A practical guide to regulatory sandboxes
In this article:
Within the AI Act, regulatory sandboxes are defined as frameworks established by competent authorities.
These frameworks allow providers or prospective providers to test innovative AI systems in real-world conditions for a limited period.
The operation of these sandboxes is governed by a “sandbox plan,” which details the objectives, conditions, timeframe, methodology, and requirements for the activities conducted within the sandbox.
This plan is a mutually agreed document between the AI provider and the competent authority, ensuring clear guidelines and expectations for the testing phase.
The Act mandates that each Member State must establish at least one AI regulatory sandbox at the national level within 24 months of the regulation’s entry into force.
These sandboxes may also be established jointly with other Member States or by participating in existing sandboxes, provided they offer an equivalent level of national coverage.
Additionally, Member States can establish further sandboxes at regional or local levels or in cooperation with other Member States.
