A regulatory sandbox is a controlled environment in which a company can test an innovative solution under the supervision of the relevant authority, with agreed conditions and a limited scope. Its purpose is to enable development where strict application of the rules would make testing impossible, while also giving the company legal certainty that it is proceeding correctly. European artificial intelligence regulation provides for such environments, and member states are to establish them, with particular attention given to smaller undertakings and startups. For a company, the main benefit is access to the regulator's interpretation before deployment, which in unclear situations is more valuable than any amount of advice. Availability, entry conditions and scope are still evolving, so check the current state with the relevant authority.
See also: AI risk classification, High-risk AI system, Oversight and Sanctions for AI.
