The regulatory sandbox for AI systems is a controlled framework set up by the competent national authority. It allows providers and deployers of AI systems to develop, learn and test an innovative AI system on the basis of a sandbox plan for a limited period of time and under regulatory supervision. It is a tool to stimulate innovation in the field of AI and to help ensure compliance with the new regulatory framework before entering the market or being used in practice.
Benefits of regulatory sandboxes:
- Developing innovative products that comply with AI legislation fosters collaboration between market players to find the right solutions.
- Product development follows from the outset the rules imposed by the regulator in the development of AI systems.
- The Regulatory Sandbox Report can be used as evidence that a product has been developed according to the principles of ethical and safe use of an AI system.
Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (AI) provides for the establishment of AI regulatory sandboxes to stimulate innovation in the field of artificial intelligence (AI) across the EU.
The Regulation requires EU Member States to establish at least one AI regulatory sandbox at national level by 2 August 2026 at the latest.
The Agency is responsible for setting up the regulatory sandboxes. The Agency will be responsible for the establishment of regulatory sandboxes for AI systems:
- Coordinate and provide support for the involvement of other market surveillance authorities, notification authorities and other inspection bodies with subject matter competence,
- launch a public call for proposals on the content of the AI regulatory sandbox area,
- provide an appropriate platform for the establishment of regulatory sandboxes,
- carry out the tasks related to the annual reports.
The selection process for the Regulatory Sandbox will be carried out in two steps:
- The first step will be a public call for the selection of the content area of the regulatory sandbox, which will be published on the Agency's website.
- The second step will be an open call for candidates to test AI systems in the selected sandbox area, based on one or more selected areas, published on the Agency's website.
Priority in the selection of bidders will be given to start-up, small and medium-sized enterprises(SME classification), which are also entitled to free access to the sandbox under the AI Act. The detailed conditions for the participation of companies in the sandbox will be published in the call for proposals.