EU delays parts of AI Act
The European Parliament voted to delay implementation of key parts of the EU AI Act, including deadlines for certain high‑risk uses like biometrics and some content‑labeling rules—creating compliance uncertainty for CIOs and vendors. The pause forces companies to choose between early investment in draft controls or waiting for the final rules to settle. (computerworld.com) (cio.com)
IMCO and LIBE committees adopted a joint report (A10‑0073/2026) on March 18, 2026, by 101 votes to 9 (8 abstentions) that fixes application dates for postponed obligations: 2 December 2027 for high‑risk systems in Annex III and 2 August 2028 for Annex I products. (ppc.land) The European Parliament’s plenary carried measures reflecting the committee position in a vote on March 26, 2026, but the package still requires formal approval from the Council of the European Union before it becomes law. (cio.com) The delay traces back to a European Commission proposal published as part of the Digital Omnibus (COM(2025)0836) on 19 November 2025 that sought to push back high‑risk obligations until supporting standards and guidance were ready. (ppc.land) European standardisation bodies CEN‑CENELEC confirmed the technical standards meant to operationalise the AI Act would not be ready by the originally targeted 2025 timetable and are expected to extend into 2026, a key justification cited for the postponement. (allaboutai.com) Rapporteurs Arba Kokalari and Michael McNamara led the committee amendments that replaced the Commission’s proposed “trigger plus six‑month” mechanism with the fixed calendar deadlines now proposed in the report. (ppc.land) Alongside the timetable changes, MEPs backed new prohibitions including bans on so‑called “nudifier” apps and non‑consensual sexually explicit deepfakes as part of the same amendment package. (theverge.com)
