---
title: "AI policy template for a small business: six rules to adapt (2026)"
url: https://while.sh/en/guides/ai-policy-template-small-business-2026/
author: "Aloïs Oddon, co-founder of Nextmap"
updated: 2026-11-18
---

# AI acceptable use policy for a small business: the template to adapt in 2026

Your employees already use AI, and some already turn it into real tools. A one-page policy is enough to say what is allowed, what is not, and where the tools go. Here is a template in six rules, to adapt to your company.

In your company, someone already uses AI. A salesperson has ChatGPT proofread their quotes, an assistant summarizes contracts with Claude, a workshop manager built themselves a small scheduling tool. In 2024, Microsoft and LinkedIn's Work Trend Index found that **78% of people using AI at work brought their own tools**, and 80% at small and medium-sized companies (source 1).

Banning does not work, and neither does letting it happen. A **one-page policy** simply says what is allowed, what is not, and where the tools AI builds go. Here is a template in six rules, to adapt to your company.

> **The short version.** A good small business AI policy fits on one page: the allowed AIs, the data to keep to yourself, checking what AI produces, a shared home for the tools built, credentials out of the code, and the person to contact. It works all the better when a tool applies it on its own.

## The six rules of the template

Each rule answers a risk you already know, and it is written in one sentence everyone understands.

1. **The allowed AIs.** Say which ones, and with which account. A company account is better than a personal one, because the company keeps control over what it contains.
2. **Data to keep to yourself.** List what never goes into an AI: client data, salaries, passwords and software credentials.
3. **Check before sending.** What AI produces gets reread before it goes to a client or into an official document.
4. **A shared home for tools.** A tool built with AI, that others use, goes to a single place everyone knows, kept for employees.
5. **Credentials out of the code.** The credentials of company software live neither in a tool's code nor in a conversation.
6. **A person to contact.** When in doubt, or if a tool must open to a client or to everyone, you contact a designated person, who approves.

## The template to copy

Here is the text of the template. Replace the brackets, keep what helps you, and remove the rest.

```text titre="AI acceptable use policy"
AI acceptable use policy of [company name]

1. We use [Claude, ChatGPT, other], with the company account.
2. We never give an AI our clients' data, salaries, or the passwords and credentials of our software.
3. We reread what AI produces before sending it to a client or using it in an official document.
4. A tool built with AI, that others use, goes on [the shared home], kept for employees.
5. Our software credentials go neither into a tool's code nor into a conversation.
6. To open a tool to a client or to everyone, or when in doubt, we contact [name], who approves.

This policy is reread every year. Last review: [date].
```

<figure class="illus"><img src="/illustrations/charte-regles-1200.webp" srcset="/illustrations/charte-regles-768.webp 768w, /illustrations/charte-regles-1200.webp 1200w" sizes="(min-width: 48rem) 46rem, 100vw" width="1536" height="1024" loading="lazy" decoding="async" alt="The six rules of a small business AI policy: allowed AIs, data to protect, checking, a shared home for tools built with AI, credentials out of the code and a person who approves" /><figcaption>Six rules, one page, and a place where they apply on their own.</figcaption></figure>

## A policy that applies on its own

A policy states the rule, but nobody rereads it when sharing a tool. It works much better when **the place where tools live applies it on its own**. That is what we wanted when we built while: rules 4, 5 and 6 are built in.

- **Rule 4, a shared home.** Your employees keep the AI of their choice, and it puts their tools on while. On the Pro plan, everyone signs in with their email account, Google or Microsoft. See [keep a tool for your colleagues](/en/guides/internal-tool-company-sign-in/).
- **Rule 5, credentials out of the code.** Software credentials [stay in while](/en/security/), and the AI never sees them. Those left in the code are detected.
- **Rule 6, a person who approves.** Opening a tool to everyone waits for approval from someone on the team, and the history shows who did what. A client uses a tool with [an access link and a password](/en/guides/share-app-with-password/), with no public link.

> You, to your AI: Put this schedule on while, for employees only.

Your tools are hosted in France by Nextmap. To go further, read [shadow AI: govern the tools your employees build with ChatGPT or Claude](/en/guides/shadow-ai-employee-tools-chatgpt-claude-2026/).

## Keep the policy alive

An AI policy ages fast, because AI changes every month. Name the person for rule 6, reread the policy every year, and start with a tool that already exists: its author puts it in the shared home, and the others follow.

## Try the shared home

To try while, give this message to your AI. The trial is **free, with no sign-up**, and your tool has its address **in 3 minutes**.

```text titre="The message to give your AI"
Read while.sh/agents and help me get started with while.
```

## Sources

1. [Microsoft and LinkedIn, Work Trend Index 2024: AI at work is here. Now comes the hard part](https://www.microsoft.com/en-us/worklab/work-trend-index/ai-at-work-is-here-now-comes-the-hard-part), 2026-11-18

## Questions

### Why does a small business need an AI policy?

Because your employees already use AI. In 2024, 78% of people using AI at work brought their own tools, 80% at small and medium-sized companies (source 1). A policy says what is allowed, instead of leaving everyone to guess.

### What should an AI policy contain?

The allowed AIs and with which accounts, the data never to give an AI, checking what it produces, where the tools built live, and the person to contact when in doubt.

### Should you ban ChatGPT at work?

A ban pushes usage into the shadows. Choosing the allowed AIs and giving the tools a home works better.

### Is a policy enough?

It states the rule, but you also need a place where the rule applies on its own: on while, tools are kept for employees through company sign-in, credentials stay in while, and opening to everyone waits for a person.

### Who should write the policy?

Management, with the person who handles IT. It fits on one page, and gets reread every year, at the pace AI changes.
