AMA calls for strict Australian health AI regulation as inquiry hears evidence

Australia's current legal framework is falling short when it comes to regulating artificial intelligence in healthcare, the Australian Medical Association told a federal parliamentary inquiry.

Giving evidence on 21 October to the Joint Select Committee on Artificial Intelligence, AMA president-elect Dr Michael Bonning stated that existing rules leave too much room for uncertainty. He urged lawmakers to classify all healthcare AI applications as high risk by default, rather than assessing products on an ad hoc basis.

The danger of software feature creep

While doctors increasingly adopt AI transcription services and diagnostic support tools to save time, Bonning cautioned that machine-generated suggestions can appear technically sound while remaining inaccurate for an individual patient.

Bonning told the committee that drawing a firm legal line between administrative software and clinical software is difficult in practice. Software vendors frequently update and expand their products to stay ahead of rivals. A tool originally bought for basic practice management or patient scheduling can gradually expand its scope, eventually guiding decisions that affect patient care. An autonomous booking tool that cancels an appointment without human review, for instance, could lead to a patient missing necessary treatment.

To address these risks, the AMA outlined five regulatory priorities for Australia:

• Treat healthcare AI as high risk by default, scaling oversight according to the autonomy of the system and its potential to cause harm.
• Establish a dedicated, clinician-led governance body to oversee health AI.
• Create an overarching statutory framework operating alongside existing regulators such as the Therapeutic Goods Administration and the Australian Health Practitioner Regulation Agency.
• Enforce clear legal liability so responsibility stays with whoever controls the system.
• Mandate strong data safeguards, keeping Australian patient information stored within Australia wherever possible.

What this means for Australian businesses

This inquiry carries direct lessons for Australian medical clinics, allied health providers and wellness businesses adopting AI tools.

First, do not rely on product marketing labels. If a vendor advertises a tool as purely administrative, review what it actually does. If the software summarises patient consultations, manages appointment triage or prompts staff about clinical follow-ups, your practice bears professional responsibility for any mistakes.

Second, keep human staff in the loop. Automated systems should draft notes, propose schedules or flag records, but a qualified practitioner must check and approve the final result. In Australia, liability follows control, which means software vendors will rarely take the legal fall for an automated error inside your clinic.

Finally, check where your customer and patient data lives. Many general-purpose cloud AI services process data through overseas servers. Small businesses handling sensitive health records must verify that their software providers meet Australian Privacy Principles and, ideally, keep data stored in domestic data centres.