EU hits snooze on AI Act rules after industry backlash

The European Union has announced a significant delay in the enforcement of key provisions within its landmark Artificial Intelligence Act, a move widely seen as a concession to industry lobbying. Initially slated for August 2, 2026, compliance deadlines for high-risk AI systems in critical sectors like biometrics, employment, and border control have been pushed back to December 2, 2027. This recalibration aims to address concerns from tech firms and industry groups that the original timeline was unworkable, potentially hindering Europe's competitive position in the global AI race. This provisional agreement, dubbed the "Digital Omnibus on AI," streamlines certain aspects of the AI Act, particularly for manufacturers in industrial sectors facing overlapping regulatory requirements. While the delay grants smaller companies more flexibility and seeks to untangle AI rules from existing product safety legislation, it does not represent a wholesale weakening of the act. For instance, the agreement solidifies a ban on AI systems used to generate non-consensual sexual deepfakes and child sexual abuse material, reflecting a continued commitment to curbing harmful AI applications. The decision to delay enforcement highlights a delicate balancing act for the EU, striving to foster innovation while ensuring citizen protection. Officials contend the adjustments are necessary to align regulatory timelines with the development of supporting standards and technical guidance. This revised approach seeks to create a more "innovation-friendly environment," as stated by Commission President Ursula von der Leyen, without compromising the fundamental safety and ethical considerations at the heart of the AI Act.