Adopted on the basis of the EU AI Act, the Commission has established the rules for the development and operation of AI regulatory sandboxes.
AI regulatory sandboxes provide a controlled, time-limited environment established by competent authorities to help providers develop, train, validate, and test innovative AI systems before placing them on the market.
AI regulatory sandboxes provide a controlled environment established by competent authorities to help providers develop, train, validate, and test innovative AI systems before placing them on the market. This regulation will set unform rules across all EU member states to overcome fragmentation and maximise the legal certainty of sandboxes for all.
SMEs are offered guidance about participation and in regards to the regulation and sandbox services, which should be individually judged on the need for such a sandbox, its regulatory learnings, and the innovation quality of the system. Competent data protection authorities should be on hand when personal data is used to train AI systems.
The eligible systems must not be already on the market, be substantially modified or deemed “high-risk”. Once the sandbox, the participants and authorities decide on the timelines, scope of activities, objectives, and real testing plans.
