In a major policy move, the European Parliament and Council have reached a provisional agreement to streamline the EU’s Artificial Intelligence Act as part of the broader “Omnibus VII” legislative package. The deal introduces pragmatic amendments aimed at reducing administrative burdens while reinforcing protections against harmful AI practices.
Key changes include:
- Extension of compliance deadlines for high‑risk AI systems: standalone systems now face a December 2, 2027 deadline, while AI embedded in products must comply by August 2, 2028 (consilium.europa.eu).
- A new prohibition on AI systems that generate non‑consensual sexual or intimate content, including child sexual abuse material (CSAM) (consilium.europa.eu).
- Reinstatement of obligations for providers to register AI systems in the EU’s high‑risk database, even when claiming exemption, and reinforcement of the “strict necessity” standard for processing sensitive personal data for bias detection and mitigation (consilium.europa.eu).
These amendments are designed to support EU businesses by reducing recurring administrative costs and ensuring legal certainty, while maintaining robust safeguards for vulnerable populations and sensitive data contexts (consilium.europa.eu).
The provisional agreement marks the first deliverable under the “One Europe, One Market” roadmap and reflects the EU’s commitment to balancing innovation with safety and rights protection (consilium.europa.eu).
What to watch next: The formal vote by both the European Parliament and Council is expected in the coming weeks, after which the updated timelines and content restrictions will become binding. Organizations operating in the EU should begin adjusting compliance strategies to align with the new deadlines and content rules.