A controlled arrangement in which a regulator lets businesses test new products with real customers under close supervision, with some flexibility in how the rules apply, so both sides can learn before anything is rolled out widely. The idea was popularised by the United Kingdom’s financial regulator in 2016 and has since been adopted for AI, including in European Union law.
Technical view
The Financial Conduct Authority opened its sandbox to applications in May 2016, offering individual guidance, restricted authorisation, waivers and consumer safeguards during live testing; data protection and other regulators later ran their own. Article 57 of the European Union’s AI Act requires each member state to set up at least one AI regulatory sandbox, a deadline moved from 2 August 2026 to 2 August 2027 by the Digital Omnibus on AI. Participants receive guidance and exit documentation that can support later conformity assessment, but remain liable for harm caused during testing. Critics raise selection bias towards well-resourced firms and regulatory capture; supporters point to faster learning on both sides.
Related
Part of AI Terminology Index, section: Governance, Law and Regulation.