Managing the ethical risks of artificial intelligence

Tabled Report 2: 2025–26

Tabled date

2025-09-24

Report year

2025-26

Information technology Whole of government

Entities audited

Not identified in the published HTML audit scope.

Recommendations

1. QAO's text Addressed to: enhances its Foundational artificial intelligence risk assessment (FAIRA) and supporting material by clarifying when and how often entities

   enhances its Foundational artificial intelligence risk assessment (FAIRA) and supporting material by

   • clarifying when and how often entities should use the FAIRA across the AI life cycle, including retrospective application
   • developing guidance for alternative risk assessments for lower-risk AI systems
   • clarifying key ethical principles that alternative frameworks should address if entities choose not to use the FAIRA, aligned where possible with national standards or leading practices in other jurisdictions

2. QAO's text Addressed to: enhances its Foundational artificial intelligence risk assessment (FAIRA) and supporting material by clarifying when and how often entities

   supports continuous improvement by assessing the effectiveness of the AI governance policy and supporting tools

3. QAO's text Addressed to: enhances its Foundational artificial intelligence risk assessment (FAIRA) and supporting material by clarifying when and how often entities

   improves its understanding of AI system use and risks across the public sector and develops risk-based advice to support entities in managing higher-risk AI systems.

4. QAO's text Addressed to: the Department of Transport and Main Roads

   enhances its governance arrangements to support responsible use of AI by

   • assessing and updating governance arrangements to ensure they are suitable to manage ethical risks and align with the requirements of the Queensland Government’s AI governance policy
   • implementing appropriate assurance frameworks to ensure its AI governance arrangements are effective at managing ethical risks, meet required standards, and operate within its risk appetite
   • improving visibility of AI systems to strengthen oversight of ethical risks and controls

5. QAO's text Addressed to: the Department of Transport and Main Roads

   improves QChat’s controls to manage ethical risks more effectively by

   • establishing monitoring activities to strengthen oversight and ensure usage is appropriate
   • developing a structured approach to enhance staff capability and promote responsible use of AI systems through training and education.

6. QAO's text Addressed to: the Department of Customer Services, Open Data and Small and Family Business

   supports entities to better manage the risks associated with using generative AI systems, such as QChat, by providing entities with access to content safety information.

7. QAO's text Addressed to: all public sector entities

   implement ethical risk assessment processes for AI systems in use or under development to more comprehensively identify and manage ethical risks.

Authoritative report — Queensland Audit Office

Report PDF on QAO