutional settlement where the links between the EU institutions, the Member States and citizens are no longer anchored in the socio-democratic principles that still underpin the national constitutions. In her analysis, European material citizenship expresses the new form of national citizenship, which occurred beneath the surface of formal constitutional amendments. There is, however, nothing deterministic about this transformation. In her analysis, EU citizenship can again mutate into a new form, which will depend on the complex dynamics of material forces.
Through revisiting the debates on ‘economic democracy’ or industrial democracy in the EU, O’Shaughnessy reminds us of the pivotal moment in European economic and social history when Keynesianism gave way to a new neoliberal hegemony. The author traces the transition from a Europe at ease with corporatism and even, to a lesser extent, with socialism. From being on the cusp of introducing a German-style system of ‘co-determination’ – that is, employee representation – on company boards across the internal market, O’Shaughnessy tracks the rupture with the old consensus, where what was politically and legally imaginable – by EU institutions, employer groups, trade unions – is now significantly narrowed.
The debate on the automation of decision-making, and the arguably inherent risk it carries of ‘algorithmic bias’, has the potential to take us beyond existing legal notions of ‘discrimination’. Lamprinoudis shows the need for domestic and EU courts to confront more systematically the bias or ‘unfair algorithmic differentiation’ that can occur when decision-making is automated in cases such as border control, taxation, social security, financial services and dismissal of employees. He offers a taxonomy to systematise the fragmented judicial landscape and to bring national courts and the CJEU into conversation on responding to new forms of bias.
Dubos, Safi and Poissonnier on the one hand, and Ippolito on the other, present two opposing views on the role of the action for failure to act (Article 265 TFEU) to challenge the EU’s failure to adopt substantive measures against the risk of genocide committed by Israel in the Gaza Strip, ascertained in early 2024 by the International Court of Justice. Both explore the role and pitfalls of this action in the EU’s system of legal remedies. They each reach a different conclusion, but from different angles they both agree that EU law fails where legal responsibility matters most and that the EU’s lack of position regarding Gaza reflects law’s weaknesses to uphold EU values. There could have hardly been a more sensitive topic to test the justiciability of the EU’s duty to act. These articles are the first of a series of ‘Encounters’ – a new format in our Dialogue and Debate section – designed to open up a discussion on selected EU law matters by confronting two different positions and perspectives on the same issue.
Are we witnessing a (re)turn to activist government in the European Union? The symposium in this issue offers varied viewpoints and answers, brought together skilfully by Dani and Antoniolli in their opening article, which is part introduction, part conclusion, part research agenda and part manifesto. What follows is a series of thought-provoking contributions by an awesome cast of authors for a symposium that is destined to make itself felt in these debates for years to come.
Open access
