Everyday AI

Using AI at work without leaking customer data: a one-page policy for small businesses

In short: Staff can use AI chatbots safely if you set three rules: no customer or confidential data in public tools, the training setting turned off (or a business plan), and a human checks everything before it's used. Copy the one-page policy below and adapt it.

Last checked: · Next check: · Settings and menus change often.

The main risks

Five steps for a small business

  1. Choose approved tools and list them in one place.
  2. Set them up properly: turn off training on personal accounts (see How to stop chatbots training on your chats), or use a business plan. OpenAI says business workspaces aren't used for training by default.
  3. Set clear "never paste" rules. See What never to paste into a chatbot.
  4. Keep a human in charge: a person checks and owns anything AI helps write before it goes out.
  5. Review every six months, or whenever you add a new tool.

One-page AI policy template

Copy this, fill in the brackets and share it with your team.

[BUSINESS NAME] - AI USE POLICY (1 page)
Version: [1.0]   Owner: [ROLE, e.g. Manager]   Review date: [DATE + 6 months]

1. PURPOSE
We use AI tools to save time on drafting, summarising and ideas.
People stay responsible for everything we send, publish or decide.

2. APPROVED TOOLS
Only use these tools for work: [TOOL 1], [TOOL 2].
Ask [OWNER] before using any other AI tool or app with work information.
Use work accounts where we have them. Turn off "train on my data"
settings on any personal account used for work.

3. NEVER ENTER INTO AN AI TOOL
- Customer names, contact details, addresses or account details
- Payment, bank or card details
- Passwords, access codes or security answers
- Staff records, health information, ID or tax file numbers
- Anything marked confidential, or supplier pricing under agreement
If in doubt, remove names and details first, or don't use AI for it.

4. CHECK BEFORE YOU USE IT
- Read every AI draft in full before sending or publishing.
- Check facts, prices, dates and anything legal, financial or safety-related.
- Don't let AI make final decisions about customers, staff or money.

5. AI AGENTS AND CONNECTED APPS
Don't connect AI tools to work email, files or payment accounts
without approval from [OWNER]. Approved connections use the
least access needed and require a person to approve sends and payments.

6. BE OPEN WITH CUSTOMERS
If customers deal directly with an AI tool (e.g. a chat assistant),
tell them, and give them a way to reach a person.

7. PROBLEMS
If you think private information has gone into an AI tool by mistake,
or AI output caused a problem, tell [OWNER] the same day. No blame.

8. ACKNOWLEDGEMENT
I have read and will follow this policy.
Name: ____________   Signature: ____________   Date: ________

Common questions

Is a business plan worth it just for privacy?

It can be, if staff handle customer information. Business and enterprise plans from the big providers usually don't use your content for training by default, and they add admin controls. Check the terms of the plan you choose.

Do privacy laws apply to my small business?

It depends on your business, so get advice if you're unsure. Treating customer data carefully is good practice either way, and it's what customers expect.

Guide only. No affiliate or sponsored links on this page. Disclaimer · Affiliate disclosure

Related recipes

Work & career

Turn meeting notes into actions

Turn messy notes into a summary with a list of who does what.

Writing & communication

Write a polite complaint or refund email

A firm but polite email that states the facts and what you want.

Writing & communication

Summarise a long email or letter

Get the key points and what you need to do from a long message.