- within Technology topic(s)
- in United States
- with readers working within the Business & Consumer Services, Insurance and Technology industries
- within Real Estate and Construction topic(s)
The Council of the European Union has now given its final green light to a targeted package of amendments to the EU AI Act, marking a decisive shift from framework-setting to implementation. The changes—part of the “Omnibus VII” simplification agenda—aim to reduce regulatory friction while preserving the Act’s risk-based architecture.
For lawyers and compliance teams, the headline development is timing. Key obligations for high‑risk AI systems are deferred, with new application dates set for December 2027 (stand‑alone systems) and August 2028 (product-integrated systems). This provides welcome breathing room, but not a pause. Organizations should be using this time to build governance frameworks and technical controls.
The amendments also sharpen substantive guardrails. New prohibitions target AI systems generating non‑consensual intimate content and child sexual abuse material, reflecting a clear policy priority around fundamental rights protection.
At the same time, the EU is tightening certain compliance expectations. Transparency deadlines for AI‑generated content are accelerated, and regulatory sandbox timelines are adjusted to better align with implementation realities.
Overall, this is a recalibration, not a rollback. The EU remains firmly committed to global AI regulation, but with a more pragmatic timeline and clearer enforcement pathway. For businesses operating in or into the EU, the message is clear: prepare now, because enforcement is coming—just on a more coordinated schedule.
Given that provisions on high-risk AI systems were due to enter into force on 2 August 2026, the co-legislators treated this part of the package with utmost priority and agreed on a fixed timeline for the delayed application of high-risk rules: the new application dates would be 2 December 2027 for stand-alone high-risk AI systems and 2 August 2028 for high-risk AI systems embedded in products.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
[View Source]