Decode & Grow

Does the EU AI Act Apply to UK Businesses? A Straight Answer

Short answer: Yes, in defined circumstances. The EU AI Act applies to businesses outside the EU where they place an AI system on the EU market, or where the output produced by their AI system is used within the EU. A UK consultancy using AI to produce work for EU clients is likely in scope. A UK business serving only UK customers with internal AI tools generally is not.

General information, not legal advice. Take advice on your specific position.

The two triggers

Trigger one: placing on the EU market. If you make an AI system or a product containing one available in the EU — selling software with AI features to EU customers, offering an AI-powered service into the EU — you're in scope as a provider, with the obligations that attach to whatever risk category it falls into.

Trigger two: output used in the EU. The broader and more commonly overlooked one. If you're a deployer of an AI system and the output it produces is used within the EU, the Act reaches you regardless of where you sit.

The practical implication: a Cardiff consultancy using AI to help produce analysis delivered to a client in Dublin is producing output used in the EU. That's the trigger.

Who is clearly out of scope

  • UK businesses with only UK customers, using AI internally, with no output crossing into the EU.
  • Purely personal, non-professional use.
  • Use of AI in a way that produces no output used in the EU — which is most local service businesses.

If that's you, the EU AI Act isn't your concern. UK data protection law still is, wherever AI touches personal data.

What UK businesses in scope actually have to do

For the great majority — deployers using off-the-shelf AI tools — the obligations are modest:

  • AI literacy. Ensure staff dealing with AI systems have appropriate training. Document it.
  • Avoid prohibited practices. A short compliance check.
  • Transparency where Article 50 applies. Disclose AI interaction, label AI-generated content where required. Enforceable since 2 August 2026.

The high-risk regime — conformity assessments, technical files, notified bodies — only engages if you deploy AI in a listed high-risk domain such as recruitment decisions, credit scoring or education access. Its deadlines have also been pushed back into 2027 and 2028.

What applies in the UK regardless

The UK has taken a sector-led approach rather than passing an equivalent horizontal AI statute. What binds UK businesses today:

  • UK GDPR and the Data Protection Act. Anything involving personal data — lawful basis, transparency, DPIAs for high-risk processing, and rights around automated decision-making.
  • Existing sector regulation. The FCA, MHRA, Ofcom and others apply their frameworks to AI within their remits.
  • Equality law. An AI tool producing discriminatory outcomes creates liability under the Equality Act regardless of AI-specific rules.
  • Consumer protection and contract law. Unchanged by the technology used to produce the output.

For most UK SMEs, UK GDPR is the more immediate and more likely enforcement risk.

The practical position

Rather than agonising over jurisdiction, build the register. List your AI tools, what they're used for, and whether any output reaches EU clients. That single document answers the scope question and forms the basis of every other obligation. Businesses that skip it end up either over-complying expensively or discovering a gap during a client's due diligence process.

Frequently asked questions

We have one EU client. Are we in scope?

If AI output forms part of what you deliver to them, likely yes as a deployer. The obligations at that level are proportionate — policy, training, transparency.

Does Brexit exempt us?

No. Extraterritorial reach was designed exactly to prevent that, in the same way UK GDPR reaches EU businesses serving UK data subjects.

Will the UK introduce its own AI Act?

Position has shifted several times. As of mid-2026 the approach remains sector regulators applying existing powers, with proposals under discussion. Watch for changes rather than assuming stability.

Not sure which side of the line you're on? Take our free AI compliance assessment.

Compliance
Made on
Tilda