What AI can and can't do for small business social posts

The draft reads well, but you can't say whether it's true. That's the usual problem with AI-written posts, and it tells you where the line sits. The tool is good with words. It isn't good at knowing your business.

Why a smooth draft can still be wrong

Text generators write what sounds likely, not what has been checked. Australia's privacy regulator warns that these systems can give false answers. It calls the risk especially high for generative AI.¹

So a post can have the right tone and the wrong opening hours. It can promise a service you don't offer. It can invent a feature.

What the tool is good for

Makers describe the same jobs. One scheduling company says its tool is for brainstorming ideas and drafting posts before you schedule them.²

In practice that covers:

• Turning a rough note into a few wordings to pick from.
• Shortening a long post for a smaller space.
• Suggesting angles when you're stuck for an idea.
• Tidying spelling and grammar in something you wrote.

None of these needs the tool to know a fact. You supply the facts. It supplies the phrasing.

What needs a person

Anything a customer could rely on needs a human check. The ACCC, the Australian consumer regulator, says social media statements must be true and accurate, and that a business has to be able to prove them.³

Its list of covered items includes prices, product descriptions, claims about value or performance, and shipping and delivery details.³ Those are lines a tool can get wrong or invent.

That rule is Australian. Other countries have their own consumer rules, so check the regulator where you trade. Labels on AI pictures are a separate matter, covered under platform rules.

A person also needs to judge:

• Whether the post sounds like your business. Try getting your voice across.
• Whether the timing is right. A cheerful post on a bad news day can hurt.
• Whether a name, date or place is correct.
• Whether anything private slipped in.

Keep customer details out

Keep customer names, messages and health details out of public chat tools. The Australian regulator's advice is to avoid entering personal or sensitive information into public generative AI tools.¹ That's Australian advice. Check the privacy rules where you operate.

Who is responsible

You are. The tool isn't a publisher and doesn't answer to your customers. Even makers keep the person in charge. One says its AI can't publish on its own, because people stay in control of reviewing, approving and publishing.²

Build a check into every post. The checklist lists what to look at, and the weekly routine shows where the check fits.