In a significant regulatory development, the European Parliament and Council have reached a provisional agreement aimed at simplifying the implementation of the EU’s landmark AI Act. The deal introduces several key adjustments:
• The deadline for establishing national AI regulatory sandboxes has been postponed to August 2, 2027, giving member states more time to prepare.
• The grace period for providers to implement transparency measures for AI-generated content has been shortened from six months to three months, with a new deadline of December 2, 2026.
• The agreement clarifies the supervisory role of the EU’s AI Office over general-purpose AI systems developed and deployed by the same provider, while preserving national authority in areas such as law enforcement, border control, judiciary, and financial services.
• To avoid regulatory overlap, the deal allows sectoral laws—such as those governing medical devices, machinery, toys, lifts, and watercraft—to take precedence over the AI Act where similar AI-specific requirements already exist. The Commission may issue delegated acts under sectoral legislation to add health and safety requirements for high-risk AI systems.
• The Commission is also tasked with issuing guidance to help economic operators of high-risk AI systems comply with the AI Act in a way that minimizes administrative burden.
This provisional agreement now awaits formal endorsement by both the Council and the Parliament, followed by legal-linguistic revision and final adoption in the coming weeks. If ratified, these changes will ease compliance for businesses—especially SMEs—while preserving the AI Act’s core protections and risk-based approach.
This development marks a pragmatic shift in EU AI policy, balancing regulatory rigor with flexibility and clarity for implementers.