Direct answer
What does Article 57 of the AI Act say about AI regulatory sandboxes?
Your question is about Article 57: AI regulatory sandboxes. That obligation applies today. Whether your system actually falls under it depends on conditions you assess yourself.
You determine this yourself
- Relevant where you are a provider or prospective provider of an innovative AI system that you want to develop, train, test or validate before placing it on the market or putting it into service, and you want up-front certainty about classification or about how you meet the requirements of this Regulation.
- Participation is voluntary. Article 57 places the duty on the Member State to provide a sandbox, not on you to join one. A Member State may also fulfil that duty by participating in an existing sandbox with equivalent national coverage. Article 57(4) leaves other regulatory sandboxes established under Union or national law unaffected.
First step: Apply to a sandbox and agree the sandbox plan.
Article 57 of the AI Act covers AI regulatory sandboxes. Member States must provide at least one national AI regulatory sandbox. For you this is a voluntary route: you develop, train, test and validate an innovative AI system in a controlled, supervised environment under a plan agreed with the competent authority, before placing it on the market or putting it into service. This provision concerns the deployer and the provider of an AI system. This provision has applied since 2 August 2026.
This applies now
- Article 57: AI regulatory sandboxesApplicable
Your first actions
- Apply to a sandbox and agree the sandbox plan. Apply to the competent authority, agree a specific sandbox plan, and record which uncertainty about the Regulation you want resolved inside the sandbox.
Record this
- Written proof of participation and the exit report
General interpretation, not legal advice. Checked against Regulation (EU) 2024/1689 and the Digital Omnibus (EU) 2026/1744; the official source remains authoritative.
Full map for your situation