Home / Allied health / Ahpra's AI guidance in plain English, for allied health

Ahpra's AI guidance in plain English, for allied health

What Ahpra and the National Boards actually expect when you use AI scribes, ChatGPT or other AI tools in practice, and a checklist you can work through in one sitting.

7 min read · Updated 10 October 2026 · Applies to any AI tool: scribes, chatbots and note software

Ahpra and the National Boards have published guidance called Meeting your professional obligations when using Artificial Intelligence in healthcare. It doesn't ban AI and it doesn't add new rules. It explains how your existing obligations under your Board's code of conduct apply when an AI tool is involved in your work.

Who it applies to

The guidance applies to every practitioner registered under the National Scheme. In allied health that includes physiotherapists, occupational therapists, psychologists, podiatrists, optometrists, chiropractors, osteopaths, medical radiation practitioners, Chinese medicine practitioners and paramedics.

Speech pathologists, dietitians, exercise physiologists and audiologists aren't registered with Ahpra. They're self-regulated through their professional associations, so the guidance doesn't bind them directly. Privacy law still applies to everyone who handles health information, and the same principles are a sensible standard for any clinician.

The five principles, in plain English

1. You're accountable

Whatever the tool produces, the clinical record and the decision are yours. If an AI scribe mishears a dose or a chatbot invents a contraindication and it ends up in your notes, it's your error. Review and edit every output before it becomes part of a patient's record or care.

2. Understand the tool

Before you use a tool clinically, know what it's designed to do, what it isn't, and its known limits. Check where data is stored and processed, whether it's kept or used to train the model, and whether the tool is regulated by the TGA as a medical device. A general chatbot isn't designed for clinical decisions.

3. Be transparent

Patients should know when AI is part of their care. That can be as simple as a line in your intake paperwork and a sentence at the start of the session.

4. Get informed consent

If a tool records a consultation (an AI scribe) or patient information goes into a tool, the patient needs to understand what's happening and agree to it. Explain what's recorded, where it goes, how long it's kept, and that they can say no without it affecting their care. Record the consent in your notes.

5. Meet your ethical and legal obligations

Privacy law, confidentiality and your code of conduct all still apply. Watch for bias in outputs, keep identifiable information out of tools that aren't approved for it, and make sure anything you publish with AI help (such as website copy) still meets the advertising rules for regulated health services.

Checklist: before you use an AI tool in practice

1. List every AI tool anyone in your practice uses, including free chatbots used "just for letters".
2. For each tool, write down what goes in, where it's stored, whether it's kept or used for training, and who the vendor is.
3. Decide which tools can receive identifiable patient information. For general chatbots without a business agreement, the safe default is none: de-identify first.
4. Write a short consent script and add a line to your intake form for any tool that records or receives patient information.
5. Set a review rule: no AI-drafted note, letter or report is signed or sent until a clinician has checked it against the session.
6. Update your privacy policy. Changes to the Privacy Act add new requirements from 10 December 2026: see the AI privacy checklist.
7. Put it in writing as a one-page practice AI policy, and revisit it when you add a tool.

A prompt to draft your patient explanation

This doesn't need any patient information. Describe your tool and the AI writes a plain-English explanation you can use on intake forms or read aloud before a session. Check it against what your vendor actually does.

I'm a [profession] in a [solo / small group] practice in Australia.
I use [tool name] to [e.g. record sessions and draft clinical notes].

How it works: [e.g. audio is transcribed, a draft note is created, audio is deleted after 7 days, data is stored in Australia]

Write a plain-English explanation for patients, under 150 words, that covers:
- what the tool does and why I use it
- what is recorded and where it goes
- that I review every note myself
- that they can say no at any time and it won't affect their care

Then write a one-sentence version I can say at the start of a session.
Don't make any claims about security or compliance that aren't in my description.

Common questions

More for allied health professionals

• AI privacy checklist for allied health practices: the 10 December 2026 changes New Privacy Act requirements about automated decisions start on 10 December 2026. Here's what they mean for a practice that uses AI scribes or chatbots, and a checklist to get ready.8 min read
• Draft de-identified SOAP notes with AI (physio guide) Turn quick shorthand after a session into a structured SOAP note, without putting patient data at risk.7 min read
• Summarise assessment results for an OT or NDIS report Get from raw assessment scores to a clear, structured summary section faster.6 min read