Book a fit chat

Free tool · One page

An AI policy your team will actually read.

Answer the pad and the sheet next to it writes your business's AI use policy as you go: which tools are approved, what never goes in, who checks the output and what happens when something goes wrong. Copy it or print it. It fits on one page.

Checked against OAIC and National AI Centre guidance · 8 Nov 2026 · Runs in your browser, nothing is sent

What should a small business AI policy say?

Four things, on one page. Which AI tools are approved, and that they are used through business accounts rather than personal ones. Which information never goes into an AI tool: at minimum health information, tax file numbers, bank and identity details, passwords, and anything a client asked you to keep confidential. When a person has to check the output before it is sent or relied on, which for most firms is anything going to a client or touching money or advice. And what to do when something goes wrong, so a mistake gets reported the same day instead of hidden. Add a line on meeting recordings if you use an AI note-taker, and a line saying no AI tool makes a decision about a person on its own. Name an owner and a review date, or the policy goes stale within a year.

Change any answer and the policy rewrites itself. Nothing you type leaves this page.

01 · The business

Your industry

02 · Approved tools

Tools the team may use for work
Accounts

03 · What never goes in

Never put this into an AI tool
Client information in approved tools

04 · Checking and telling

A person checks before it's used
Also include
Telling clients
Review the policy every

The method

What goes in the policy, and why.

Why one page?

A policy only works if the person at the desk remembers it when they're about to paste something. Long policies get read once at induction. This one has a short section for each decision a staff member actually faces, and nothing else.

Which tools should be approved?

The ones you already pay for, on business accounts. Business plans from Microsoft, Google, OpenAI and Anthropic don't train their models on your business content by default; personal and free accounts follow different terms. Start with what is built into your email and documents, and add a standalone tool only if the team needs it. The policy also asks for approval before anyone switches on a free trial, a browser extension or a new AI feature in an app you already use.

What should never go into an AI tool?

The OAIC's guidance on commercially available AI says the best practice is not to enter personal information, particularly sensitive information, into publicly available AI tools. The default list covers health information, tax file numbers, bank, card and identity details, passwords and keys, confidential or privileged material, and staff records. Everything is ticked to start with; untick only what your approved tools are genuinely set up for.

When does a person have to check the output?

At minimum, before anything goes to a client or the public, or affects money or advice. AI drafts can be confidently wrong about facts, figures, names and dates, so the policy says to check those and makes the sender responsible for what goes out.

What do the industry clauses add?

  • Accounting or bookkeeping: professional standards and confidentiality apply to AI-drafted work, and figures are re-checked against the source.
  • Law or conveyancing: confidentiality and legal professional privilege apply in full, and a lawyer checks anything drafted for a matter, citations included.
  • Allied health or a clinic: health information is sensitive information, stays out of general AI tools, and AI never makes or changes a clinical decision.
  • Finance or insurance broking: client financial details stay out of unapproved tools, and advice and disclosure obligations apply to AI drafts.
  • Trades, retail or hospitality: quotes, prices, measurements and safety information are checked before they reach a customer.

What happens when something goes wrong?

The policy asks for a same-day report with no blame attached, because a mistake that gets hidden costs more than one that gets fixed. If personal information may have gone somewhere it shouldn't, the owner decides whether it is a data breach that has to be reported under the notifiable data breaches scheme.

Sources

General information, checked 8 November 2026. It is not legal advice. For the reasoning behind each section, read what a small firm's AI policy should actually say.

Straight answers

Before you ask.

Is there a free AI policy template for Australian businesses?
Yes. The National AI Centre's Guidance for AI Adoption includes an AI policy template and an AI register template. They are thorough and written for organisations with someone to run them. This builder produces something shorter: one page a small team will read.
Do we need an AI policy if only a couple of us use AI?
Yes, and it is quicker to write now. Two people pasting client details into personal accounts is already the problem a policy prevents. The OAIC's guidance says businesses using AI products should have clear policies, and that the best practice is not to enter personal information, particularly sensitive information, into publicly available AI tools.
Should we just ban AI tools?
A ban mostly moves the use onto phones and personal accounts, where you can't see it. Approving one or two tools on business accounts, and being clear about what never goes in, gives you more control than a ban does.
Is this policy legal advice?
No. It is general information and a starting point. If you are in a regulated profession, check it against your professional and client obligations, or have your adviser look over it, before you adopt it.

The policy is the easy half

Want help making it true?

A policy says which tools are approved. Setting those tools up so the team uses them properly is the work. Fifteen minutes on a fit chat tells us both whether I can help.

More free tools: automated decisions privacy check · which AI subscription to pay for

WhatsApp0450 253 297