EU tightens AI rules
The EU moved to streamline its landmark AI Act while lawmakers pushed bans on non‑consensual deepfakes and took first steps to outlaw AI‑generated child sexual abuse images — a major regulatory tightening announced that also drew coverage on the child‑abuse ban reported. Industry groups are already lobbying for carve‑outs for sectors like manufacturing and healthcare, warning the rules could duplicate existing safety oversight argued.
The Council’s negotiating mandate came as part of the Omnibus VII legislative package [press release] consilium.europa.eu, and it explicitly lets the Commission delay application of high‑risk AI rules by up to 16 months, setting long‑stop dates of 2 December 2027 for stand‑alone high‑risk systems and 2 August 2028 for systems embedded in products [text]. consilium.europa.eu The mandate inserted a new ban on AI practices that generate non‑consensual sexual or intimate content and on AI‑generated child sexual abuse material, a move EU governments described as a concrete legal step following public outcry; Reuters reported the governments’ proposal on 13 March 2026. consilium.europa.eu Industry groups pressed for carve‑outs and lighter obligations, with DIGITALEUROPE warning of high compliance costs for manufacturers and smaller firms and urging changes to protect industrial competitiveness [statement], while a group of ten member states including Germany, France, Italy, Spain and Poland has asked for clearer overlap rules with existing product safety regimes [updates]. digitaleurope.org The Council reinstated provider obligations to register high‑risk systems in the EU database even where providers claim an exemption and kept a strict‑necessity test for processing special category data for bias detection [press release], while legal advisers have flagged that the omnibus’s conditional delays and undefined “adequacy” tests leave compliance timing and obligations uncertain for manufacturers and health‑sector firms [analysis]. consilium.europa.eu
