Trilateral pilot project for AI regulatory sandbox simulation successfully concluded
Year of issue 2026
Date of issue 2026.03.17
The Bundesnetzagentur, the Hessian Ministry for Digitalisation and Innovation and the Federal Commissioner for Data Protection and Freedom of Information have concluded their joint pilot project for the simulation of an AI regulatory sandbox. The aim of the project, which commenced in May 2025, was to take two start-up cases as a basis and to simulate the key requirements, processes and challenges of AI regulatory sandboxes for the purposes of the European AI Act.
One result of the pilot project is the creation of a detailed roadmap. It shows how the requirements of the AI Act and the Medical Devices Regulation work together when developing a medical device with an integrated high-risk AI system.
"Regulatory sandboxes are an important tool for promoting innovation that is compatible with fundamental rights. AI offers great opportunities while also harbouring significant data protection risks. As the data protection authority, we want to keep contributing to a competitive AI landscape that is also trustworthy and focused on fundamental rights. That is why I look forward to continuing our collaborative and trustful cooperation with the Bundesnetzagentur’s AI regulatory sandbox and with other authorities,"
explained Professor Dr Louisa Specht-Riemenschneider, Federal Commissioner for Data Protection and Freedom of Information.
Hessian Digitalisation Minister, Professor Dr Kristina Sinemus, said, "We have done real pioneering work by simulating an AI regulatory sandbox, because together we want to speed up innovative processes, create legal certainty and at the same time enable responsible AI applications for businesses. We now need clarity as to the structure of the AI Act as soon as possible so that we can know the regulatory framework."
"We are taking responsibility for an AI environment that is conducive to innovation yet at the same time reliable,"
said Klaus Müller, President of the Bundesnetzagentur. "In doing so, we are focusing on close cooperation with other authorities and the data protection authority in order to interlink regulation and innovation constructively. We will create a secure environment with our AI regulatory sandbox, in which companies can develop and test new AI systems. In this way we are strengthening Germany’s leadership role in the responsible use of AI."
Important findings
An AI regulatory sandbox should promote innovation and at the same time provide insights for regulatory practice. Experience obtained from the pilot project shows that those projects that are particularly suitable have a high potential for innovation and a greater perspective for broad application. Concurrently, they should address outstanding regulatory issues relating to the AI Act, often in conjunction with other areas of law.
Successful cooperation in the AI regulatory sandbox depends on trust and professional communication. A structured maturity model that takes the development status of AI systems into account increases the efficiency of regulatory oversight. It also allows for the targeted bundling of legal, expert and technical AI competence within the authorities.
Equally important is close cooperation between the participating authorities where the AI Act and other sector-specific regulations intersect. Only in this way can practical solutions be developed. The participating providers or prospective providers of AI systems will benefit in particular from a comprehensive exit report and written proof of the activities carried out in the regulatory sandbox. These will be of assistance to them in drawing up their technical documentation and may significantly speed up the conformity assessment process.
The results of the pilot project provide an important basis for future arrangements and the operation of the Bundesnetzagentur’s AI regulatory sandbox. The findings will also be made available to the federal state authorities and public institutions, with the intention of supporting the establishment of more AI regulatory sandboxes and of fostering collaboration.
AI regulatory sandbox compliant with the AI Act
The European Union requires all Member States to set up at least one AI regulatory sandbox within the meaning of the AI Act no later than 2 August 2026. Through the federal government’s draft act on AI market surveillance and promotion of innovation, the Bundesnetzagentur is taking on a key role in the national AI ecosystem and setting up an AI regulatory sandbox. In this protected test environment, undertakings can develop, test and validate innovative AI systems under regulatory supervision before they are made available on the market. In this way, companies - especially SMEs and start-ups - can speed up the process of innovation while reliably observing the requirements of the AI Act.
The joint final report on the pilot project and additional information on the AI regulatory sandbox under the AI Act can be found at www.bundesnetzagentur.de/ki-reallabor.