Skip to content

A one-page AI use policy you can fill in today

Most owners have never written down what their staff can and can't paste into ChatGPT. This is the one-page policy we fill in with owners during the workshop, and it works fine without us. Fill in the blanks, share it with your staff, and nobody's left guessing what they can paste into ChatGPT.

How to use it

  1. 1. Fifteen minutes

    Set aside about fifteen minutes. Most of the blanks are things you already know.

  2. 2. One owner

    Pick one person to own the policy. Their name goes wherever the template says [name], so your staff know who to ask.

  3. 3. Every blank

    Fill in every blank, including the ones that feel obvious. A blank left empty tells your staff that rule doesn't matter.

  4. 4. Somewhere visible

    Share it where your staff will see it, like the shared drive they open every day or the wall by the time clock. An email from three months ago doesn't count.

  5. 5. A review date

    Put a review date on the calendar before you close the file. These tools change often, and your answers will too.

The template

This is a starting point written by workshop instructors, not legal advice. If your business handles regulated data, have a lawyer read your filled-in version.

[Company] AI Use Policy

Effective [date]. Applies to all employees and contractors.

This policy is owned by [name].

1. Approved tools

We use these AI tools for company work, on company accounts only: [tools and account types]. Personal accounts are not used for company work. To get a tool added to this list, ask [name].

2. What can go in

Fine to put into approved tools: [examples: drafts, public information, templates, job postings]. Never put in without written approval from [name]: customer personal information, employee records, company financials, passwords or credentials, anything under an NDA, [anything specific to your business]. If you're not sure which list something belongs on, ask before you paste it.

3. Check the output

You are responsible for anything you use or send, whether AI drafted it or not. Before AI output leaves the company: read the whole thing, check every number, name, date and price against the source, and fix what's wrong. These tools state wrong things confidently. Checking is part of the job.

4. Disclosure

[State when, if ever, you tell customers AI was used. For example: "Client-facing work is always reviewed by a person before it goes out, and we say so if asked."]

5. Mistakes

If AI output caused an error that reached a customer or a decision, tell [name] the same day. The point is fixing the process, not blaming the person who caught it.

6. Exceptions

[Name] decides exceptions to this policy, in writing.

Reviewed every [quarter / six months]. Last reviewed [date].

Download the template as a Word file

It's a .docx file that opens in Word or Google Docs, and we don't ask for your email to download it.

In the workshop, writing this policy is part of Sunday morning, after you've built your data map, so the blanks get filled in against what your business actually handles. If you'd rather start with us, the free live session is below.

Free live session

Start your data map with us, free

In the free live session we run ChatGPT, Claude and Copilot on real business paperwork, and you start sorting what your company can put into them, which is what this policy is built on.

Next session announced soon.

Optional.

Optional. For event reminders and workshop announcements by text.

Text message consent

Read the Privacy Policy and Terms.

Read the Privacy Policy and Terms.

By joining the waitlist you agree to receive email from AI Weekend Workshop about upcoming workshops and, occasionally, other services from Philips-Spalding, LLC. Unsubscribe anytime.