An AI use policy for a small office: a template you can copy

A page of clear rules prevents most accidents. Copy this, change the bracketed parts, and have everyone read it.

This template is general and is not legal advice. Adapt it to your organisation, your clients' agreements and the law that applies to you. Review it every six months, because tools and rules change.

Why a policy

Staff will use AI tools whether you have a rule or not. Without a rule, someone will paste a client file into a personal account. A short policy tells them what is allowed and what to do when they are unsure.

The template

Purpose. This policy explains how [organisation] staff may use AI tools such as chatbots and image generators at work.

1. Approved tools. Staff may use only these tools for work: [list]. They must use the work account, not a personal one, where [organisation] has paid for one. Other tools need approval from [name].

2. Allowed uses. Drafting and editing text, brainstorming, summarising public documents, explaining concepts, writing formulas and small scripts, and preparing training material.

3. Data that must never be entered. Identity numbers (Aadhaar, PAN, passport), bank and card details, passwords and keys, clients' or students' personal details, health records, salary data, unreleased results and exam papers, and anything covered by a confidentiality agreement. If in doubt, do not enter it.

4. Removing details. Where AI help is useful on a real case, staff must first replace names, numbers and addresses with placeholders.

5. Checking. The person who uses an AI draft is responsible for the final text. Before use, check every fact, figure, name, date, rate and legal reference against an official or reliable source. Nothing is sent to a client, an authority or the public without being read in full by a person.

6. No invented sources. Staff must not put a quotation, case, law or statistic in a document unless they have found it in the original source.

7. Honesty. Staff must not present AI output as a person's personal experience or as real photographs of real events or products. Where clients or students ask how we prepared something, we answer truthfully.

8. Settings. Staff must turn off the use of chats for model training where the tool allows, use temporary or incognito chats for sensitive questions, delete chats that are no longer needed, and not click feedback buttons on chats with sensitive content.

9. Records. For important work, note which tool was used and who checked the result.

10. Mistakes. If someone enters data that should not have been entered, they must tell [name] at once and delete the chat. Reporting a mistake is expected and will not be punished; hiding one will be treated seriously.

11. Review. [Name] reviews this policy every six months.

How to introduce it

1. Share it in a short meeting and explain the reasons with two or three real examples.
2. Ask everyone to read and sign.
3. Put the approved tools and the contact person on a card near the computers.
4. Review after a month: what is working, what is annoying, what is missing.

For schools

Add a section for students: when AI is allowed for homework, how to say that it was used, and the consequences of passing AI work off as your own. Add a rule for teachers about not entering student data.

For clients

Tell clients if you use AI tools on their work and what you do to protect their data. A short sentence in your engagement letter or terms is enough: "We may use AI tools for drafting; we do not enter your personal or confidential data into public tools."

This is an independent guide. It is not connected with OpenAI, Anthropic, Google, Microsoft or any other AI company; product names belong to their owners. Statements about companies are taken from their own help pages as read on 13 and 14 November 2026. The prompts are our own wording and were not run against every tool. Not legal, financial or professional advice.