On 19 November 2025, the European Commission proposed targeted simplification measures to amend the EU AI Act (Regulation 2024/1689), strengthening AI regulatory sandboxes and expanding real-world testing of high-risk AI systems. The initiative addresses providers and prospective providers of high-risk AI systems under Annex III and Annex I, small and medium-sized enterprises (SMEs) and small mid-caps (SMCs), as well as Member State authorities that supervise AI-enabled products and infrastructures. The Regulation would allow the AI Office to set up an EU-level sandbox for AI systems falling under Article 75(1), with priority access for SMEs and start-ups. Sandbox plans would have to be integrated into real-world testing plans, and national sandboxes would need to be coordinated more closely on the basis of common implementing rules. Real-world testing outside sandboxes would be extended to high-risk AI systems covered by Union harmonisation legislation in Annex I Section A. For AI-enabled products under Annex I Section B, a voluntary regime would be created based on written real-world testing agreements between interested Member States and the Commission that set out the conditions, access to public infrastructure and the detailed testing plan. The Proposal would also alter the implementation timeline of general testing requirements for high-risk AI systems, calculating implementation periods by reference to the Commission adopting a decision confirming that adequate measures in support of compliance were available. High-risk AI systems classified according to Annex III would be required to comply 6 months after the Commission decision, while high-risk AI systems classified according Article 6(1) would be required to comply 12 months after the Commission decisions. In any case, the provisions would apply at the latest from 2 December 2027 for Annex III high-risk systems, and from 2 August 2028 for Article 6(1) high-risk systems.
Original source