Deakin to lead new $5 million ARC Training Centre for AI governance
The Australian Research Council has awarded Deakin University funding to lead a new $5 million ARC Training Centre focused on AI governance, oversight and assurance, under the Industrial Transformation Research Program.
Deakin said the centre will train a specialised workforce and develop governance, assurance and certification approaches aimed at supporting safer deployment of AI in three priority sectors: healthcare, fintech and defence.
Deakin Distinguished Professor Matthew Clarke, Deputy Vice-Chancellor Research and Innovation, said the centre would build on Deakin’s AI expertise and more than a decade of research in AI assurance.
“At Deakin we are dedicated to exploring the new frontiers of artificial intelligence. Our researchers are hard at work on pioneering technologies that solve real-world industry challenges and benefit society; and it’s critical those two considerations operate in tandem.
“This new centre addresses Australia’s need for trustworthy AI and expertise in governance, assurance, certification and human-AI teaming.”
The ARC Training Centre for AI Governance, Oversight and Assurance will be led by Professor Sunil Gupta from Deakin’s Applied Artificial Intelligence Initiative. Deakin said it will involve 18 industry collaborators and 12 Australian and international research partners, including the University of Wollongong, Swinburne University and the University of Queensland.
Gupta pointed to recent incidents and controversies to argue for stronger oversight of automated systems.
“AI is transforming these critical sectors offering substantial improvements in efficiency, prediction, and decision-making. Yet recent failures expose the risks of inadequate oversight,” Professor Gupta said.
“An OpenAI agent autonomously breached an Australian Government Medicare statistics portal, the Robodebt scheme demonstrated how flawed automated decision-making can generate unlawful debts and cause widespread social harm, and studies have identified racial and skin-tone biases in healthcare AI systems.
“These incidents raise serious concerns about the safety, fairness and accountability of increasingly automated technologies. There’s a growing mismatch between AI’s technical potential and society’s ability to assure that these systems act lawfully and ethically.”
Deakin said expected outcomes include practical frameworks, tools and testbeds intended to support transparent and compliant AI, alongside “safe, secure and privacy-preserving” approaches in critical sectors.
Gupta said skills shortages in governance and compliance are a constraint on domestic capability.
“Only 7% of AI professionals nationally have formal training in risk, ethics, or regulatory compliance. This deficit constrains Australia’s ability to design, audit, and certify AI systems domestically,” Professor Gupta said.
“Australia needs sovereign capability in this space, in part to ensure compliance with Australian laws, as well as alignment with Australia’s social priorities, such as Indigenous data governance and public sector accountability.
“Through the ARC Training Centre for AI Governance, Oversight and Assurance we can build a future workforce equipped with the training and tools to ensure AI is developed in Australia in a safe, ethical and effective way.”
About Author
.png "Agentic AI’s first job in Australian healthcare is fixing the fragmented patient journey")
