Most writing about AI governance is aimed at large organisations with compliance teams, risk committees and budgets to match. Small organisations are often left with two bad options: copy a framework built for a multinational, or do nothing and hope for the best.

There is a middle path. A small business, charity, practice or public body can set out its approach to AI on a single page. It will not answer every question, but it will make sure the most important ones have been asked, and that someone is responsible for the answers.

I advise leadership teams adopting AI responsibly. This article sets out the one-page starting point I would suggest for a small organisation.

This article is general information, not legal advice. Organisations should check the legal and regulatory requirements that apply to them.

Why small organisations need AI governance too

It is easy to assume governance is only for organisations at scale. But AI tools are now available to almost everyone, often free or built into software people already use. In a small organisation, staff may be using several AI tools without anyone having decided that they should.

That is not necessarily a problem. Many of those uses will be sensible. But without a shared understanding, an organisation cannot know what information is going where, who is checking the output, or what it would say if a client or customer asked how AI is used.

Governance, at its simplest, is just deciding these things on purpose rather than by accident.

An AI governance framework for small businesses in the UK: the one-page version

The page has eight short sections. Each one should be answerable in a few lines.

1. Purpose

What are we using AI for, and why? List the tasks. Be specific: drafting customer emails, summarising meeting notes, analysing survey responses. If you cannot say what a tool is for, you cannot judge whether it is working.

2. Inventory

Which AI tools are in use, and by whom? Include tools built into software you already pay for. You may be surprised by the list.

3. Information rules

What information must never go into which tools? At minimum, decide how you will treat personal data, client or customer information, and anything confidential. Where personal data is involved, UK data protection law applies whether or not AI is used.

4. Human review

Which outputs must a person check before they are used, sent or published? Anything that affects a customer, a client, a member of staff or a decision about someone should be reviewed by a person who understands it.

5. Ownership

Who is responsible for AI use in the organisation? In a small organisation this may be one person, often the founder, director or operations lead. What matters is that someone is named.

6. Transparency

What will you tell people about your use of AI? Decide when you will tell customers, clients or the public that AI tools were involved in something they receive. A short public statement can help. I publish my own on How I use AI.

7. When things go wrong

What happens if AI output causes a problem: a wrong figure sent to a customer, a confidential detail entered into the wrong tool, a decision questioned? Decide who staff should tell and what happens next.

8. Review

When will you look at this page again? AI tools and the rules around them are changing quickly. A review every few months is reasonable for most small organisations.

How to use the page

Write it with the people who actually use the tools, not just for them. They know where AI helps and where it causes problems.

Keep it short. A one-page policy that people read and follow is worth more than a long one nobody opens.

Share it. Staff should know where to find it, and new staff should see it when they join.

Revisit it. Treat the first version as a draft. The questions will stay the same, but the answers will change as your use of AI changes.

What the one-page version does not cover

A single page is a starting point, not a complete governance programme. Organisations in regulated sectors, those handling large amounts of sensitive data, and those using AI in decisions that significantly affect people will need to go further. That might include more formal risk assessments, procurement checks on AI suppliers, staff training and closer attention to sector-specific rules.

Public sector organisations face particular expectations of accountability and transparency to the people they serve. I have written separately about what a council-wide Copilot adoption programme teaches about AI governance.

How I work with organisations

Through Institute of AI, I advise CTOs, COOs and Directors of Digital in public sector organisations, private-sector firms and small businesses, including NHS bodies and legal technology firms, on adopting AI responsibly.

Organisations can engage me for:

  • AI Strategy Workshops
  • Advisory Retainers
  • AI Governance Keynotes

This is AI governance advisory, not legal advice.

If you would like help turning this one-page starting point into something that fits your organisation, read more on the Advisory page get in touch about advisory work, or book a 15-minute call. You can also read more about Institute of AI.

More writing on this theme is collected under AI ethics and Public-sector AI.