How to write an AI policy for a small team

A one-page AI policy a 5 to 50 person UK firm can adopt this week: approved tools, what never goes in, human review, copyright and data protection.

A workable AI policy for a small firm fits on one page. It names the tools staff may use, lists what must never be typed into them, says a person checks anything that leaves the building, and gives one named person the job of keeping it current.

You need one because your staff are almost certainly using AI already. The Office for National Statistics found that 55% of employees reported using AI for work or education in mid-2026, while only 35% of businesses with 10 or more staff said the business itself used it.

  • Approve specific tools on business accounts, so people are not left to use personal ones.
  • Ban three things from every prompt: personal data (unless the tool is approved for it), client confidential material, and passwords or keys.
  • A named human is responsible for every output. AI gets things wrong and states them confidently.
  • Write down what you decided and review it every six months, because the tools and the law are both moving.
Staff are ahead of their employers
55%of employees in Great Britain reported using AI for work or education
35%of businesses with 10 or more staff said the business used AI, June 2026

Source: ONS, Artificial intelligence in UK businesses: 2023 to 2026, July 2026

Why a small firm needs an AI policy at all

The gap between what staff do and what the business has approved has a name. The National Cyber Security Centre (NCSC) calls it “shadow AI”, and in a September 2026 blog it cited a Microsoft study in which 71% of employees reported using AI tools their employer had not approved.

The NCSC’s concern is practical. When staff put company or customer information into consumer AI services, that information “may be stored, retained or used to improve the service” outside your control “unless specific privacy controls are in place”. Its advice is not to ban AI: it is explicitly “not recommending that individuals stop using AI”. It says the goal “should be to reduce risk rather than assume it can be eliminated”, and that staff need approved tools that meet their needs.

Acas, the workplace advisory body, says the same from the employment side: “Employers should develop clear policies regarding the use of AI at work and should consult workers and any representatives on its introduction.”

A one-page AI policy you can copy

The policy on one page

Step 1 of 6

Tools

Name the approved tools and the business accounts staff must use.

Step 2 of 6

Never paste

Personal data, client confidential material, credentials and sensitive figures stay out.

Step 3 of 6

Check

Whoever uses an output is responsible for checking it before it leaves.

Step 4 of 6

Disclose

Say when customers are dealing with AI and when clients should be told.

Step 5 of 6

Report

Tell a named person the same day if something goes in that should not.

Step 6 of 6

Review

Keep a log of decisions and revisit the policy within six months.

Use the headed sections below as your outline. Replace the examples with your own tools and names, and keep the finished document to a single page.

1. Purpose and scope

One or two sentences. “This policy covers the use of AI tools (chat assistants, AI features inside our software, image and transcription tools) for any work done for the company, on any device. It applies to employees, contractors and temps.”

2. Approved tools

List them by name, with the account type. For example: “the company’s paid chat assistant, signed in with your work email” and “the AI features in our office suite”. State that personal or free accounts are not to be used for work, and say who to ask if someone wants a new tool added. Before approving a tool, check its business terms for whether prompts are used for training, how long data is kept and where it is stored.

3. What must never be pasted in

  • Personal data about customers, staff or applicants, unless the tool is on the approved list for that purpose.
  • Client confidential information, including anything covered by a contract or non-disclosure agreement.
  • Credentials: passwords, API keys, access tokens, bank details.
  • Commercially sensitive material such as unpublished accounts, pricing and source code, unless the tool is approved for it.

Acas puts the reason plainly: workers “should be careful entering any information that is business sensitive or personal into public tools”, because it “could be made public or used by others”.

4. Human review and accuracy

“Whoever uses the AI output is responsible for it. Check facts, figures, names, quotes and references against a real source before sending or publishing.” The NCSC warns that these tools can “present incorrect statements as facts”, and Acas advises that “outputs should be checked for accuracy, tone and bias”. List any work where AI must not make the decision, such as hiring, disciplinary matters, credit or anything with a legal effect on a person.

5. Telling people when AI was used

Decide your own rule and write it down. A sensible default: tell customers when they are talking to a chatbot, tell clients if a deliverable was substantially AI-generated where your contract or their expectations call for it, and never present AI output as a named person’s professional opinion without that person reviewing it. Acas suggests that “AI should be cited when used”. For advertising, the Advertising Standards Authority said in May 2025 that there is “no blanket legal requirement in the UK to disclose the use of AI in ads”, but the content must not mislead.

6. Copyright and ownership

“Do not ask AI tools to copy a named author’s, artist’s or competitor’s work. Do not paste in third-party material we have no right to use. Add meaningful human input to anything we want to own and protect.”

7. Data protection

“UK GDPR applies whenever personal data goes into or comes out of an AI tool. Any new use involving personal data needs sign-off from [name] before it starts.”

8. Security and reporting

“Keep AI tools behind your work login with two-step verification. Do not connect them to email, files or other systems without approval. If you paste in something you should not have, tell [name] the same day.” A no-blame reporting line matters. As the NCSC puts it: “You cannot manage what you do not know.”

9. Owner, record and review date

Name the policy owner, keep a short log of tools approved and decisions made (date, what, why), and set a review date no more than six months ahead.

The government consulted on copyright and AI and published its Report on Copyright and Artificial Intelligence in March 2026, as required by the Data (Use and Access) Act 2025. Two points matter for a small business.

  • Training data. The report says a broad copyright exception with an opt-out for rights holders is no longer the government’s preferred way forward, and that it will not reform copyright law until it is confident the changes meet its objectives. The existing law therefore still applies.
  • Who owns AI output. UK law currently gives some protection to “computer-generated works” with no human author, under section 9(3) of the Copyright, Designs and Patents Act 1988. The report proposes removing that protection, while copyright would continue to protect works created with AI assistance.

As of November 2026 we found no legislation implementing either point, so treat this as a direction of travel. The practical lesson is the same either way: work with real human authorship is on firmer ground than raw AI output, so have your people shape, edit and add to what the tool produces. That applies to everyday marketing copy too, as our guide to writing product descriptions with AI explains. For anything valuable (a logo, a product name, software you sell), take legal advice.

What the ICO expects

The Information Commissioner’s Office (ICO) treats AI as a data protection matter like any other. All of the Data (Use and Access) Act 2025’s data protection provisions were in force by 19 June 2026, and the ICO is still consulting on or finalising several related pieces of guidance, including on automated decision-making. Its guidance on AI and data protection explains how the ordinary UK GDPR principles apply to information used in AI systems. The ICO notes that this guidance was last updated on 15 March 2023 and is under review because of changes made by the Data (Use and Access) Act, so check it again when you next review your policy.

If you are putting an AI agent in front of customers, our AI agents and UK GDPR checklist goes through them one by one.

Rolling it out

A policy nobody has read protects no one. Talk it through at a team meeting, and ask what people are already using and why. Acas notes that if you expect certain roles to start using AI, “that could mean a change of terms and conditions”, so involve staff early.

Give people a good approved tool on the same day you restrict the unapproved ones. Our explainer on what AI costs to run will help you budget for it, and our summary of how UK small businesses are using AI shows which uses are most common.

Common questions

Is an AI policy a legal requirement?

No law says you must have a document with that title. Your existing duties on data protection, confidentiality and consumer law apply to AI use regardless, and a written policy is the simplest way to show you have thought about them.

Does the policy cover AI built into software we already use?

It should. Assistants inside email, office and accounting software read the same data as the product itself. List them under approved tools and check their settings. The same applies to customer-facing tools such as an AI chatbot on your website or a voice agent answering your phone.

A note and an offer

This is general information, not legal advice. If you would like help choosing tools and turning this outline into a policy that fits your firm, our startup consulting service does exactly that.

Sources

→Keep reading

Related guides.

All insights
AI explained

UK AI regulation and the EU AI Act: what applies to a UK business

The UK still has no AI Act, but the EU's reaches UK firms that sell into Europe. What applies, the 2026 changes and delays,… Read
Startups

How to validate a business idea before you build anything

Five cheap ways to test demand before you spend on a product, what counts as real evidence, when to stop, and the UK name… Read
Startups

Funding a UK tech startup: SEIS, EIS and Innovate UK in brief

The main UK funding routes for an early tech company in late 2026: SEIS and EIS limits, Innovate UK grants and loans, and Start… Read

→Get in touch

Got a question? Ask us.

Tell us what you’re trying to do. Every project starts with a free audit: someone who does the work replies with a few ideas, an honest view and a written plan with prices.

What happens next

  1. 1
    You send the formIt takes about two minutes.
  2. 2
    We do our homeworkWe look at your website and your market.
  3. 3
    You get a short planWhat we’d do first, with prices. Yours to keep.
  • Free, with no obligation
  • We won’t add you to a mailing list

We only use your details to reply to you. See our privacy policy.