The most common problem we find when we ask an AI tool about a health practice isn’t that the practice is invisible. It’s that the AI is confidently wrong. It lists a physio who left two years ago. It says the clinic bulk bills when it doesn’t. It describes a psychologist as “specialising in trauma” because a directory listing said so.
Patients read that answer before they ever reach your website. So for health practitioners, AI search is as much an accuracy and compliance question as a visibility one.
Where the wrong information comes from
AI tools don’t invent your practice details from nothing (usually). They stitch together whatever they can find: your website, your Google Business Profile, Healthdirect and other directories, booking platforms, old staff bios on a previous employer’s site, a local news story from 2021, your reviews.
When those sources disagree, the AI picks one, and it doesn’t always pick the newest. The practical consequences we see most often:
- Departed practitioners still listed on a booking platform or an old page that was never deleted.
- Outdated fees or billing (bulk billing, gap fees, NDIS or DVA status) from a page that hasn’t been touched since the last price change.
- Wrong hours or locations after a move, especially when the old Business Profile was never closed.
- Services you’ve stopped offering still described on a forgotten blog post.
- Language that’s yours but now isn’t compliant, picked up from old copy written before you knew the rules.
That last one is the one that matters for AHPRA.
The AHPRA problem with being “visible”
Most of the generic advice for appearing in AI answers is fine for a plumber and risky for a health practitioner. “Show strong reviews”, “highlight results”, “say what you specialise in” all run straight into section 133 of the National Law.
AHPRA’s guidelines for advertising a regulated health service say advertising must not be false or misleading, must not use testimonials, must not create an unreasonable expectation of beneficial treatment, and must not encourage unnecessary use of health services. Gifts and discounts need their terms stated.
This is where that bites when you’re writing for AI visibility:
“Specialist” and “specialises in”. Unless you hold specialist registration, AHPRA’s guidance is that words like “specialist”, “specialises in” or “specialised” are likely to mislead. The guidelines suggest phrasing like “substantial experience in” is less likely to be a problem. So “Jane has a particular interest and substantial experience in perinatal mental health” rather than “Jane specialises in perinatal mental health”. It’s slightly clunkier. It’s also the sentence an AI will repeat about you, so it needs to be right.
Outcome claims. AI answers love outcomes (“this clinic helps people recover from back pain faster”). Don’t give them that sentence to quote. Describe what you do, who it’s for and what an appointment involves, not what the patient will get.
Testimonials. This is the big one. Under AHPRA’s testimonial guidance, a review that mentions symptoms, diagnosis, treatment or outcome is a testimonial and can’t be used in your advertising. Reviews that only talk about customer service (“easy to book, friendly reception”) aren’t testimonials.
This is general information, not legal advice. If you’re unsure about specific wording, talk to your indemnity insurer or your professional association.
Reviews, AI summaries and who’s responsible
AI tools increasingly summarise reviews into a sentence or two. You might find an AI answer that says “patients say Dr Lee fixed their chronic shoulder pain”. You didn’t write that, and the AHPRA testimonial guidance says advertisers aren’t responsible for reviews on platforms they don’t control, or on sites that aren’t advertising a regulated service. Patients are allowed to post reviews.
Where you are responsible is anything you control. That includes:
- Your website. No embedded review widgets that pull in clinical reviews automatically.
- Your practice social media accounts.
- Anything you share, quote or “feature”.
AHPRA also warns that engaging with reviews on third-party sites can be treated as using them to advertise. So thank people politely, don’t repeat their clinical details back, and never reply with something like “so glad your migraines have gone!” We cover replies in more detail in what you can and can’t do with Google reviews as a health practitioner.
Making your website the source of truth
The fix for inaccurate AI answers is boring and effective: make one place on the web clearly correct and keep it current. That should be your website, backed by your Business Profile.
For each practitioner, a profile page with:
- Full name and profession as registered (for example “Psychologist” or “Clinical Psychologist” only if they hold that endorsement)
- AHPRA registration number, so patients and tools can verify it
- Days they work and at which location
- Areas of interest, written in compliant language
- What a first appointment involves and how long it takes
For the practice as a whole, a fees and billing page that states plainly whether you bulk bill, what the gap is likely to be, whether you see NDIS, DVA or WorkCover patients, and whether a referral is needed. If an AI is going to describe your billing, give it a current, unambiguous page to describe.
Google’s guidance on its AI search features is that these features draw on its normal ranking and quality systems, value first-hand, expert content, and can use Business Profile information for local businesses. In other words, well-written, accurate, regularly updated pages are the method. There isn’t a special AI version.
Then clean up the other sources: close duplicate Business Profiles, update your listing on booking platforms and directories, and ask any previous employer or clinic to remove old bios. Delete or update old blog posts that describe services you no longer offer.
Auditing what the tools say about you
Pick three questions a new patient would ask and put them to ChatGPT, Gemini and Google’s AI Mode. For example:
- “Who are the psychologists at [practice name] and what do they see people for?”
- “Does [practice name] in [town] bulk bill?”
- “Good [profession] in [town] for [common reason people book]”
Note anything wrong and trace where it came from. Ask the tool directly: “What’s your source for that?” Most will tell you. Then fix the source, not the AI. There’s no form to submit to correct a chatbot, but there is usually a stale directory listing you can update.
If the third question never brings up your practice, that’s a visibility problem, and our AI search readiness checklist walks through the website, profile and directory work in order.
The trade-off worth accepting
Compliant copy is sometimes less punchy than what your competitors publish. Some of them are using “specialist” when they shouldn’t, or embedding five-star reviews about pain relief. In the short term an AI might quote them more readily. Over time, inaccurate claims are a regulatory risk, and accurate, specific, well-structured pages are what both Google and patients reward.
We designed and built the website for North Coast Psychology, and AHPRA-aware copy is part of how we approach health sites through our web design service. When an AI tool describes your practice wrongly and you can’t find the source, an AI search audit can trace it with you and keep the fixes within the advertising rules.