Communiques
26 June 2026 meeting
The AI Review Committee held its inaugural meeting on 26 June 2026. Established under the Artificial Intelligence Plan for the Australian Public Service, the Committee supports the safe, responsible and practical adoption of artificial intelligence (AI) across the Australian Public Service (APS).
The Committee is designed to consider use cases at different stages of development, from early conceptual thinking about potential capabilities through to testing capabilities in a pilot. Its observations reflect the use case at the stage it was considered. The Committee aims to provide advice at a stage where it can meaningfully inform the agency’s ongoing consultation, refinement and decision-making on the use case’s suitability, design, safeguards and implementation arrangements.
This communique provides a summary of key discussion themes and should not be read as an endorsement or approval of any specific use case.
Members discussed the Committee’s role in providing independent, practical and non binding advice to agencies considering residual high-risk, highly sensitive, novel or complex AI use cases. Members noted that agencies remain responsible for their own decisions, controls and assurance arrangements. The Committee also noted the importance of strong cyber hygiene and organisational readiness as AI capabilities continue to evolve.
The Committee considered its terms of reference and agreed they should be reviewed annually, or earlier by agreement of members, so the Committee remains responsive to emerging technology, risks and operating experience. Members also considered 2 agency use cases.
Attorney-General’s Department
The Attorney-General’s Department discussed the potential to use AI to support staff productivity and wellbeing by assisting with document analysis, matter triage and workflow optimisation.
Members acknowledged the department’s focus on supporting staff wellbeing and improving administrative efficiency in work that may involve confronting material. The Committee encouraged continued consideration of human oversight, quality assurance, legal and transparency implications, staff consultation, staged implementation, monitoring arrangements and security controls. Members also noted the importance of ensuring AI generated summaries support, rather than replace, human judgment.
Services Australia
Services Australia presented a proof of concept using AI to enhance search and virtual assistant capability across its Operational Blueprint, a library of operational instructions for staff. The agency advised that the proof of concept is being tested with a small cohort of experienced users. It is intended to help staff find and apply operational guidance more efficiently within a controlled environment.
Members noted the potential value of the use case, particularly where it improves staff access to complex guidance and supports more consistent service delivery. The Committee encouraged Services Australia to strengthen assurance and implementation planning as the trial continues. This includes supplementing user satisfaction measures with robust quality assurance, clear criteria for scaling, ongoing monitoring of privacy and data guardrails, and consideration of workforce capability, business continuity, accessibility and broader workforce impacts.
The Committee will continue to consider relevant AI use cases and share lessons that support safe, responsible and practical AI adoption across the APS.