The European Commission is set to postpone a central element of its Artificial Intelligence Act, following sustained pressure from the US administration, technology companies, and industry lobby groups.
According to two Commission officials, the proposal, expected on 19th November, would push back the application of high-risk AI requirements by at least one year.
The Commission is planning to introduce the delay as an amendment within a wider digital simplification package. Any revision still requires approval from EU countries and the European Parliament before it becomes law. If endorsed, the shift moves implementation of high-risk rules to August 2027, signalling a change in Europe’s regulatory posture.
The EU had previously positioned itself as an early mover in AI oversight. However, concerns that stringent rules might impede competitiveness, particularly against the US and China, are influencing the Commission’s revised timeline. Critics of the existing schedule argue that the technical standards firms need to demonstrate compliance are not ready for a summer deadline, making a delay unavoidable.
The initial provisions of the EU AI Act came into effect on 2nd February 2025. From that date, systems classified as unacceptable are banned. These include:
- Social rating tools used by public or private bodies
- Individual predictive policing models
- Emotion recognition systems in workplaces or educational settings
- AI exploiting user vulnerabilities or applying manipulative or subliminal methods
- Real-time facial recognition for identification in public spaces
- Biometric categorisation used to infer characteristics such as race, political beliefs, religious beliefs, sexual orientation, or trade union membership
The implementation of the legislation, considered the most ambitious AI regulation in the world to date, was set to be gradual.
Jane Smith, Chief Data & AI Officer at ThoughtSpot, comments on the implications of the delay: “In delaying the AI Act, the EU is abandoning the one area where it could have led globally: rigour. Speed was never going to be Europe’s advantage in AI; standards were. While China competes on scale, and the US on capital and innovation, Europe is unlikely to win in either of those areas.
“By delaying key provisions of the AI Act, the EU is losing its only competitive advantage and creating unnecessary uncertainty. Rather than simplifying the regulatory landscape as claimed, this move leaves everyone unclear about what to expect. Will these rules return unchanged in a year, in a different guise, or is this a complete change of approach?
“It feels like Europe has capitulated to big tech pressure, which is a shame. The EU was one of the few major bodies taking a stand for people over corporate interests, and this represents a significant step back from positioning itself as the global leader in safe and responsible AI.”
