“Sandbox”: A Virtual Environment for Implementing Responsible Artificial Intelligence

By Avv. Helga Zanotti

The European Commission’s April 2021 proposal for a regulation on artificial intelligence opens the door to monitoring AI through “sandboxes”: controlled virtual environments in which the potential risks arising from its use can be assessed. The main areas considered are health, safety and fundamental rights.

Spain has offered to act as pilot country for the “AI sandbox” project, which aims to pursue the following objectives:

• clarify and give practical substance to the requirements set out in the proposed AI Regulation for AI systems;
• support compliance with technical know-how by applying AI rules to companies that are already implementing AI or planning to do so;
• encourage innovation and promote the development of trustworthy AI systems;
• build expertise and begin a consultation process in Spain for the creation of a national supervisory authority;
• test future obligations and compliance requirements in a controlled environment, including through the development of guidelines and national and European standards.

Cooperation between Spain and the European Commission will be close and ongoing. Other European countries will also be allowed to observe or participate directly, giving the sandbox a pan-European dimension.

The sandbox will be structured around two focus groups:

1. the first will implement the digital environment in practical terms, with a focus on the regulatory framework and on testing different solutions, also in cooperation with the relevant authorities;
2. the second will prepare the overall structure of the sandbox, study its development and share know-how during the implementation of the relevant requirements, including technical requirements.

Spain will act as a scientific hub of excellence for digital innovation, building on the Spanish National AI Strategy. Europe, meanwhile, appears to be taking on the role of connecting innovators and regulators in order to create a controlled digital environment that facilitates development, testing and validation, while ensuring compliance with the requirements identified by the proposed AI Regulation and with any national rules that may be adopted.

For further information, please contact Paula Vega at p.vega@bmvinternational.com.