EU BLINKS ON AI ACT — PUSHES ENFORCEMENT DEADLINE TWO YEARS AND QUIETLY REWRITES THE RULES

The European Union reached a political agreement on May 7, 2026 to significantly revise its landmark AI Act, pushing the compliance deadline for high-risk AI systems from August 2026 to December 2027. For AI systems embedded in regulated products under Annex I, the deadline gets pushed even further, to August 2028. The revision covers the most consequential category of AI deployments: systems used in hiring, credit scoring, healthcare, education, law enforcement, and critical infrastructure.

This is a significant retreat from what was supposed to be the world’s most comprehensive AI regulation. Brussels spent three years writing and negotiating the AI Act. It was supposed to set the global standard. Now, facing a flood of industry objections and the reality that enforcement infrastructure was nowhere near ready, EU lawmakers gave the industry a two-year reprieve. The August 2026 deadline is not gone entirely. Chatbot transparency rules and watermarking requirements for AI-generated content still take effect this year. But the teeth of the Act, the obligations that would have forced companies to document, test, and register high-risk AI systems, are pushed into a future that looks increasingly uncertain as the AI development pace outstrips the regulatory calendar. The law that was meant to tame AI launched in a different AI era. The question now is whether any revision can catch up.

Keywords: EU AI Act enforcement deadline 2026, EU AI Act compliance extension, AI regulation Europe, high-risk AI systems rules