Queensland Health Artificial Intelligence Framework G2: Where governance sits and who holds the roles: low or medium risk solutions used in a single Hospital and Health Service within its risk appetite may be governed there, while high risk or potentially statewide projects need eHealth and departmental advice; clinical AI follows rigorous clinical governance pairing technical and clinical expertise plus Therapeutic Goods Administration requirements; consumer representation is considered across the lifecycle and in governance, design and testing; and human oversight is secured through the AI initiative owner, coordinating team, executive sponsor, decision maker with financial delegation, system owner and a statewide AI Oversight Committee that refines the Framework, develops the risk tool, enforces the principles, assesses risks, approves and monitors projects and ensures legal compliance

What does G2 require?

Governance is scaled to the size, scale and risk of the solution. Where a solution is low or medium risk, will be used only within a single Hospital and Health Service and sits within that service's risk appetite, governance may rest with the service alone; high risk projects and projects that may scale across several services or statewide require advice from eHealth and other departmental units, with open communication about risk mitigation. Because inaccurate or misleading AI in clinical care can harm patients, clinical governance must follow a rigorous process, pairing technical with clinical expertise according to the complexity and risk of the clinical application, and Therapeutic Goods Administration safety regulation applies and is checked for the latest advice. Queenslanders are both excited and nervous about AI, so health consumer representation is considered in every initiative across the whole lifecycle, as stakeholders and potentially in project governance and in the design and testing of applications, consistent with the Clinical Senate's recommendation of a community advisory function and its principles of careful, safe, supported adoption that engages the community and reflects equality, diversity, inclusion, human rights and trust. Human oversight runs through defined roles: the AI initiative owner responsible for the initiative through the stages; the AI coordinating team or unit that helps it through the gates and coordinates governance and approvals; the executive sponsor or lead who reviews each initiative for strategic alignment, principles, prioritisation, affordability and formal support; the decision maker who reviews and approves at each gate and must hold the financial delegation to identify funding; and the system owner who operates and maintains the statewide technical and human resources that make AI solutions available and monitor them. A departmental or system-wide AI Oversight Committee refines and updates the Framework, develops the Queensland Health AI risk assessment tool, enforces the guiding principles, conducts regular risk assessments of AI systems and projects, reviews and approves projects, monitors AI performance and outcomes, ensures compliance with laws, regulations and standards, facilitates knowledge sharing, provides guidance, builds public trust and represents Queensland Health externally.