On 9 September 2026, the European Commission proposed a Council Recommendation on common principles on regulatory sandboxes (the Proposal). A regulatory sandbox is a supervised space in which companies can test an innovation products in real-world conditions before it goes on the market. Published alongside the European Innovation Act, the Proposal looks like a modest, non-binding exercise in administrative coordination. I argue that the Proposal is more interesting than that – it is a quiet attempt at bottom-up regulatory simplification which could support European innovation by building “legislative infrastructure” with common definitions, procedures and channels for learning by national regulators. And, if Member States and national authorities engage with the implementation of the recommendation, this experimentalist instrument can help build a shared European innovation culture.

Regulatory Sandboxes, Cross-Border Learning and Experimentation

The Proposal aims at enabling regulatory sandboxes as tools for providing businesses (including SMEs and startups) with testing environments and regulatory clarity that facilitates the development of new technologies and accelerates their entry to the market. Crucially, the Proposal emphasises that sandboxes are an experimentation tool and as such are not deregulatory. Instead of repealing rules or engaging in harmonisation in a traditional sense (by unifying substantive rules), the Proposal aims to contribute to the simplification by standardising the use of sandboxes by national competent authorities as a coordinated procedure for deviating from national rules in areas where there is no EU legislation.

Consider, for example, a local municipality that wants to deploy a new flood-detection system. Several startups are developing solutions based on sensors, drones, and AI – and the European Innovation Act’s R&D procurement rules let the municipality run a competitive process to compare suppliers’ progress and award the contract to the best performer. The startup that wins the contract still needs to test its product. Such testing – that often requires deploying the system in the real-world – would mean that the startup has to navigate the permits, certifications, and civil-protection rules that, more often than not, were not designed with such an innovative product in mind. Testing, in other words, is likely to lead our innovative startup straight into the wall of “inconsistent and restrictive regulations” that the 2024 Draghi Report referred to as one of the obstacles behind the European “innovation gap”.

A regulatory sandbox is meant to address exactly this problem. The startup and the competent authorities (e.g. the municipalities and the local water management authority) could trial the system in real conditions under the competent authority’s supervision, on terms that have been fixed in advance (duration of the tests, liability in case of an omission or a false alarm, and the termination of the tests if things go wrong). Sandbox testing does not provide a shortcut for obtaining a permit or a certification, but offers an opportunity for both the company and the regulators to learn about the innovation and the regulatory requirements that apply to its subsequent deployment, to make an informed and evidence-based decision.

The Proposal sets out the common principles for how such sandboxes should operate. It introduces the first common EU definition of regulatory sandboxes, and offers guidance on experimentation, duration, and conditions for participation in a regulatory sandbox, thus responding to the documented challenges of the use of regulatory sandboxes in EU Member States and obstacles to cooperation.

The Proposal also anticipates and supports further increase in the use of cross-border regulatory sandboxes in the EU and responds to the emergence of divergent interpretation and use of the concept of a “regulatory sandbox” across different sectors and among Member States. It also recognises that the action by individual Member States in non-harmonised areas is likely to increase fragmentation and coordination costs for Member States seeking to cooperate. The Proposal is thus offering a framework for a coordinated approach to reduce transaction costs for European firms as well as public authorities using sandboxes.

In addition, a shared approach is expected to “foster collective action” and accommodate “the different degrees of experience and preparedness among Member States in accelerating innovation,” easing joint regulatory experimentation across jurisdictions and reducing costs for firms operating under multiple national regimes. It also builds on the AI Act and the Interoperable Europe Act, which recognise the limits of the national regulatory sandboxes for “cross-border regulatory learning, mutual recognition of outcomes, or the scaling of innovative solutions across the Single Market.” In our flood-detection system example, if the river crosses a border, water management authorities of both Member States could run one joint sandbox and agree in advance who supervises what (based on point 6 of the Proposal), thus building one body of evidence and potentially a better-tested product for public buyers like municipalities.

Legal Basis and Objectives: A Simplification Tool in Disguise?

The Proposal is based on Article 292 and Article 173(3) TFEU. As such, the framework is aimed at contributing to the EU’s and Member States’ obligation under Article 173 TFEU to ensure the necessary conditions for competitiveness of the EU’s industry. As an Article 292 instrument – Council Recommendation – the Proposal does not propose any extension of EU’s regulatory power nor has binding force on the Member States. Member States and their competent authorities thus retain discretion over implementation, based on their national circumstances. I find this design intriguing (in a positive way) for two reasons.

First, the Proposal would be adopted by the Council alone and does not depend on the Innovation Act’s fate. The Act relies on Article 114 TFEU as its legal basis, and was put forward over two negative opinions from the Regulatory Scrutiny Board, and although narrowed before publication, continuously struggles to reconcile its key R&D and pre-commercialisation procurement mechanisms with the Public Procurement Act proposal, and is likely yet to face the fight for its survival. Adoption of the common principles on regulatory sandboxes as a Council Recommendation is both pragmatic and also aligns with the experimentalist governance logic (more on that below).

Second, and more substantively, the Proposal requires Member States to “collect and analyse the results of, and lessons learnt from, the experimentation within a regulatory sandbox” in order to “assess the need to integrate those learnings into the national regulatory framework applicable to the innovation that was subject to the experimentation”. In other words, the Proposal directs the use of regulatory sandboxes towards regulatory (or legislative) change at Member State level to facilitate further innovation. In my opinion, this regulatory simplification logic behind the Proposal deserves particular attention for its potential to redirect the regulatory momentum away from the Commission and towards empowering Member States to simplify their rules and administrative procedures bottom-up. Unlike the Commission’s omnibus packages, which simplify by amending or repealing existing law, the Proposal touches no existing rule. Instead, the Proposal is designed as a residual, horizontal instrument. As such, it applies specifically to “areas where there is no Union legislation”. This sets it apart from the growing number of sector-specific sandbox provisions already embedded in EU law – the AI Act, the DLT Pilot Regime, the Net-Zero Industry Act, the Interoperable Europe Act – which carve out bespoke experimentation rules for a single technology or market (did someone say “simplification”?). Read together with its objectives to reduce coordination costs and encourage cross-border experimentation, this horizontal design appears to address the fragmentation and regulatory rigidity identified in the Draghi report as barriers to innovation. In that sense, I support the reading of the Proposal as an attempt, and an opportunity, to build connective tissue – a shared vocabulary and shared procedure – through which national regulators might, over time, develop something closer to a common European innovation culture.

Legislative Infrastructure for European Innovation Culture

Despite its non-binding nature as an Article 292 TFEU instrument, the Proposal’s recitals focus on the functioning of the internal market, including non-discrimination, free movement, and “unjustified competitive advantages”. This invites a reading of the Proposal as a de facto harmonisation instrument. Belgium made a similar argument in Belgium v Commission (C-16/16 P) unsuccessfully claiming that a Commission Recommendation on online gambling was, in substance, a harmonising measure. AG Bobek’s opinion had argued for exactly that substance-over-form approach, which the Court declined to follow.

I think the opposite is a better reading. Rather than a competence creep, the Proposal – which coordinates only the procedural architecture for departing from national rules, leaving substantive derogation decisions with the Member States’ competent authorities – can instead be read as empowering national regulatory competence. Thus, instead of displacing national-level provisions, the Proposal suggests that further EU-level intervention to address slow authorisation procedures and outdated rules in areas currently not regulated by the EU will not be necessary – conditional on effective cross-jurisdictional coordination in deploying sandboxes. This reading is supported by the text of recital 11 of the Proposal, which “welcomes the Commission’s intention to monitor relevant developments and assess where appropriate whether further measures at Union level may be necessary”. This indirect simplification effort that some call “simplification by pre-empting” should be welcomed from the constitutional perspective as it avoids the “legal vacuum protected by the supremacy of EU law”, that often results from top-down harmonisation measures.

Now, the reader might be wondering whether it will work. The picture is indeed mixed here. For one, the sandbox mandates have multiplied in EU law (the “regulatory sandbox phenomenon”), and so have fragmentation, arbitrage risk, as well as the weak implementation of the existing mandates (as the record of Article 57 of the AI Act to date suggests). Moreover, the proposed framework is non-binding. This, however, should not be seen as fatal, because the Proposal borrows from experimentalist governance: it sets broad goals, allows local actors to experiment, and results are compared and fed back into revised goals. That works when four elements are in place: goals with metrics, local autonomy, reporting and peer review, and periodic revision. The Proposal sets broad goals but no metrics are attached to measure achievement of the stated goals, and while the Proposal provides for reporting obligations, it lacks clear peer review arrangements (beyond a passing reference to the Council Working Party) and any scheduled mechanism for revising the framework itself. Although these omissions may be justified by the instrument’s non-binding nature, they are concerning in light of the broader critique, that evidence for experimentation – and sandboxes in particular – delivering on stated goals for both regulators and firms remains thin even where implemented.

These gaps are not inherent to the non-binding instrument. In my view, the Council could (and should) amend the Proposal to attach indicative benchmarks to its stated goals, formalise a structured peer-review process, and add a scheduled evaluation clause mirroring Article 40 of the Innovation Act proposal.

Read alongside the European Innovation Act, the Proposal is more than a coordination exercise and more than an indirect simplification tool. It is an attempt to reenforce legislative infrastructure for a shared European innovation culture, and national authorities should embrace it as a chance to drive evidence-based regulation and regulatory transformation Europe so badly needs. The success of the new framework – and the broader emergence of regulatory sandboxes as “legislative infrastructure” for closing the innovation gap – will depend on broad participation, sound methodological design, and an effective peer-review process for periodic revision.