On 13 March 2026, the Council of the EU agreed its negotiating position on a proposal to streamline certain rules on artificial intelligence, often described as the "AI Omnibus". The proposal is intended to adjust parts of the EU AI Act's implementation approach as the market and governance infrastructure develops.
The Council mandate proposes long-stop dates for when high-risk requirements would apply: 2 December 2027 for standalone high-risk AI systems under Annex III, and 2 August 2028 for high-risk AI systems embedded in regulated products under Annex I. This would extend the original schedule by up to 16–24 months, depending on system category.
The Council position also adds a new prohibited AI practice relating to the generation of non-consensual sexual or intimate content, including child sexual abuse material, reflecting political focus on so-called "nudify" use cases.
Two operational changes are particularly relevant for compliance teams. First, the Council reinstates a registration obligation for AI systems in the EU database even where a provider claims their system is exempt from high-risk classification. Second, it reinstates a stricter "strictly necessary" threshold for processing special category data for bias detection and correction.
Next steps: Parliament committees (IMCO and LIBE) are expected to adopt their position in mid-March 2026, after which trilogue negotiations begin.
Acompli perspective: Treat this as a programme recalibration, not a pause. Dates may move, but the evidence expectations for classification, documentation, and transparency controls remain. Organisations should use this window to strengthen their AI governance framework, review impact assessment requirements, and ensure their teams are prepared for regulatory changes.