Article 57 requires each Member State to establish at least one national AI regulatory sandbox by 2 August 2027 (postponed from 2026 by the Digital Omnibus, Regulation (EU) 2026/1744). Sandboxes allow providers to develop, train, test, and validate innovative AI systems under the supervision of national competent authorities in a controlled environment before market entry.
Regulation (EU) 2024/1689 — Article 57(1)
“Member States shall ensure that their competent authorities establish at least one AI regulatory sandbox at the national level, which shall be operational by 2 August 2027 (as amended by Regulation (EU) 2026/1744).”
- Sandboxes provide a controlled environment to test innovative AI without triggering full compliance obligations
- Member States may establish joint sandboxes with other Member States
- Participation does not exempt providers from liability for harms caused to third parties
- Sandbox findings must be documented and may be used to inform compliance approaches after exit