AI Regulation Tracker  /  Regulator guidance

New Zealand Medical Council says doctors stay fully responsible for clinical decisions when they use AI

The Medical Council of New Zealand has published guidance telling registered doctors they may use AI in patient care but remain accountable for every decision, must tell patients in some situations, and must not let AI impersonate a doctor.

By Anthony Guerriero  ·  Reviewed by The Leveraged Years Editorial Desk  ·  Published July 7, 2026  ·  Last updated July 7, 2026

The Medical Council of New Zealand has published guidance telling registered doctors that they may use artificial intelligence in patient care, but that responsibility for the resulting clinical decisions stays with them. The document, Guidance on using artificial intelligence (AI) in patient care, was published on 10 March 2026 by the Council, which regulates every doctor licensed to practise in New Zealand. It does not ban any tool. It sets the professional standard a doctor is expected to meet when a tool is used.

What the guidance requires

The Council frames AI as something a doctor may adopt only on the doctor's own professional judgment. According to the guidance, the doctor remains responsible for all their clinical decisions and actions. That principle sits at the center of the document and shapes the rest. Before relying on a tool, a doctor is expected to be satisfied that it is safe and suitable for the particular patient in front of them, rather than assuming that a product cleared for general use fits every case.

The guidance also addresses the patient relationship directly. It tells doctors to obtain informed consent for the use of AI in some specific situations, which places a disclosure duty on the doctor rather than on the vendor or the institution. The guidance does not attempt to list every such situation, so a doctor is expected to apply judgment about when the use of a tool is significant enough that a patient should be told and asked. Alongside consent, the Council asks doctors to consider potential bias and the effect of a tool on health equity, and to ensure adequate privacy and data security safeguards are in place for the patient information the tool processes. Together these expectations treat an AI tool as part of the clinical encounter, not as a back-office utility that patients never need to hear about.

The line on impersonation

One of the sharper instructions is that doctors must not use AI tools in ways that impersonate a doctor. As AI systems increasingly draft messages, answer patient questions, and generate clinical text, the Council is drawing a boundary. A patient should be able to know when they are dealing with a clinician and when they are dealing with software. This matters most for patient-facing chat tools and automated messaging, where the risk of a patient mistaking a machine for their doctor is highest. The instruction is also a warning to vendors who design tools to sound like a named clinician, because the doctor who deploys such a tool carries the professional consequence if a patient is misled.

What it does not do

The guidance is not legislation and does not create a new medical-device approval pathway. It does not certify or prohibit specific products, and it does not shift liability onto manufacturers. It works through the Council's existing authority over professional conduct, which means a doctor who ignores it risks being assessed against it in a competence or conduct process, not fined under a statute. The document supplements the standards doctors already work under rather than replacing them, so existing duties of competence and good practice continue to apply to AI use as they do to any other clinical tool.

The cross-border angle

For a United States reader, the guidance does not bind US doctors, and it does not carry the force of the Food and Drug Administration's device rules. Its value is as a signal. New Zealand has now stated, in plain regulatory language, that accountability for an AI-assisted decision rests with the licensed clinician. It reads as a trans-Tasman companion to Australia's Therapeutic Goods Administration work on AI as a medical device, with New Zealand addressing the practitioner's duty where the TGA addresses the product. Health-technology vendors selling into New Zealand should expect their clinical customers to ask harder questions about safety, bias, privacy, and consent, because the Council has now made those questions the doctor's responsibility to answer.

What clinical leaders should do

Clinical directors and hospital risk teams can treat the guidance as a checklist. Map every AI tool currently in use against the Council's areas of expectation, confirm who is accountable for each decision the tool touches, and make sure the consent and disclosure practice is documented where the guidance calls for it. The standard is now written, and the Council expects doctors to be able to show they have met it.

Browse the full AI Regulation News tracker

Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.