European Commission Implementing Regulation on the Operation of AI Regulatory Sandboxes Published

mohamed-nohassi-5K_ijgjiDko-unsplash

The European Commission Implementing Regulation (EU) 2026/2487 has been published in the Official Journal of the European Union, establishing common rules and laying down detailed arrangements for the establishment, development, implementation, operation and supervision of AI Regulatory Sandboxes, as provided for under Regulation (EU) 2024/1689 (the Artificial Intelligence Act – AI Act).

The adoption of the Implementing Regulation marks an important step towards the implementation of the AI Act, as it establishes a common European framework for the operation of AI Regulatory Sandboxes, with the aim of preventing fragmentation, ensuring the consistent application of the AI Act across the European Union and facilitating the mutual recognition of the results achieved through
participation in these sandboxes.

AI Regulatory Sandboxes are controlled environments established by competent authorities, which enable providers and prospective providers of AI systems to develop, train, validate and test innovative AI systems, including, where appropriate, under real-world conditions, on the basis of a specific plan and under regulatory supervision, for a limited period before the systems are placed on the market.

Under the AI Act, each Member State must ensure that at least one AI Regulatory Sandbox is operational at national level by 2 August 2027.

In Greece, Law 5321/2026 provides for the establishment of the AI Regulatory Sandbox within the Hellenic Telecommunications and Post Commission (EETT). The EETT and the Hellenic Data Protection Authority (HDPA) are designated as the competent authorities responsible for its operation. The law also provides for the possibility of cooperation with the Greek Artificial Intelligence Factory, “PHAROS AI FACTORY”, the Hellenic Centre for Defence Innovation (ELKAK S.A.) and the European Digital Innovation Hubs (EDIHs).

Regulatory sandboxes can serve as an important instrument for promoting responsible innovation, by providing an organised framework for cooperation between innovation stakeholders and competent authorities. Through this cooperation, participants can gain greater legal certainty and a better understanding of regulatory requirements, while competent authorities can acquire valuable experience and insights into the capabilities, emerging risks and potential impacts of AI deployment.

Furthermore, the Implementing Regulation encourages the use of the wider European and national AI ecosystem through cooperation with testing and experimentation facilities, European Digital Innovation Hubs, centres of excellence, AI Factories and other AI innovation initiatives. In this way, Regulatory Sandboxes can serve not only as a mechanism for supporting compliance with the AI Act, but also as an instrument for regulatory learning, responsible innovation and the wider uptake of AI
across the European Union.

Particular emphasis is placed on small and medium-sized enterprises (SMEs), including start-ups, which are given priority access to Regulatory Sandboxes. The regulatory guidance, advice and supervision services provided under the Regulation are to be offered free of charge, in accordance with its terms and conditions.

It should be noted that participation in a Regulatory Sandbox and successful completion of the sandbox do not constitute an authorisation to place an AI system on the market or put it into service, nor do they create a presumption of conformity. Providers remain fully responsible for ensuring compliance with all applicable legislation.

What the New Implementing Regulation Provides

Implementing Regulation (EU) 2026/2487 further specifies the framework governing the operation of AI Regulatory Sandboxes and provides, inter alia, for:

  • Common conditions for access and participation, ensuring transparent and non-discriminatory procedures.
  • Specific eligibility and selection criteria, taking into account, among other factors, the regulatory and practical challenges associated with the project, its potential to generate regulatory learning, its public interest or societal benefit, the maturity of the proposal and the innovative nature of the AI system.
  • The preparation of a sandbox plan, specifying the objectives and scope of the activities, the roles and responsibilities of the parties involved, the relevant regulatory requirements, the risk management safeguards and, where applicable, the plan for real-world testing.
  • The possibility of conducting real-world testing, subject to the conditions laid down in the applicable European and national regulatory framework.
  • The issuance of written evidence and an exit report upon completion of participation, documenting the activities carried out and contributing to regulatory learning.
  • Mechanisms for the monitoring, evaluation and dissemination of lessons learned from the operation of the sandboxes.

Full text of Commission Implementing Regulation (EU) 2026/2487

Share on: