In a significant policy shift, the European Parliament on March 26, 2026, voted overwhelmingly—569 in favor, 45 against, and 23 abstentions—to adopt its negotiating position on the Commission’s Digital Omnibus proposal, which amends the EU Artificial Intelligence Act (AI Act). The amendment delays the enforcement of obligations for high‑risk AI systems, citing the absence of finalized technical standards and insufficient supervisory capacity across member states.(dentons.com)
Under the new timeline, stand‑alone high‑risk AI systems—such as those used in biometrics, education, employment, law enforcement, and critical infrastructure—will now face compliance requirements by December 2, 2027, instead of the original August 2, 2026 deadline. AI systems embedded in regulated products (e.g., medical devices, machinery) will have until August 2, 2028.(zpravy.kurzy.cz)
The delay responds to practical implementation hurdles. The European Commission has yet to publish essential guidance on high‑risk classification, and only eight of the 27 member states have designated national supervisory authorities or contact points—creating a regulatory vacuum if enforcement were to proceed as originally scheduled.(delbion.com)
Despite the postponement, one critical requirement remains unchanged: Article 4’s mandate for AI literacy training. All personnel involved in the development or deployment of AI systems must complete AI literacy training by August 2, 2026. This obligation applies across sectors and company sizes and is not affected by the delay.(delbion.com)
The Parliament’s vote now sets the stage for trilogue negotiations with the Council. If finalized, the amended timeline will provide much‑needed breathing room for businesses and regulators to align on standards and enforcement mechanisms. However, organizations must still prioritize compliance with unchanged provisions, particularly staff training, to avoid penalties and ensure readiness for the eventual rollout of high‑risk AI rules.
This development marks the most consequential AI policy shift in the past 24–48 hours, reflecting the EU’s pragmatic approach to balancing regulatory ambition with operational feasibility.