Governing artificial intelligence in Australian healthcare organisations - a qualitative study of staff perspectives.

OBJECTIVE

This study aims to examine how healthcare staff conceptualise artificial intelligence (AI) governance and to develop an understanding of organisational requirements to support safe and responsible AI use within health services.

METHODS

This qualitative study was conducted within a major Australian healthcare organisation delivering care across three hospital campuses and community-based services. Twenty-three semi-structured interviews were undertaken with staff and representatives across strategic leadership, clinical implementation, research and evaluation, governance and regulation, operational delivery, legal oversight, and consumer engagement. Interviews lasted between 30 and 70 min. A grounded theory approach guided data collection and analysis. Two investigators independently coded transcripts, with codes and themes developed iteratively and refined through research team review and consensus.

RESULTS

Participants characterised AI governance as a multifaceted and ongoing organisational function rather than a single approval step. Five interrelated domains were identified as critical to effective operational AI governance within healthcare organisations: risk assessment, infrastructure and data, ethics, capability and culture, and governance structures. Participants emphasised the need for AI-specific risk assessment pathways, lifecycle monitoring, stronger data governance, clearer approaches to consent, fairness and transparency, greater internal capability and system navigation support, and dedicated governance structures with defined oversight responsibilities. Existing processes were commonly described as fragmented, ad hoc, and insufficiently tailored to the ethical, technical, and operational challenges posed by AI.

CONCLUSIONS

Healthcare staff view AI governance as an organisation wide function requiring clear pathways, defined accountabilities, and sustained oversight. The five domains identified provide a foundation for Australian health services seeking to operationalise safe and responsible AI.