The Australian healthcare system is unprepared to maximise the opportunities and manage the risks of artificial intelligence (AI), according to a new report putting forward dozens of recommendations for change.
The report, titled ‘Don’t Walk. Run. AI in healthcare – The 3rd National Policy Roadmap’, calls for urgency and agility in responding to AI disruption.
Melissa Sweet writes:
As the rapid uptake of artificial intelligence (AI) disrupts health systems and services, the safety and quality of healthcare in Australia cannot be guaranteed under existing regulatory and governance systems, warns a new report released today.
“There is a clear and increasingly urgent role for Australian governments and agencies to set the conditions for safe and effective adoption of AI services in healthcare,” it says.
Australia is also at risk from our dependence on international technology providers and needs to urgently develop sovereign capacity in AI in healthcare, it says.
“If our nation’s citizens receive most of their health information from online sources which can be algorithmically manipulated by foreign corporations or nations, then the nation is at risk of direct interference,” says the report.
It identifies significant ethical concerns, such as AI platforms harvesting patients’ health information for other uses, and AI being embedded in healthcare without patients’ consent.
“For example, consenting to an MRI already implies consent to the many algorithms within the MRI software,” it says.
As the use of AI scribes becomes increasingly common in healthcare settings, the report recommends that “conscientious objection to the use of AI by consumers should be respected”.
The report also identifies an urgent need to upskill the health workforce in AI design, development, implementation, evaluation, and governance.
It calls for a national Chief AI Officer training program to be resourced to produce a new generation of executive leaders and to upskill existing leaders, to drive procurement, governance, and implementation.
It suggests the National Health & Medical Research Council and Medical Research Future Fund aren’t up to speed on AI, noting that the 2026 update of the National Health and Medical Research Strategy “is largely silent on how we should support AI research or support the use of AI in research”.
Context
This is the third edition of a report produced by the Australian Alliance for Artificial Intelligence in Healthcare, presenting a national policy roadmap for AI in healthcare, following iterations in 2021 and 2023.
The Alliance hoped the 2023 report would lead to a national plan for an AI-enabled Australian healthcare system, but says that in 2026 there is still no co-ordinated national approach to advancing the safe and effective use of AI in healthcare.
“Australia appears still unprepared to manage the opportunities and risks of an AI-enabled health system,” the report says.
The 2026 report was developed through engagement with government, industry, peak bodies, and consumers. More than 96 organisations contributed to an initial survey in 2025 and these draft recommendations were then tested in focus groups and interviews with 38 people from 20 peak bodies and government agencies.
Macquarie University, the Digital Health Cooperative Research Centre, CSIRO’s Australian eHealth Research Centre, and RMIT University supported this work.
“In the last two years it has become clear that the pace of industry growth far exceeds our current capabilities to ensure the technology is safe and effective,” Professor Enrico Coiera, the Alliance’s founder, writes in the 2026 report foreword.
“It is becoming clear that system disruption is arriving fast. A handful of ‘hyperscaling’ international companies dominate the AI market and have turned their gaze to healthcare. They have ample resources to move quickly and at scale in their attempt to become the de-facto front- door for consumers to manage their health.
Coiera, Professor of Medical Informatics and Director of the Centre for Health Informatics at Macquarie University’s Australian Institute of Health Innovation, said: “For Australia’s healthcare system to remain sustainable and resilient, it will need to shift gears, and quickly, from catching up with overseas industry, to moving ahead and defining our own pathway.
“If there is to be wholesale change to healthcare, then it must be on our terms.”
The report puts forward 28 recommendations under the themes of leadership, governance, sovereign capabilities, accelerating adoption of AI across healthcare, workforce, consumers, and research and development.
These are timely given widespread concerns, as reported at Croakey recently, that health and healthcare do not appear to have high priority in the Federal Government’s recently announced plans for national standards for artificial intelligence (AI).
The Alliance’s report does not address the public health issues associated with rapid AI uptake, whether in healthcare or wider society, being focused on AI’s impacts on healthcare, rather than providing a health impact assessment of AI. Nor does it address the impacts of AI upon the health workforce itself.
The report recommends research funding should prioritise Aboriginal and Torres Strait Islander led research with support from Aboriginal Community-Controlled Organisations as lead researchers to produce high-impact, culturally grounded evidence to guide AI policy and practice. It acknowledges the importance of Indigenous Data Sovereignty.
Indigenous Data Sovereignty
It comes as a new publication by Aboriginal and Torres Strait Islander scholars, titled ‘Indigenous Data Sovereignty and the Governance of Artificial Intelligence: Toward Equitable Benefit for Indigenous Peoples’, underscores the critical importance of Indigenous governance of AI systems.
The article’s authors – Associate Professor Jacob Prehn, Professor Bronwyn Carlson, Professor Maggie Walter, Professor Ray Lovett and Dr Bobby Maher – are executive members of Maiam nayri Wingara, the Aboriginal and Torres Strait Islander Data Sovereignty Collective, and of the Global Indigenous Data Alliance.
They say Indigenous peoples have the right to govern how data are created, collected, interpreted, and applied, ensuring this is done in ways that are accountable to community values, cultural protocols, and collective governance structures.
“It is important to emphasise that AI is not rejected outright; it can have many benefits, but it should be engaged critically and strategically through Indigenous perspectives,” they write in Journal of Sociology.
“Such interventions shift Indigenous peoples from objects of algorithmic classification to architects of AI design and governance, showing that, under Indigenous jurisdiction, AI can support cultural continuity, self-determination, and nation-(re)building.”
Leadership
The AI roadmap report calls for development of an ambitious, well-resourced National AI in Healthcare strategy, and a National Health AI Expert Group to be set up to advise government on “a coherent and strategic approach to national adoption of AI in healthcare”.
“Left to its own devices, our fragmented health system has defaulted to solving the AI challenge in silos that poorly coordinate with each other, double up effort and miss critical pieces of the solution,” says the report.
“Governments and agencies are setting up multiple AI committees and sub-committees, and parallel work is underway in the non-government sector.
“We need harmonisation of these activities to avoid waste, duplication, and inertia.
“A fully funded national plan, crafted in collaboration by the key stakeholders from government, agencies, the health system, industry, and consumers would go a long way to solving this problem.”
Governance
The report also makes recommendations to promote governance of AI for safety, quality, ethics and security, including ensuring that healthcare organisations have robust clinical governance processes.
Effective post-market surveillance of AI is also urgently needed “so that new risks are rapidly detected and communicated”.
The report highlights regulatory gaps and overlaps with multiple agencies involved, including the Therapeutic Goods Administration (TGA), the Australian Commission on Safety and Quality in Health Care (ACSQHC) and the Australian Digital Health Agency (ADHA).
While clinical AI is subject to TGA regulation, non-medical generative AI falls into a grey zone. Although used for clinical purposes, it evades scrutiny because it is a general purpose technology not explicitly intended for healthcare.
“Uploading or processing sensitive patient data into a non-medical AI chatbot hosted in servers outside Australia is also problematic from a privacy and consent perspective,” says the report.
“The current model for regulating AI was designed over 30 years ago in an environment where technology was single function and predictable.
“Today, governance needs to move from a ‘certify once’ model to one which ensures adaptive AI remains fit for purpose as it evolves where in use. Localisation of AI models by health services will be a common strategy to deal with data biases, and governance is needed to ensure local AI services function as expected.”
Ethical guidelines will need to grapple with many issues, including patient consent for secondary use of their health data and AI history for model training.
Routine patient safety incident reports are a potential source of monitoring data for AI but are fragmented, with each state running their own process, and no formal mechanism for primary care reporting.
Harmonising and integrating patient safety reporting should be a national priority, says the report, and generative AI products, such as AI scribes, require extra scrutiny because safety assessment will have to rely more heavily on post-market mechanisms.
Sovereign capabilities
The report recommends Australia develop independent national sovereign capabilities to build and operate large-scale AI.
Australian healthcare organisations must have access to AI that is calibrated to the characteristics of specific local populations, services, and health needs, says the report.
“There is a view in some quarters that all we need to do as a nation is adopt the best of what is produced internationally, and that we do not need deep AI sovereign capabilities,” it says.
“Nothing is further from the truth. Without some degree of algorithmic sovereignty (the capability to produce or modify AI in Australia), the nation is exposed to new risks and will miss one of the most significant industrial revolutions of our times.”
Other recommendations
The report recommends developing and sharing best practice expertise in the selection, implementation, and operation of clinical AI, as well as procurement guidance to help health services identify proven AI technology.
It highlights the importance of training for diverse workforces, including clinicians, health service administrators, public health workers, business managers and investors, science and technology researchers and developers.
“To prepare the sector for the increased use of AI, we will need to support the creation of national consensus on foundational clinical competencies, scopes of professional practice, and codes of professional conduct to use AI, and provide a basis for patient safety, service quality, and practitioner credentialling.”
The report highlights the challenges for consumers in navigating AI-driven health services, including the risks of misinformation and bias.
“For example, if they use any of the common publicly available GenAI chatbots, they may read information tailored to US and not Australian conditions, such as vaccination schedule recommendations, which differ for Australia and the US,” it says.
On the positive side, it also describes the potential for AI to enhance access to and navigation across and within critical services such as My Aged Care and 1800 Medicare.
It says the community also needs upskilling: “If patients, carers, consumers, and citizens are to be involved in co-designing AI-enabled services, and advising on ongoing management, this will require a level of literacy currently inaccessible to most Australians.”
Some progress
The report acknowledges policy progress since the 2023 road map, including:
- Adoption of a whole of government approach to AI oversight, led by the Federal Department of Industry, Science and Resources.
- Completion of a Legislation and Regulation Review for the safe and responsible use of AI in Healthcare by the Federal Department of Health, Disability and Ageing.
- Creation of an AI Safety Institute by the Australian Government, focusing on AI safety policy for government across all industry sectors.
- Launch of AI implementation guidelines in hospitals by the Australian Commission on Safety and Quality in Healthcare.
- Release of the Medical Software Industry Association and Medical Technology Association of Australia AI governance code, a voluntary medical software industry framework.
- Establishment of an AI taskforce by NSW Health to inform use of AI in the NSW public health system.
- Establishment of a Technical Reference Group for AI by the TGA, and provision of advice by the TGA on the safe use and regulation of generative AI.
- Establishment of an Artificial Intelligence Enabled Care Expert Advisory Group by the Australian Digital Health Agency.
- Publication of an Advisory on the use of unregulated AI in the Victorian Public Health System.
- Inclusion of specific advice to all health professionals on their obligations when using AI by the national professional regulator Ahpra.
- Professional guidance on the safe and effective use of AI scribes and LLMs by the Royal Australasian College of General Practice, the Royal Australian and New Zealand College of Radiologists, the Australian Physiotherapy Association, and the Australian College of Nursing.
- The funding of four national AI Adopt Centres, intended to support Australian businesses. Unfortunately, says the report, no healthcare-specific centre was funded.
• Marie McInerney will cover The Royal Australian and New Zealand College of Radiologists conference on AI on Friday for the Croakey Conference News Service. See her preview article: Radiology conference asks hard questions about AI use in healthcare.
See Croakey’s archive of articles on AI and health
Add Comment