EU Reaches Provisional AI Act Amendments Deal

European Union negotiators from member states and the European Parliament reached a provisional agreement Thursday in Brussels on amendments to the Artificial Intelligence Act, delaying rules for high-risk systems while adding prohibitions on certain harmful applications.

The deal, finalized after nine hours of talks, requires formal approval from EU governments and the Parliament before taking effect. It postpones obligations for high-risk AI systems, including those used in biometric identification, critical infrastructure, and law enforcement, from August 2 this year to December 2, 2027. Negotiators also agreed to exclude machinery from the AI Act's scope, as such equipment falls under existing sector-specific regulations.

These changes respond to concerns raised by businesses about overlapping rules and administrative burdens that could undermine competitiveness against companies in the United States and Asia. The adjustments form part of the European Commission's broader initiative to streamline digital regulations. The AI Act entered into force in August 2024, with provisions phased in over time.

In parallel, the agreement strengthens protections against misuse of AI. It...