All insights

Weekly7 August 20261 min read

This week in AI: the EU AI Act deadline that moved

The high-risk rules in the EU AI Act were due to apply on 2 August. A late change pushed most of them to December 2027. Here is what changed and why we are carrying on regardless.

What happened

Until a few weeks ago, 2 August 2026 was the date when the EU AI Act's rules for high-risk AI systems were due to apply. These cover uses such as recruitment, credit scoring, education and access to public services.

That date has moved. The Digital Omnibus on AI entered into force on 27 July. It moves the obligations for stand-alone high-risk systems to 2 December 2027, and those for AI built into regulated products, such as medical devices, to August 2028. The rules for general-purpose model providers, which began in August 2025, are unchanged.

Our take

For UK organisations that only operate in the UK, the AI Act does not apply directly. But many UK suppliers sell into the EU, and UK buyers increasingly ask the same questions the Act asks. How was the system tested? What data does it use? Who oversees it? How do users know they are dealing with AI?

So the delay changes timetables more than it changes good practice. Our advice to the organisations we work with is to keep a simple record for every AI tool: what it does, the risks you considered, how a person stays in control and how you tell users. In the UK, the Algorithmic Transparency Recording Standard already provides a template for most of this.

If you have paused that work because the deadline moved, this is a good moment to restart it, without the pressure of a date.

Keep reading

Related insights