Many UK business owners assume the EU AI Act is someone else’s problem. It isn’t. If your business uses a chatbot, an AI customer service tool, or any AI system whose outputs reach users in the European Union, the Act applies to you — regardless of where your company is registered. Brexit created no exemption here.
The next significant deadline is 2 August 2026. From that date, transparency requirements for “limited-risk” AI systems come into force. With just over a month to go, now is the time to act.
What Is a Limited-Risk AI System?
Under the EU AI Act, “limited risk” covers AI systems that interact directly with people or generate content that people consume. In practice, that means:
- Customer-facing chatbots, including AI assistants on your website
- AI tools that generate text, images, or video intended for human consumption
- Recommendation engines that customers interact with directly
- Any tool that creates synthetic or AI-manipulated content of real people
If you’re using any of these in a way that reaches EU customers — even just occasional French or German visitors to your website — the transparency requirements apply from August.
What Do You Need to Do by 2 August?
The August obligation is primarily about disclosure. Users must be clearly informed when they’re interacting with an AI system. Specifically:
Chatbots: Users must be told upfront that they’re talking to an AI, not a human. A clearly labelled bot notice at the start of the conversation is typically sufficient, but it needs to be prominent and unambiguous — not buried in a terms page.
AI-generated content: If your business uses AI to generate content presented to users as human-created — marketing copy, personalised recommendations, product descriptions — appropriate disclosure is required.
Deepfake or synthetic media: Any content that manipulates how real people appear or sound must be flagged as AI-generated.
The good news: the high-risk AI deadline has been delayed to December 2027. If you operate AI systems in areas like recruitment, credit scoring, or critical infrastructure, you have more time. But the limited-risk obligations are still landing in August — and most customer-facing AI tools fall into this category.
What Happens If You Don’t Comply?
Fines under the EU AI Act can reach €35 million or 7% of global annual turnover, whichever is higher. Enforcement will initially focus on larger platforms rather than small businesses, but the framework is in place and EU digital regulation has a consistent track record of broadening enforcement over time. GDPR started similarly.
Beyond fines, there’s the reputational angle. Customers increasingly expect to know when they’re talking to an AI. A clear disclosure label builds trust rather than undermining it, and getting ahead of compliance means you’re not scrambling in September.
Getting Prepared
The practical steps are manageable for most SMEs. Start by auditing which AI-facing tools your business runs: your website chatbot, any AI email or ticket handling, any AI-generated content pipelines. Check whether each one has appropriate disclosure language in place at the point of customer interaction.
If you need help mapping your AI use to the compliance requirements, ApplyAI works specifically with UK businesses on practical AI adoption and governance — they can help you identify which tools need a disclosure label and which fall outside the scope of the Act.
For the legal paperwork — updating your privacy notices, terms of service, or drafting an AI usage policy that reflects your actual use — Smallprint offers legal document templates designed for UK small businesses that can be adapted quickly.
The Bottom Line
The EU AI Act is not a future concern. If your business uses AI systems that reach EU users, the August 2026 deadline is less than six weeks away. In many cases, a clearly labelled chatbot and an updated privacy notice is all it takes to stay compliant. The effort is small. The risk of inaction isn’t.
