EU Agrees to Relax AI Rules And Delay Key Obligations

EU countries and the European Parliament have reached a provisional agreement to ease rules on artificial intelligence, including a delay to some key obligations. The deal came after nine hours of negotiations and still needs formal approval from both member states and the European Parliament.

The changes mainly concern rules for high-risk artificial intelligence systems, including those used in biometrics, critical infrastructure and law enforcement. Their application has been postponed from 2 August 2026 to 2 December 2027. The agreement also exempts some machinery from the Artificial Intelligence Act (AI Act), as it is already covered by sector-specific regulation.

The adjustments are part of the European Commission’s effort to simplify digital rules after businesses complained that excessive red tape was weakening European firms’ ability to compete with rivals in the US and Asia. Critics of the deal, however, say the EU has bowed to pressure from Big Tech companies.

Lawmakers also approved a ban on artificial intelligence systems that create unauthorized sexually explicit images. The measure will take effect on 2 December and also responds to cases of deepfake content generated by Grok, the chatbot on X. Mandatory watermarking of AI-generated content will apply from the same date.

Even after the changes, the EU’s rules on artificial intelligence remain widely regarded as the strictest in the world.

(reuters, max)