AI Use Policy Starter
A one-page AI policy your team will actually read.
Five short sections: why, which tools, what never goes in, when a human reviews, and where exceptions get written down. Fill in the blanks, adapt the words, and you have a working policy today — improvable forever.
Free
- No email
- Works offline
- ~30 minutes
Download my filled copy
Nothing you type here is stored or sent. Download a filled copy before closing the page.
Before you start
This is a starting policy, not a legal document, one page your team can actually read, covering the five decisions that matter. Fill the underlined fields, delete what doesn’t apply, and resist the urge to make it longer: a policy that fits on one page is a policy that gets followed.
- AI use at Effective · Owned by · Next review 1 · Why this exists
We use AI tools to work faster and think better. This page keeps that safe: it says which tools we use, what never goes into them, and when a person checks the output. It protects our clients’ trust and our own name.
- 2 · The tools we use
We use the tools on our inventory, nothing else for work. approves new tools; approval takes one message and adds one line to the inventory.
Guidance: pair this with the AI & Tool Inventory from this same free set, the policy points at the inventory instead of listing tools that will be stale in a month.
- 3 · What never goes in
Client names tied to their situations. Passwords and keys. Financial account details. Other people’s personal information. Anything under NDA. When unsure, we ask first — asking is free.
Guidance: edit this to match your own never-enter list from the Safe Input Guide.
- 4 · When a person reviews AI work
Anything AI helped produce that reaches a client, a contract, a price, or the public is read by a person who understands the work before it goes out. No exceptions when we’re busy, busy is when it matters.
- 5 · Exceptions and open questions
Exceptions are allowed — and written down in our decision log the same day, with who approved them. An exception that isn’t written down didn’t happen.
- Acknowledged by
Keeping it alive
Put the review date in the calendar now. A policy without a re-read date is a screenshot, not a rule.
Read it aloud once with the team. Two minutes. The questions people ask are the policy working.
Change it when it’s wrong. If an exception keeps recurring, the exception is the real rule — write it in.
Educational operating guidance only. Not legal advice, cybersecurity certification, or regulatory compliance certification.
This page runs entirely in your browser and sends nothing anywhere. It summarizes, in its own words, ideas from public frameworks, the NIST AI Risk Management Framework and Playbook, the NIST Generative AI Profile, and CISA’s small-business resources. The originals are linked in Official Sources, part of this same free set.
- Free Sunnyworks guide