The bottom line first: Most UK service businesses — tradespeople, professional service firms, sole traders, local service companies — are not operating high-risk AI systems as defined by the EU AI Act, and do not need to take compliance action for August 2. But if you use AI tools that inform decisions about hiring, credit, or customer scoring, you need to check your vendor's classification before the end of July.

What August 2 actually means

August 2, 2026 is the date originally set for full compliance with the high-risk AI system obligations in the EU AI Act — specifically Articles 9–17 (provider requirements: risk management, data governance, documentation, transparency, accuracy, robustness, cybersecurity) and Article 26 (deployer requirements: monitor, log, report).

The high-risk categories under Annex III of the Act include:

  • AI used in recruitment, CV screening, or employment decisions
  • AI used in credit scoring or access to financial services
  • AI used in biometric identification
  • AI used in education and vocational training assessment
  • AI used in law enforcement, border control, or administration of justice
  • AI used in critical infrastructure management

What the Digital Omnibus extension means

The EU's Digital Omnibus on AI (Council of the EU final endorsement: 29 June 2026, European Parliament formal endorsement: 16 June 2026) delays the most demanding high-risk provider obligations from August 2, 2026 to December 2, 2027. The legislative text is pending publication in the EU Official Journal, after which the delay becomes formally binding.

Important nuance: the delay primarily benefits AI system providers (the companies building and selling AI tools). Deployers — businesses using those tools — have a different compliance path, and some deployer obligations may still apply from August 2 depending on the final published text.

The practical implication for UK businesses: do not assume the extension covers everything. If your AI vendor has told you their tool is high-risk under the Act, ask them for their updated compliance timeline and what obligations remain live from August 2.

What about UK businesses — does the EU Act apply?

The EU AI Act applies extraterritorially. A UK business is covered if:

  • You place a high-risk AI system on the EU market (you sell AI-powered products or services to EU customers)
  • You deploy a high-risk AI system that affects people in the EU (for example, using an AI hiring tool that screens EU-based applicants)
  • You use a general-purpose AI system with systemic risk that generates outputs used in the EU

For the typical UK service business — a chimney sweep, a professional services firm, a trades company — the honest answer is that the EU AI Act's high-risk provisions are unlikely to apply. You are not operating high-risk AI systems; you are using productivity tools, customer communication automation, and scheduling software.

Where it might apply: if you use an AI tool for candidate screening, or if you offer a scored AI product or service to EU customers. Check with your vendors if uncertain.

The two things every UK business should do before August 2

Even if the high-risk provisions do not apply to you, two low-effort actions are worth completing before the end of July:

  • Ask your AI vendors whether their tools are classified as high-risk. Most major providers (Anthropic, Google, OpenAI, Microsoft) have published EU AI Act classification statements. If your vendor has not, that is a risk signal.
  • Build your AI register. List every AI tool you use, what it does, and who in your business is accountable for reviewing its outputs. This satisfies the spirit of UK and EU AI governance requirements simultaneously. See today's fifth story for the two-hour template.

Quick self-assessment

Do you use AI for hiring or CV screening? If yes: check your vendor's EU AI Act classification now.
Do you sell AI-powered services to EU customers? If yes: take legal advice on whether you are a deployer or provider under the Act.
Do you use AI for general productivity (writing, scheduling, customer communication)? Then the high-risk provisions almost certainly do not apply — but an AI register is still worth building for UK regulatory purposes.
Penalties (for completeness): Up to €35 million or 7% of global annual revenue for the most serious violations. Not applicable to businesses that are not operating high-risk AI systems.