Back to blog

— NEWS

AI Content Policy: A Plain-English Guide for Small Businesses

By Matt Canty •

Hand-drawn policy and AI tool motifs

An AI content policy is a short, written set of rules for how your business uses AI to create or edit anything you publish: web pages, blogs, ads, emails, social posts and images. It says what AI can be used for, who checks the result, what data stays out of the tools, and when you tell customers AI was involved. For a small business in Wollongong or anywhere else in Australia, one to two pages is enough, as long as people follow it.

We build websites and run SEO for local businesses, so we see how quickly AI-written content ends up on a site with nobody having decided the rules. This guide gives you a policy you can write this week.

CantyDigital
Make Your Website AI Search Ready
CantyDigital provides signal-based SEO and websites built with SEO and LLM compatibility in mind for businesses seeking better search visibility.

TL;DR

  • An AI content policy is a one to two page document covering approved tools, human review, disclosure, privacy and copyright.
  • Australia has no standalone AI law. Existing rules, such as consumer law, privacy and copyright, apply to what you publish however it was written.
  • Every piece of AI-assisted content should have a named person who checks facts, claims and tone before it goes live.
  • Never paste customer details, private business information or other people’s copyrighted material into public AI tools.
  • Review the policy a few times a year. The tools change quickly, and so should your rules.

Table of Contents

Why a small business needs one

Most businesses already use AI. Someone on the team drafts a blog post with a chatbot, someone else generates an image for Facebook, and a freelancer uses a tool for product descriptions. None of it is wrong. The problem is that nobody has agreed on the limits, so the quality and risk depend on whoever happens to be typing.

A written policy fixes three things:

  1. Consistency. Your website sounds like one business, not five different chatbots.
  2. Accountability. When a wrong claim goes live, it’s clear who should have caught it.
  3. Protection. Your customers’ details and your own commercial information stay out of public tools.

It also helps your search visibility. Search engines and AI assistants favour pages that are accurate, specific and clearly written by people who know the subject. Generic AI filler works against that. We cover how this plays out in our guide to AI Overviews and Australian SEO.

What the rules already say in Australia

Australia doesn’t have a separate AI Act. The government’s approach has been to rely on existing laws and voluntary guidance. That means the laws you already follow still apply when AI writes the words.

Three areas matter most for a small business publishing content:

  • Consumer law. The Australian Consumer Law prohibits misleading or deceptive conduct. It doesn’t matter whether a person or a chatbot wrote the claim. If your page says you have “20 years of experience” or “the cheapest rates in the Illawarra” and it isn’t true, the business is responsible.
  • Privacy. If you put customer names, emails, enquiry details or health information into a public AI tool, you may be disclosing personal information. Your privacy policy should also match how you actually handle data. If you need one, see our guide to a privacy policy for Australian websites.
  • Copyright. AI tools can produce text and images that borrow heavily from existing work, and putting someone else’s material into a prompt can create its own problems. Our article on copyright infringement in Australia explains the basics.

The practical lesson: the government’s voluntary guidance on disclosing AI use is optional, but the laws underneath it are not. Don’t wait for new legislation before setting your own rules.

The National AI Centre, a federal body, publishes free guidance on creating an AI policy and being clear about AI-generated content. Search for it on ai.gov.au if you want to read the official version. We’ve used the same general structure below, simplified for a business with a handful of staff.

What to put in your policy

Keep each section to a few lines. A policy nobody reads protects nobody.

1. Purpose and who it covers

One short paragraph. Say the policy applies to staff, contractors and agencies who create content for your business. Say that you support using AI where it helps, within these rules. A policy that only says “no” tends to get ignored or worked around.

2. What AI can and can’t be used for

Split tasks into three groups. This is the most useful part of the whole document.

Group Examples Rule
Fine to do Brainstorming headlines, outlining a blog, summarising your own notes, fixing grammar, drafting a first version of social posts Allowed, with the review steps below
Needs sign-off Service pages, pricing descriptions, case studies, testimonials, anything with legal, medical or financial claims Owner or manager approves before publishing
Not allowed Fake reviews, invented testimonials, impersonating real people, images of real people without their consent, content using customer data Banned

The logic is risk. The more damage a mistake could do, the more checking it needs. A caption for a Friday coffee special doesn’t need the same scrutiny as a page explaining your warranty.

3. Approved tools

List the tools staff may use for work, and say who can approve a new one. Ask staff to use business accounts rather than personal ones. That keeps your content history in one place and lets you adjust settings, such as whether the tool can use your inputs for training.

4. Data rules

This is the section that prevents the most expensive mistakes. State plainly what never goes into a public AI tool:

  • Customer names, contact details, enquiry content or account information
  • Staff records
  • Pricing agreements, contracts or anything under a confidentiality agreement
  • Login details and internal systems information
  • Other people’s copyrighted material, such as a competitor’s full article pasted in for rewriting

A simple rule works well: if you wouldn’t be comfortable seeing it on a public page, don’t paste it in.

5. Human review

Every AI-assisted piece gets read by a person who knows the subject before it goes live. Spell out what they check:

  • Facts. Are names, numbers, dates, locations and legal references correct? AI tools state wrong things confidently.
  • Claims. Can you back up every claim about your business, such as “fully licensed”, “award-winning” or “same-day service”?
  • Local accuracy. Does it get suburbs, services and regulations right? Chatbots regularly mix up Wollongong with other places.
  • Voice. Does it sound like your business, or like every other website?
  • Originality. Does it say something specific from your experience, or only what any competitor could say?

Name the person. “The marketing manager reviews” is better than “content is reviewed”. In a business of three, it might simply be the owner.

If you want more practical ideas for building these checks into a content workflow, there are useful controls for AI-assisted marketing content you can adapt to your own process.

6. Disclosure

Decide when you tell people AI was involved. The sensible approach is proportionate: the more AI contributed, and the more the content could affect someone’s decision, the clearer the disclosure.

A workable starting point:

  • AI-assisted (a person wrote it, AI helped with editing or ideas): no label needed on a blog post, but record it internally.
  • AI-generated images or video that could be mistaken for real photos or footage: label them.
  • Chatbots and automated replies that talk to customers: make it obvious they’re not a person.
  • Anything that changes the meaning of existing content, such as editing a quote or a testimonial: don’t do it. If it’s a real customer’s words, they stay as written.

Never use AI to create a fake person, fake review or fake before-and-after photo. That’s the fastest way to turn a productivity tool into a legal problem.

7. Images, voices and likenesses

Generating or editing images of real people, including staff, customers and public figures, needs documented consent. Add one sentence to your policy: no AI recreation of a real person’s face or voice without written permission. This also applies to old photos of a founder, a late relative or a former employee.

Who owns AI-assisted work isn’t always clear-cut. It can depend on the tool’s terms, what you fed into it and how much human input went into the final result. For your policy, keep it practical:

  • Contractors and agencies must tell you if they’ve used AI on your work.
  • Your agreements should say who owns the final content.
  • Staff don’t paste in material they don’t have the right to use.

If the content matters a lot, such as a brand name, logo or a core product description, have a lawyer review ownership. That’s one place a general policy shouldn’t stand in for legal advice.

9. Reporting problems

Give staff an easy way to say “this AI output looks wrong” or “I accidentally pasted something I shouldn’t have”. Make it a no-blame process. A quick report on day one costs far less than a problem found months later.

10. Review date

Put a date on the document and a reminder to revisit it. A couple of times a year is realistic for most small businesses, and again whenever you adopt a new tool.

A worked example: a Wollongong plumber

Say you run a plumbing business with six staff and a website. Your office manager uses AI to draft blog posts and service pages. Here’s how the policy changes things.

Before: She asks a chatbot for a page about hot water systems. It produces 600 smooth words, including a claim that you offer “24/7 emergency service across the Illawarra” and a line about a manufacturer’s warranty terms. She publishes it.

After: Under the policy, the draft is a first version. She marks the service and warranty claims for sign-off. The owner confirms that after-hours cover only runs to certain suburbs, and the warranty line is removed because nobody has checked the manufacturer’s actual terms. She adds a real job from Thirroul, with the customer’s permission, and a photo taken by one of the team. The page now says things only your business can say.

Before-and-after AI content review workflow

That page is more accurate and safer. It’s also more likely to be cited by search engines and AI assistants, because it contains specifics instead of generic filler.

Common mistakes

  • Banning AI outright. Staff keep using it anyway, just without any rules.
  • Copying a corporate template. A 20-page policy built for a government department won’t suit a five-person business. Keep yours short.
  • No named reviewer. If everyone is responsible, nobody is.
  • Publishing unedited drafts. Raw AI output is the main reason websites read the same.
  • Forgetting contractors. Your freelancers and agencies use AI too. Put the rules in their briefs.
  • Setting and forgetting. Tools, laws and guidance move quickly. An out-of-date policy can be worse than none, because it gives false comfort.

How the policy supports your SEO

A good AI content policy isn’t only about risk. It’s also a quality filter. Search engines reward pages that show real experience, accurate detail and a clear author. AI assistants tend to quote sources that are specific and consistent. Your policy pushes content in that direction by requiring:

  • Local detail: suburbs, projects, real staff and real jobs
  • Verified facts and claims
  • A consistent voice across every page
  • Sensible disclosure that builds trust rather than raising questions

If you’re planning to be visible in AI-driven search, read our 30 to 90 day AI search visibility playbook and our guide to implementing AI SEO strategies in 90 days. Both fit neatly alongside a content policy.

Where we can help

Where we can help

We’re a Wollongong-based digital agency with 12 years of experience in web design and SEO, including optimisation for AI search. Our contracts are flexible with no lock-in, so you’re not tied in while you work out what suits your business.

If you want your website and content built on solid foundations, our Wix web design is a $590 one-off. For ongoing SEO, the SEO Essentials Plan is $170 per month. If you’re curious how your site shows up in AI answers, our generative engine optimisation (GEO) service is built for exactly that, and you can start with a free GEO audit.

CantyDigital

When a news release is part of your plan, our press release distribution across Australian platforms is $290 one-off. It pairs well with a policy that keeps your claims accurate, because a press release is the kind of content that needs a careful read before it goes out.

Our view

Most of the businesses we speak to aren’t worried about AI. They’re worried about being left behind by it. So they use it quickly, and they skip the part where someone decides what “good” looks like.

We think the winning approach is simple. Use AI for speed, then add what it can’t: your local knowledge, your real jobs, your real customers and your honest claims. That’s what search engines and AI assistants are increasingly looking for, and it’s what customers trust.

A written policy is how you make that repeatable. It doesn’t need to be long. It needs to name who checks the work and what never goes into a public tool.

This is also why we build websites and SEO with AI search in mind from the start, rather than bolting it on later. Pages that are accurate, specific and well structured serve readers and machines alike.

Matthew

FAQ

What is an AI content policy?

It’s a short document that sets the rules for using AI to create or edit your business’s content. It covers which tasks AI can help with, who reviews the output, what information stays out of AI tools, and when you disclose AI use. For a small business, one to two pages is plenty.

No, there’s no law that requires a small business to have one. But the laws that apply to your content, including consumer law, privacy and copyright, apply regardless of whether AI wrote it. A policy is a practical way to show you take those obligations seriously and to keep mistakes from reaching your website.

Do I have to tell customers I used AI?

There’s no single rule for every case, so a proportionate approach makes sense. Label AI-generated images that could pass as real, make it clear when customers are talking to a chatbot, and never use AI to invent reviews or testimonials. For a blog post a person wrote and checked, with AI helping on editing, a label is usually unnecessary, though recording it internally is smart.

What should never go into ChatGPT or similar tools?

Customer personal information, staff records, confidential contracts or pricing, login details and other people’s copyrighted material. Treat anything you type into a public tool as something that could become visible to others. If you’re unsure, leave it out or remove the identifying details first.

How often should I update the policy?

Review it a couple of times a year, and whenever you start using a new tool or a new type of content. Put the review date in the document so it doesn’t get forgotten.

Sources

Want results like these?

Let's talk about how SEO and AI search optimisation can grow your business.

Get in touch