Our approach to GenAI at UWA 

Generative AI represents a transformative technology that can enhance the research, teaching and service delivery we provide to our university community. We believe that UWA should adopt AI tools mindfully, with particular focus on preserving academic integrity, protecting intellectual property, maintaining data security and privacy, and supporting human creativity for both staff and students.

While we remain excited by the unprecedented possibilities of GenAI, we recognise that rapid advancements generate uncertainty. Our guiding principle is to prioritise optionality – to ensure the University and our staff maintain as much flexibility as possible. To safeguard the University’s core values, we have recommended a strengthening of existing processes so that AI-augmented outcomes remain grounded in human-driven expertise (often called “human in the loop”).

A key element of our approach is to leverage established frameworks – both within and beyond UWA – to inform the necessary guardrails for the technology. Rather than building new mechanisms from scratch, we see value in refining policies to accommodate AI’s emerging capabilities while protecting core academic standards. For example, where TEQSA standards govern higher education, we believe their scope already includes maintaining high academic standards in an AI-influenced environment (as opposed to new regulatory approaches).

Equally crucial is our commitment to inclusiveness, transparency and respect for diverse perspectives. Whether it’s safeguarding sensitive data or ensuring that AI-generated content reflects the communities it serves, we believe the adoption of AI tools make it critical that all members of our community are responsible agents of data and information. For this reason, we advocate for a tiered approach to AI tools with different levels of security and data loss prevention. We continue to highlight that AI tools must not compromise personal privacy, data security, intellectual property or cultural integrity, but rather are the right tools for the job based on the classification of data they use with the tool.