EU Softens AI Act: Extended Deadlines and Refined Rules for High-Risk Systems

By

A New Timeline for High-Risk AI Compliance

In a significant move to ease the transition for businesses, European Union lawmakers have agreed to push back the most stringent deadlines under the AI Act. The provisional deal, struck between representatives of the European Parliament and the European Council, grants enterprises additional time to prepare for compliance with high-risk AI regulations. The new deadlines, which still require formal adoption, are set for December 2, 2027 for stand-alone high-risk AI systems and August 2, 2028 for AI used in products covered by EU sectoral safety rules—replacing the original deadline of August 2, 2026.

EU Softens AI Act: Extended Deadlines and Refined Rules for High-Risk Systems
Source: www.computerworld.com

“Today’s agreement on the AI Act significantly supports our companies by reducing recurring administrative costs,” said Marilena Raouna, Cyprus’s deputy minister for European affairs, speaking on behalf of the Council presidency. The breakthrough comes just nine days after previous negotiations collapsed, underscoring the urgency to balance innovation with regulatory oversight.

Refining the Scope: Fewer Overlapping Rules

The provisional agreement introduces several clarifications to reduce duplication and streamline compliance. Notably, it removes overlapping requirements for AI systems integrated into machinery products. Instead, these systems will now follow only sectoral safety rules, with additional safeguards to ensure equivalent health and safety protection.

Redefining “Safety Components”

The deal narrows what qualifies as a “safety component” under the AI Act. AI features that merely assist users or improve performance will not automatically be treated as high-risk, provided their failure does not create health or safety risks. This change is expected to provide relief for many manufacturers and developers.

Mechanism for Sectoral Overlaps

For wider sectors such as medical devices, toys, lifts, machinery, and watercraft, the co-legislators agreed on a mechanism to resolve conflicts between the AI Act and existing sectoral laws. This ensures that companies do not face contradictory obligations while maintaining high safety standards.

Revised Key Dates: Sandboxes and Watermarking

The agreement also adjusts timelines for supporting frameworks and transparency obligations:

  • Regulatory sandboxes: Member states now have until August 2, 2027 to establish AI regulatory sandboxes—a one-year extension from the original deadline. Sandboxes allow companies to test innovations under regulatory supervision.
  • Watermarking obligations: In a move to enhance transparency sooner, requirements to watermark AI-generated content will apply from December 2, 2026 instead of the previously proposed February 2, 2027. This is intended to help users identify synthetic content earlier.

Supporting Small and Mid-Size Companies

Mid-size firms gain additional breathing room. Exemptions that were previously available only to small and medium-sized enterprises (SMEs) now extend to small mid-cap companies, defined as those with up to 499 employees. This expansion is designed to reduce the compliance burden on a broader set of businesses.

EU Softens AI Act: Extended Deadlines and Refined Rules for High-Risk Systems
Source: www.computerworld.com

Central and National Oversight

The deal clarifies the division of enforcement responsibilities. The EU AI Office will supervise general-purpose AI systems centrally, ensuring consistent application across the bloc. Meanwhile, national authorities retain responsibility in areas including law enforcement, border management, judicial authorities, and financial institutions. This dual approach aims to combine efficiency with local expertise.

“With this agreement, we show that politics can move just as quickly as technology,” said Arba Kokalari, the Parliament’s co-rapporteur for the Internal Market and Consumer Protection committee. The formal adoption process is expected to be completed before the original August 2 deadline, after which the new timelines will take effect.

For more details on the AI Act’s provisions and compliance guidance, see our sections on scope changes and revised key dates.

Tags:

Related Articles

Recommended

Discover More

Flutter Embraces Swift Package Manager: CocoaPods Era Ends with v3.44JackRabbit MG Cargo: The Featherweight Hauler That Defies ExpectationsV8 Engine Achieves 2.5x JavaScript Boost by Overhauling Math.random Number StorageThe Case for Detailed Climate Data in Corporate Resilience PlanningThe Hidden Costs of Running AI in the Public Cloud