The regulatory sandbox, or controlled testing space, is a structured framework that allows experimentation with innovative technologies, products, services, or models in real or near-real conditions, for a defined period and scope, under regulatory supervision and with appropriate safeguards.
In recent years, this concept has gained increasing relevance in sectors where the pace of innovation is generating new regulatory challenges, such as artificial intelligence, energy technologies, the circular economy, autonomous mobility, or new industrial models.
What does it offer compared to existing instruments?
The main value is that it helps reduce the risks associated with the innovation process, facilitating that companies, administrations, and regulatory bodies generate knowledge jointly before a solution reaches the market. But what does it really offer compared to the experimentation mechanisms that already existed? A sandbox does not have as its primary objective to simply determine whether a technology works. For that, there are tools such as laboratories, which allow validating technologies under highly controlled conditions; pilot plants, which evaluate performance at a larger scale; testbeds, aimed at analyzing performance, interoperability, or scalability; or living labs, where users and other actors participate in experimentation processes.
All of these instruments primarily respond to one question: does the technology work? The sandbox introduces another question: can this technology operate within the existing administrative, operational, and regulatory framework? Are there regulatory barriers? Is it necessary to interpret certain requirements? Should any procedure be adapted to enable the safe application of a new technology? This is the main difference between a regulatory sandbox and other experimentation spaces: the existence of a specific regulatory issue and the involvement of a competent authority to supervise the process.
Therefore, a sandbox is not simply a technological pilot in a real environment. For regulatory experimentation to occur, an uncertainty or regulatory barrier must be identified, what is to be analyzed from a regulatory standpoint must be defined, and a collaboration framework between innovators and regulators must be established.
A Controlled Environment: Rules, Boundaries, and Safeguards
In practice, specific requirements are set to ensure that experimentation stays controlled: the solution must not be in a commercial phase, the action must be reversible, and if it requires installing devices or equipment, these must be removable at the end of the test, leaving the environment in its initial state. Moreover, participating in a sandbox does not imply an obligation for the Administration to purchase, and the costs associated are typically borne by the promoting company.
These conditions are fundamental to separate experimentation from public procurement, since participating in a sandbox does not mean selling a solution to a government nor guarantees a future purchase. The main objective is to generate knowledge and evidence that allow determining whether a solution can be deployed later and under what conditions. To that end, the process must include a test plan (sandbox plan) that defines objectives, methodology, indicators, responsibilities, monitoring mechanisms, and exit criteria, always maintaining the necessary safeguards in areas such as safety, data protection, the environment, consumers, or fundamental rights.
Even when a test does not achieve the expected result, the learning generated is valuable. Identifying technical limitations, operational difficulties, or regulatory barriers allows improving the solution and anticipating problems before large-scale implementation. This shared learning is precisely one of the distinctive elements of the sandbox: the so-called regulatory learning, which enables both the innovator and the regulator to understand how a technology operates in real-world conditions and how the regulatory framework should evolve.
Europe Drives Sandboxes as a Tool for Regulatory Innovation
This approach is gaining growing importance at the European level. The Council Recommendation proposal submitted by the European Commission on September 9, 2026 (COM(2026) 568 final) sets out common principles for the development of regulatory sandboxes in the European Union, with the aim of reducing fragmentation among member states and facilitating a more coherent framework for this type of experimentation. The proposal emphasizes that sandboxes are not about deregulating, but about creating safe spaces where innovation and regulation can evolve together.
An illustrative example is the European AI Act, which incorporates the concept of an AI regulatory sandbox, where providers of AI systems can develop, train, validate, and test innovative solutions under the supervision of competent authorities for a limited period. In this case, the objective is not solely to verify that the system works technically, but to analyze key regulatory aspects such as data governance, technical documentation, transparency, human oversight, risk management, or future conformity assessment.
Thus, a sandbox does not replace pilots, living labs, or testbeds. They complement them by incorporating a dimension that traditionally emerged later in the innovation process: the interaction between the technology and the regulatory and administrative environment in which it will have to operate. In a European context that seeks to accelerate the market entry of new technologies without reducing safety, environmental protection, or citizens’ rights, sandboxes can become a key tool to close the gap between innovation, demonstration, and deployment.
Olga Rivera, Project Manager, European Funds.
