Write a simple AI use policy your team will follow

Your team is probably already using AI at work, with or without permission. That's the point. A use policy isn't there to slow them down. It draws bright lines, so nobody pastes a customer's private data into a public chatbot. You can write a useful one in an afternoon.

What to include

• Approved tools.Name which AI tools are okay for work, and which aren't. A short allow-list beats "use good judgment" every time.
• What never to paste in.Draw the red line clearly: customer data, passwords, contracts, and anything confidential. Don't paste it unless the tool is approved and privacy settings are right.
• What needs a human check.Say which outputs a person must review before they're used. Don't skip customer-facing work, legal language, financial numbers, or decisions about people.
• Disclosure.Say when you'll tell customers or staff that AI was involved. Keep it tied to the work, not a vague principle.
• Who owns the output.Make ownership blunt. AI output is a draft your team owns, and it doesn't ship unread.
• Who to ask.Name the person for questions and exceptions. If you don't, people will guess.
• How to ask for an exception.Give people a simple way to ask, "Can I use X for Y?" Then the policy can bend without breaking.

A one-page starting template

Keep it short enough to enforce

A policy nobody reads protects no one. Keep it short. Fit it on a page, use plain words, and make the two rules impossible to miss. What can't people paste? What needs a human check? You can add detail later when a real situation demands it.

Review it when the tools change

AI tools change monthly. Don't treat the policy like stone. Put a review date on it, then keep that date. A new tool, feature, or type of work can move a bright line. A living one-pager beats a perfect document that's already stale.