Blog · Policy
How to write a one-page AI use policy your team will actually read
Most firms we talk to either have no AI policy at all, or a twelve-page document nobody has opened since the day it was emailed out. Neither one changes what people do on a Tuesday afternoon.
A good AI use policy fits on one page. It tells people what they can use, what they must never paste in, and who to ask when they're not sure. That's it. Here's how to write one.
1. Say why it exists, in one sentence
Start with the point, not the legal background. Something like: "We want everyone to use AI to save time, without putting client data or the firm at risk." It sets the tone: this is about using AI well, not banning it.
2. List the approved tools
Name them. "ChatGPT Team, Microsoft Copilot, GitHub Copilot", whatever you actually pay for. Say that work happens in work accounts, not personal ones, and that anything else needs a quick OK first. A short list people can see beats a long list of rules they have to interpret.
3. Spell out what never goes in
This is the part that matters most, so be concrete:
- Client confidential information, unless the tool is approved for it
- Personal data about clients or staff (health, financial, ID numbers)
- Passwords, API keys and other credentials
- Anything under an NDA or a contract that limits sharing
4. Keep a human in the loop
AI drafts; people decide. Anything AI writes that goes to a client, into production code, or into a contract gets reviewed by a person first. If you use AI agents that can send email, change records or spend money, a human approves those actions.
5. Name an owner and a review date
Someone owns the policy and the tool list. Put their name on it, and a date to look at it again. AI tools change every few months, so a policy without a review date is out of date before anyone notices.
A template you can copy
[Firm name] AI use policy · Owner: [name] · Next review: [date]
Why: We want everyone to use AI to save time without putting client data or the firm at risk.
Approved tools: [tool 1], [tool 2], [tool 3]. Use your work account. Ask [owner] before trying anything else.
Never paste in: client confidential information (unless the tool is approved for it), personal data, passwords or API keys, anything under NDA.
Always: review AI output before it reaches a client, goes into code, or is signed. A person approves any action an AI agent takes on our behalf.
Not sure? Ask [owner] or post in [channel]. Asking is always fine.
Print it, pin it in your team chat, and go through it in your next team meeting. Five minutes of talking it through does more than any PDF.
Want a second pair of eyes on how your firm uses AI? Book a free 20-minute call.