Using AI in Your Small Business Marketing? 5 UK Rules to Know Before You Hit Publish

AI can save a small business owner hours.

It can help you brainstorm campaign ideas, turn rough notes into a first draft, create design variations and suggest different advert headlines. Useful stuff when you are also answering emails, serving customers and wondering where the last clean mug disappeared to.

But AI does not take responsibility for what gets published. You do.

The UK’s advertising rules apply whether an advert was written by you, an employee, an agency or an AI tool. Here are five practical rules to keep in mind before you hit publish.

> “The same rules apply in the same way, regardless of whether AI is used or not.” , ASA/CAP

This is general marketing guidance, not legal advice. ASA and CAP guidance can change, so check the current official guidance for any campaign where the risk is high or the wording is unclear.

1. Treat AI-generated advertising like any other advert

The CAP Code covers non-broadcast advertising, including many social media adverts, websites, emails and online promotions. The BCAP Code covers broadcast advertising.

Neither code gives you a free pass because a machine created the content. Existing rules on misleading claims, harmful content, endorsements and social responsibility still apply.

That means you should review:

• The words in the advert

• The images and videos

• The offer and pricing

• The audience being targeted

• Any claims about results, quality or performance

• The landing page customers reach after clicking

AI is a useful assistant, not a magic doughnut button. Use it to speed up the work, not to replace judgement.

For more context, read the ASA’s guidance on generative AI and advertising.

2. Check every claim and keep the evidence

AI tools can produce confident-sounding nonsense. They may invent statistics, qualifications, customer results or product benefits without warning.

Before publishing an objective claim, ask:

• Is it accurate?

• Can I prove it?

• Is the evidence current?

• Does the wording match what customers are likely to understand?

• Have I accidentally exaggerated the result?

For example:

• A shop: Do not claim clothing is “100% organic cotton” unless your supplier documentation supports it.

• A tradesperson: “Same-day boiler repairs” needs to reflect your real availability, not an optimistic AI suggestion.

• A café: “The best coffee in town” may be puffery, but a claim such as “serving 500 customers a week” should be based on real records.

• A B&B: Check that “five minutes from the beach” is genuinely accurate for a normal walking route.

• A professional service: Do not publish invented testimonials, qualifications or case study results.

Keep screenshots, reports, product information, customer feedback and other evidence in one folder. If someone challenges a claim, you should not have to search through 47 versions of “final-final-really-final.docx”.

The ASA’s advertising codes and rulings are a useful place to check how misleading claims are assessed.

3. Do not use AI to fake an endorsement

Deepfakes and realistic synthetic images make this rule especially important.

Do not use an AI-generated likeness, voice or video of a real person in a way that suggests they have recommended your business when they have not. That includes celebrities, local public figures, former customers and competitors.

You also need permission before using a recognisable person in advertising. A small business does not become safer because the fake image was created in three clicks.

Be particularly careful with:

• Fake customer testimonials

• AI-generated “before and after” images

• Synthetic videos of experts recommending your service

• Images that make a product appear larger, better or more effective than it is

• Fake reviews or invented customer stories

Disclosure may help in some cases, such as clearly labelled comedy or fictional content. However, adding “AI-generated” does not fix a fundamentally misleading advert.

The ASA explains this risk in its guidance on disclosing AI in advertising.

4. Be transparent when hiding AI could mislead people

There is currently no blanket UK requirement to label every advert made with AI.

You do not normally need to disclose that AI helped tidy a sentence, remove a background object or create several headline options. Those uses are similar to ordinary editing or design work.

The better question is:

> Would the audience be misled if they did not know AI was involved?

Consider clear disclosure where:

• A realistic AI-generated person appears to endorse a product

• A deepfake is used, even for humour

• An AI-generated influencer is presented as a real person

• A product demonstration could be mistaken for a genuine result

• The use of AI is a prominent part of the advert’s message

Make the disclosure clear and close to the relevant content. Do not hide it in tiny text at the bottom of a busy advert.

5. Check for bias, harmful stereotypes and poor audience fit

AI tools learn from large volumes of existing material. That means their output can repeat lazy assumptions or harmful stereotypes.

Review generated content for:

• Stereotypes about gender, age, race, disability or income

• Images that exclude or misrepresent your customers

• Language that could shame or pressure vulnerable people

• Content that is unsuitable for children or sensitive audiences

• Unfair assumptions about who needs a product or service

This matters even if you did not intend to cause offence. The audience sees the advert, not your prompt or your good intentions.

Also check whether the platform, audience and timing are suitable. An edgy AI-generated joke might be acceptable in a private brainstorming document but completely wrong for a children’s activity provider, care service or financial adviser.

The ASA has highlighted the need to sense-check AI-generated advertising for bias.

A simple pre-publish AI checklist

Before publishing AI-assisted marketing, ask:

• Did a human read and approve the final version?

• Are all factual claims accurate and supported by evidence?

• Have I checked prices, dates, locations, availability and qualifications?

• Does the image show a realistic product or result?

• Could anyone think a real person endorsed this when they did not?

• Should I disclose the use of AI in this particular context?

• Does the content avoid stereotypes, offence and harmful assumptions?

• Does it sound like my business rather than a generic robot?

• Have I checked the current ASA/CAP guidance for this campaign?

• Does the advert lead to a clear, accurate page on my website?

Use AI for ideas, first drafts, design options and advert variations. Keep your real customer knowledge, evidence and final approval firmly in human hands.

If your marketing needs a clearer plan, stronger wording or a website that matches what your adverts promise, explore Donut Pig’s brand and marketing strategy service or website design and hosting service.

We make your brand sizzle like bacon.

FAQs

Do I have to label every advert made with AI?

No. UK advertising rules do not currently require a blanket AI label on every advert. Disclosure may be needed where hiding the AI use could mislead the audience.

Am I responsible if an AI tool makes a mistake?

Yes. The business placing the advert remains responsible for its claims, images, targeting and overall impression.

Can I use AI-generated images of customers?

You can use fictional images, but make sure they do not suggest real people, real results or genuine endorsements. Avoid creating realistic likenesses of identifiable people without permission.

Can AI write my social media posts?

Yes, it can help with ideas and drafts. You should still check the facts, tone, claims, spelling and suitability before posting.

Where can I get more help?

Start with the ASA and CAP advice pages. For practical help applying your brand voice and marketing plan, book a call with Donut Pig Digital Marketing.