EU Delays High-Risk AI Rules to 2027: What UK Businesses Need to Know

What happened

The EU Council published formal proposals this week to delay the most demanding parts of the AI Act by up to 16 months. Under the changes, high-risk standalone AI systems would face a new compliance deadline of December 2027, down from the original August 2026 date. AI systems embedded in physical products would have even longer, until August 2028.

The reason for the delay is straightforward: the EU has not yet published the technical standards that businesses need in order to demonstrate compliance. Pushing the deadline while those standards are still being written would have created an impossible situation for many companies.

The proposals also extend regulatory exemptions. Relief currently available to small and medium-sized businesses would be expanded to cover small mid-cap companies, reducing the compliance burden for a wider range of organisations. The package also adds new prohibitions specifically targeting AI-generated non-consensual sexual content and child sexual abuse material, areas the original Act left partially unaddressed.

The European Parliament is expected to finalise its position ahead of a June vote, with the changes targeted for publication in July 2026.

What this means for your business

If your business uses AI in a way that touches the EU, whether you sell into Europe, have EU customers, or deploy AI tools supplied by EU providers, the AI Act still applies to you. The UK is not inside the EU regulatory framework, but the AI Act has extraterritorial reach: if your AI system or its outputs are placed into the EU market, the rules follow.

What this delay means in practice is breathing room. If you were scrambling to meet an August 2026 deadline for a high-risk AI deployment, you now have until late 2027 or 2028. But the direction of travel has not changed. If your business uses AI to screen job applicants, assess credit, or make decisions in health or education, these are likely to be classified as high-risk systems under the Act.

The smart move now is to understand whether your AI use falls into the high-risk category and to begin mapping what compliance will look like, even if the deadline has shifted. Waiting until 2027 to start that work will still leave you short of time.