← All guidesAI & BusinessNovember 2026 · 5 min read

Should You Tell Customers Your Business Uses AI?

Usually you do not have to, but sometimes you should, and occasionally a label will not save you. The UK advertising regulator says there is no blanket legal requirement to disclose the use of AI in ads. What matters is whether a customer would be misled. If you drafted an email with an AI assistant, checked it and sent it under your name, nobody needs a label. If a customer could wrongly believe they are talking to a person, looking at a real photograph or reading a real review, you have a problem that needs fixing. This is general information, not legal advice.

What the advertising rules actually say

The Advertising Standards Authority (ASA) and its sister body, the Committee of Advertising Practice (CAP), have no AI-specific rules. Their position is that the existing advertising codes apply in the same way however a piece of marketing was made, and that you remain responsible for your marketing even when it is produced or sent out by automated tools.

In its advice on disclosing AI in advertising, CAP suggests two questions to ask yourself:

  1. Is the audience likely to be misled if the use of AI is not disclosed?
  2. If so, does the disclosure make the message clearer, or does it contradict it?

CAP says a disclosure on its own is very unlikely to rescue a message that is misleading at heart. Its example is an AI-generated image showing what a cosmetic product does, when real results do not look like that. Adding "made with AI" in small print does not put that right.

CAP also points out that advertisers have retouched images and edited sound for years without labelling every change. Tidying a background or brightening a product photo with an AI tool is treated in much the same way, as long as the result does not mislead. Other countries and some advertising platforms have their own labelling rules, so check the current rules wherever you advertise.

When you do not need to say anything

If a person at your business reads, corrects and approves the words before they go out, they are your words. This covers most everyday use:

  • drafting emails, product descriptions and social posts that you then edit;
  • tidying grammar, shortening copy or changing the tone;
  • brainstorming titles, subject lines and ideas;
  • small, honest edits to your own photographs.

The test is simple: is what the customer reads true, and do you stand behind it? If you are not sure which tasks fit here, the guide to everyday jobs for an AI assistant is a good place to start.

When you should be open

Use CAP's first question as your guide. A customer is likely to be misled, and you should say something or change what you are doing, in these situations.

  • They are talking to software. If a chat window or phone line is answered by an automated assistant, say so at the start and give a way to reach a person. A line such as "This is our automated assistant. Ask for a member of the team at any time" does the job. There is more on this in the guide to AI customer service with a human touch.
  • A picture looks like evidence. If an AI-generated image could be taken for a photograph of your actual product, premises, staff, customers or results, do not use it that way. Show the real thing, or make it obvious the image is an illustration.
  • A "person" is not a person. CAP notes that making clear an influencer or presenter is AI-generated could help avoid a misleading impression.
  • People rely on the advice. If you publish guidance on health, money or legal matters, have a suitably qualified person review anything AI helped to write, and consider saying who checked it.

What no label will fix

Some uses are off the table, disclosed or not.

  • Fake reviews. The Competition and Markets Authority's guidance on consumer law describes fake reviews, meaning reviews that look or claim to be genuine but are not, as banned content. If you publish customer reviews on your site, you are expected to take reasonable and proportionate steps to prevent and remove fake ones.
  • Invented testimonials. Under the CAP Code you must hold documentary evidence that a testimonial is genuine, have contact details for the person who gave it, and have their permission to use it. Writing a review of your own product while posing as a customer breaks the Code.
  • Made-up facts. AI assistants can produce statistics, case studies and quotes that sound right and are not real. If you cannot trace a claim to a source, leave it out.

Tell people how their information is used

This part is separate from advertising, and it is a legal duty. If customers' personal details pass through an AI tool, for example when you paste an enquiry into an assistant to draft a reply, data protection law applies. The Information Commissioner's Office (ICO) is clear that there is no AI exemption. Your privacy notice should say what you use people's information for, how long you keep it and who you share it with, and that includes the suppliers of the tools you use. The practical steps are in the guide to using AI tools safely with customer data.

A short statement you can adapt

A few honest lines on your About or Contact page cover most situations. Adjust this to match what you really do:

We use AI tools to help us draft and organise some of our content and replies. A member of our team reviews what we publish and send. Our reviews and testimonials come from real customers. If you would like to speak to a person, contact us using the details on this page.

Only say what is true. If nobody reviews the chatbot's answers, do not claim they do.

What to do this week

  1. List every place AI touches something a customer sees: website copy, images, emails, chat, adverts, review replies.
  2. For each one, ask CAP's first question: would a customer be misled if they did not know?
  3. Label any automated chat and add a clear route to a person.
  4. Replace any AI image that could pass for a real photograph of your product or results.
  5. Check that you can prove every testimonial on your site is genuine.
  6. Update your privacy notice, and write your approach into a simple AI policy for your team so everyone follows the same rules.

Sources

Comments

Leave a comment

Comments are checked before they appear. Your name and comment are shown publicly. Your email address, if you give one, is never shown and is only used if we need to reply to you. See our privacy policy.

Free newsletter

New guides, straight to you

Practical tips for selling online, and each new guide as it is published. Occasional emails, nothing to buy.

Free. We email you a link to confirm. Unsubscribe any time. See our privacy policy.