How to write a simple AI policy for your team

By the end of this, you will have a clear, one-page AI policy your team can follow, written in plain English and shaped around how your people actually work.

If you manage a team, a short, friendly policy does two useful things at once. It frees people to use AI with confidence, and it protects the business from the few things that genuinely matter, such as leaking client data. It does not need to be long, and it certainly does not need a lawyer to start.

What you will need

• Any AI chat tool to help you adapt the wording: Claude, ChatGPT, or Gemini.
• A rough sense of what your team does day to day, and any rules your industry already imposes.
• About 20 to 30 minutes.

Step 1: Decide what it needs to cover

A good small-team policy answers five questions in plain terms: what is encouraged, what to keep out of AI tools, who is responsible for the output, when to be open about using it, and who to ask when unsure. Jot down anything specific to your world before you start, for example client confidentiality in a law firm or safeguarding in a school.

What to expect: a short list of your own concerns to fold into the template below. If you are not sure what matters, the template covers the common ground.

Step 2: Start from this template

Copy the starter policy below. It is deliberately short and human. The parts in square brackets are for you to fill in.

[Team or company name]: our approach to using AI

Why this exists
We want everyone to use AI tools (such as ChatGPT, Claude, and Gemini) with
confidence and good judgement. This is a short, living guide, not a rulebook.

What we encourage
- Use AI to draft, summarise, brainstorm, and take the grind out of routine work.
- Treat it as a fast assistant whose work you always check before it goes out.

Protecting information
- Do not paste confidential, personal, or client information into public AI tools.
- When in doubt, leave it out, or ask [name or role] which tool is approved.
- Use our approved tools for anything sensitive, as public tools may keep and
  learn from what is typed in.

Checking the work
- You are responsible for anything you send, whatever helped you produce it.
- Always check facts, figures, names, and quotes, because AI can sound sure and
  still be wrong.

Being open
- Mention it when AI has meaningfully shaped work that others rely on as your own
  judgement. There is no need to flag routine help like fixing grammar.

Where to draw the line
- Do not let AI make final decisions about people, such as hiring or performance,
  without a human weighing in.
- Do not present unchecked AI output as expert advice.

Questions
- Unsure about anything? Ask [name or role]. It is always fine to ask.

Last reviewed: [date]

What to expect: a complete first draft you could use almost as-is. It is written to be understood by everyone, not to impress a compliance team.

Step 3: Adapt it to your team

Now make it yours. Paste the template into an AI tool along with your notes from Step 1 and let it tailor the wording. A clear prompt does most of the work.

Here is a draft AI policy for my team, and some notes about how we work.
Please adapt the policy to fit us, keep it to one page and in plain
English, and flag anything important I might have missed for a
[your industry] team.

[paste the template and your notes]

What to expect: a tailored version in your team’s language. Read it through and change anything that does not sound like you.

If it does not work: if it becomes long or corporate, reply with “Shorter and warmer, please, and no jargon.”

Wrapping up

Share the policy somewhere everyone can find it, and tell people it is a living document they can question. A policy people understand and revisit beats a detailed one that sits unread.

Keep it human: the goal is confidence, not fear. Pair the policy with a quick chat about what AI is good and bad at, so your team feels equipped rather than policed. Our guides on what AI can do for your job and keeping the human in charge are good to share alongside it.