EU AI Act: 'High-Risk' Sector Implementation Postponed to December 2027

The EU has agreed to postpone the implementation of regulations for high-risk AI sectors under its pioneering AI Act until December 2027.

Thu May 07 2026

EU AI Act: High-Risk Sector Implementation Delayed

The European Union (EU) provisionally agreed on May 7 to revise its comprehensive AI (Artificial Intelligence) Act, the world's first such legislation. The agreement postpones the start of regulations for high-risk AI applications from this August to December 2027. This move is seen as an effort to increase regulatory flexibility, particularly with the ongoing AI development competition against countries like the United States and China.

Classification and Implementation Timeline for 'High-Risk' and 'Safety Component' AI Systems

The AI Act, set to take effect in 2024, categorizes AI usage into four risk levels, with regulations applied incrementally. The postponement, agreed upon by the European Parliament and the EU Council, affects AI systems classified as 'high-risk', including those used for biometric identification, education, employment, and border control. Additionally, AI incorporated as 'safety components' in products such as medical devices, automobiles, aviation, railways, and toys will see its implementation further delayed until August 2028. However, 'safety component' products that do not pose a significant risk to health or safety in case of failure will not be classified as 'high-risk'.

*Source: 朝日新聞 (2026-05-07)*

Tags: #EU #AI Act #AI regulation #European Parliament #High-risk AI