Macquarie University has launched an AI safety research centre with backing from OpenAI, focusing on alignment, interpretability, and governance of frontier AI models deployed in Australian enterprises.
Tech & Ideas Desk is a contributing writer covering tech and public affairs for The Sydney Times.
Macquarie University has launched an AI safety research centre with backing from OpenAI, focusing on alignment, interpretability, and governance of frontier AI models deployed in Australian enterprises. The centre will receive AUD 15 million in funding over five years, with AUD 10 million from OpenAI and AUD 5 million from the Australian Government's National Reconstruction Fund. The centre is located at Macquarie Park in Sydney's northern suburbs, positioning it near the technology precinct that includes the headquarters of Optus, Microsoft, and other technology companies that are investing in AI research and development.
The research programme includes alignment research that aims to ensure frontier AI models behave in accordance with human values and intentions, interpretability research that aims to make model decision-making transparent and auditable, and governance research that aims to develop policy frameworks and technical tools for responsible AI deployment. The centre will employ 40 researchers, including machine learning engineers, philosophers, lawyers, and social scientists, reflecting the interdisciplinary nature of AI safety research. The interdisciplinary approach is designed to address the technical and social dimensions of AI safety, which cannot be separated in real-world deployment contexts.
The alignment research programme focuses on the problem of ensuring that frontier AI models pursue the objectives that their designers intend, even when the models are operating in complex environments with incomplete information. The problem is acute for enterprise AI deployments, where models are trained on proprietary data and expected to make decisions that align with corporate values and regulatory requirements. The alignment research includes work on reward modelling, constitutional AI, and human-in-the-loop oversight mechanisms that can keep frontier models aligned with intended objectives during deployment.
The interpretability research programme focuses on making model decision-making transparent to regulators, auditors, and enterprise risk managers. Current frontier models are black boxes that produce outputs without explaining the reasoning process that led to those outputs, which is a barrier to regulatory approval and enterprise risk management. The interpretability research includes mechanistic interpretability, which analyses the internal representations of neural networks to understand how they encode concepts and make decisions, and post-hoc explanation methods, which generate natural language explanations of model decisions after the fact. The research is designed to produce tools that enterprises can use to satisfy the transparency requirements of the AI regulation framework.
The centre has established partnerships with Australian enterprises including Commonwealth Bank, Telstra, and Optus, which will provide real-world deployment scenarios for the research and access to data that is representative of Australian business contexts. The partnerships are designed to ensure that the research is grounded in practical problems rather than abstract theoretical questions, and they provide a pathway for commercialising research outcomes through enterprise deployments. The centre will also work with the AI Safety Institute to align its research with the national AI safety evaluation programme, ensuring that research outputs contribute directly to the regulatory framework.
OpenAI's involvement in the centre is part of the company's broader strategy to support AI safety research at universities and research institutions around the world. The company has established similar partnerships with the University of Oxford, the University of California Berkeley, and other leading research institutions, and the Macquarie University centre is its first dedicated partnership in the Asia-Pacific region. The partnership gives OpenAI access to research talent and domain expertise that is specific to the Australian market, and it provides a channel for the company to engage with Australian regulators and enterprises on AI safety issues.
The centre's establishment is part of a broader effort to develop Australian talent in AI safety and alignment, which are specialisations that are in high demand globally but have limited supply in Australia. The centre will offer PhD scholarships and postdoctoral fellowships in AI safety, and it will develop undergraduate and postgraduate coursework that prepares students for careers in AI safety research and enterprise AI governance. The talent development programme is designed to address the skills shortage that is constraining AI adoption in Australian enterprises, and it will create a pipeline of professionals who can work in both research and industry roles.
The centre is located in the Macquarie Park technology precinct, which is home to the headquarters of Optus, Microsoft, and other technology companies that are investing in AI. The location is designed to facilitate collaboration between researchers and industry practitioners, with the centre hosting regular industry workshops and seminars that bring together researchers, enterprise practitioners, and regulators to discuss AI safety challenges and solutions. The collaboration is intended to accelerate the translation of research into practice, and to ensure that Australian enterprises have access to the latest AI safety knowledge and tools. Explore more Australian tech analysis at the Tech & Ideas hub
For Macquarie University AI safety centre details, see Macquarie University. OpenAI's research partnership programme is at OpenAI research. The AI Safety Institute's national programme is published at AI Safety Institute.
Direct inquiries, corrections, or documentation concerning this dispatch to our editorial newsroom desk.
Australian AI startups are facing a funding contraction in 2026 as venture capital firms tighten due diligence requirements, with pre-revenue companies facing the steepest reduction in available capital.
The global AI chip shortage is easing in 2026 as NVIDIA H100 and Blackwell allocations stabilise, but export controls on advanced GPUs to Southeast Asia are redirecting demand toward alternative suppliers and custom silicon.
AI-generated code ownership disputes are emerging in Australian software contracts, as enterprises and vendors negotiate intellectual property rights for code produced by frontier models during outsourced development projects.
AI regulation compliance costs are rising as enterprise legal budgets expand to cover impact assessments, conformity documentation, and ongoing monitoring requirements for high-risk AI deployments.