What does AI do well in social media work?
The tasks AI handles best are ones where you supply the facts and it supplies the wording. Turning a voice note about a finished job into a caption, rewriting a Facebook post as a shorter Google Business Profile update, suggesting angles for a quiet month, or matching a written style guide are all reasonable uses. It also helps with repurposing: taking one longer piece, such as a newsletter, and drafting several short posts from it. In each case the output is a draft. The time saved is in producing the first version, and the checking still takes a person.
- Drafting captions from photos and notes you supply
- Adapting one post for Facebook, Instagram, LinkedIn and Google Business Profile
- Suggesting post ideas from a list of your services and seasons
- Keeping wording consistent with a written voice guide
Where does AI get social posts wrong?
Language models predict likely wording, so they fill gaps with plausible details. In social posts that shows up as invented prices, wrong suburbs, opening hours that changed last month, claims such as 'same-day service' that nobody approved, and statistics with no source. A model can also misread a photo and describe a bathroom as finished when it shows work in progress. Tone can drift into generic marketing language that sounds unlike the business. None of this is visible unless someone compares the draft with the facts. The ACCC says a business must be able to prove any claim it advertises, however the words were produced.
What should never go into an AI tool?
Customer names, addresses, phone numbers, health details and anything from a customer's file should stay out of general AI tools. The OAIC recommends, as best practice, that organisations do not enter personal information, and particularly sensitive information, into publicly available generative AI tools. It also notes that privacy obligations apply to personal information put into an AI system and to output that contains personal information. For social posts this is rarely a constraint, since a caption about a deck repair does not need the owner's name or street. Strip identifying details from notes before they go in, and check photos for house numbers and number plates.
Who checks the drafts, and against what?
Every draft needs a check by someone who knows the facts. Australia's guidance for AI adoption says an organisation remains accountable for how and where it uses AI, and recommends testing before use, monitoring afterwards and maintaining human control. For a small business that means one named approver, a short list of checks and a record of what was approved. The approver should be someone who knows the job or the offer in each post. The checks can be the same every time:
- Every price, date, location and service matches current facts
- The photo shows what the caption describes
- There is no promise about timing, availability or results
- Anyone identifiable has agreed to appear
- You would be comfortable reading the post to the customer involved
Should you tell people AI helped write your posts?
Platforms label some AI content themselves. Meta requires people to disclose photorealistic video or realistic-sounding audio that was digitally created or altered, and labels images when it detects industry-standard AI indicators. Those rules do not cover a caption drafted with AI and checked by a person. The government's guidance for AI adoption recommends disclosing AI use as standard practice, especially where content could meaningfully affect people or be mistaken for a human. Many businesses describe how they produce content on their website instead of in each post. Our guide on labelling AI content covers each platform's rules.
What about AI-generated images?
Generated images of your work are the riskiest use for a trade or service business. A photorealistic image of a kitchen you did not build, or a team you do not employ, misrepresents what you offer, and the ACCC lists images and descriptions of what is offered among the information that must be accurate. Generated images may also carry metadata that platforms read and label. Use real photos of real jobs, with permission, for anything that shows your work. Keep generated or stock images for clearly illustrative uses, such as a seasonal reminder graphic, and never present them as your own projects.