Free template

What Should an AI Acceptable Use Policy Include? (Free Template)

Updated November 16, 2026

Short answer: A small firm's AI acceptable use policy needs to answer six questions on one or two pages: which AI tools are approved, what client and firm information may go into them, what is never allowed, who checks AI output before it reaches a client, what to do if something goes wrong, and who owns the policy. Name the approved tool, list the information in the words your staff use, and keep it short enough that people read it. A free template you can copy is below.

Why have one at all?

Because your staff are already deciding for themselves. In a 2024 survey, 52% of employed people said they had received no training on safe AI use (Infosecurity Magazine). And IBM's 2025 breach study found that "63% of breached organizations either don't have an AI governance policy or are still developing a policy" (IBM).

The U.S. government's AI risk framework recommends the same thing. NIST's Generative AI Profile lists, among its governance actions, "Establish transparent acceptable use policies for GAI that address illegal use or applications of GAI" (NIST AI 600-1). NIST's framework is voluntary, but it is a useful outside reference if a client or regulator asks how you set your rules. For the AI data privacy concerns a policy is there to handle, see AI privacy concerns at work.

The six things the policy has to answer

  1. Which tools are approved? Name them, and the plan. "ChatGPT" is not enough, because a personal ChatGPT account and a firm's business plan have very different terms. Say that anything not on the list is not approved for firm work.
  2. What may go in? Sort information into a few groups staff can recognize, and say which tools each group may go into. Use the real nouns of your work.
  3. What is never allowed? For example: client information in any personal account, anything a client contract forbids sharing, and passwords or login details anywhere.
  4. Who checks the output? AI drafts can be wrong. Say that a person reviews anything AI produced before it goes to a client or into a filing.
  5. What if something goes wrong? Tell staff to report a mistake straight away, to whom, and that reporting quickly will not be held against them. Fast reporting matters if your rules give you a breach notice deadline.
  6. Who owns it? Name the person who approves new tools and answers questions, and set a review date.

Mistakes that make a policy fail

  • Banning everything. A policy that approves no tool is a policy staff will ignore. Name at least one approved tool.
  • Writing for lawyers instead of staff. If it takes more than five minutes to read, most people won't.
  • Approving a brand instead of a plan. Personal and business plans from the same company keep data differently. See Can staff use their personal ChatGPT accounts for work?
  • Forgetting AI inside other software. AI features in email, CRM or practice software count too. Say that switching one on needs approval.
  • Never reviewing it. AI tools and their terms change every few months. Put a date on it.

Free AI acceptable use policy template

Copy this into your own document and fill in the bracketed parts. Delete anything that doesn't fit your firm.

[Firm name] AI Acceptable Use Policy

Effective: [date]. Owner: [name, role]. Next review: [date, six to twelve months out].

1. Why we have this policy. We use AI tools to work faster, and we hold information our clients trust us to protect. This policy sets out which AI tools we use for firm work and what may go into them.

2. Who it covers. Everyone who works for or with [firm name], including partners, staff, contractors and temps, on any device, whenever they are doing firm work.

3. Approved tools. These are the only AI tools approved for firm work:

  • [Tool and plan name, for example "[Product] on the firm's business account"]. Approved for: [all work / internal work only].

Any other AI tool, including a personal account with the same company, is not approved for firm work. To ask for a new tool, contact [owner]. AI features inside other software (email, CRM, practice management, note-takers) must be approved by [owner] before they are switched on.

4. What may go into AI tools.

Information Personal or unapproved AI Approved firm tool
Public information Allowed Allowed
Internal firm information with no client details Not allowed Allowed
Client information: [list your nouns, e.g. names, account numbers, census files, statements, returns, contracts] Never Allowed for client work, unless the client's contract or your industry's rules say otherwise
Anything a client contract or NDA says must not be shared Never Only if the contract permits it
Passwords, login codes, keys Never Never

5. Rules for using AI.

  • Use only your firm account for approved tools. Never sign in to an approved tool with a personal account.
  • Use the least client information the task needs. Leave out names and numbers when the task works without them.
  • A person must review AI output before it is sent to a client, filed, or relied on for advice. You are responsible for the work, not the tool.
  • Don't present AI output as a client's or colleague's own words.
  • Don't install AI browser extensions or add AI note-takers to client calls without approval from [owner].

6. If something goes wrong. If client information goes into an unapproved tool, or an AI tool behaves unexpectedly, tell [owner] the same day. Say what went in, which tool, and when. Reporting quickly will not count against you; hiding it will.

7. Leaving the firm. Your access to approved AI tools ends on your last day. Don't keep firm or client information in any AI account.

8. Questions. Ask [owner]. If you are not sure whether something is allowed, ask before you paste.

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

How to roll it out

  1. Pick the approved tool first. The policy is only as good as the tool it names. What makes an AI tool safe to approve has the questions to ask.
  2. Talk it through in a team meeting. Ten minutes, with examples from your own work.
  3. Get it signed. Keep the signed copies with your other staff policies.
  4. Check in. At the review date, ask what tools people use and update the list.

If your firm is an insurance agency, an investment adviser or a tax practice, your industry's rules may ask for more, such as vendor due diligence or a written information security program. The sector risks page points to the rules for each.

About Faradex. Faradex is a private AI workspace for small and mid-sized firms that handle confidential client work. Each firm gets its own Workspace on dedicated infrastructure in the United States. Files and conversations are held in memory and are gone when the Workspace is reset. AI model: Currently Opus via AWS Bedrock, more information is available on our website. AWS does not store prompts or responses, and model providers have no access to them. Pricing: from $120 a month, no seat minimum; see faradex.ai/pricing.