QChat — Queensland's Government Generative-AI Assistant, 19,000 Users In and an Audit Warning on Missing Safeguards

QChat is the Queensland Government's in-house generative-AI assistant for public servants, run by the Department of Customer Services, Open Data and Small and Family Business (CDSB). It was piloted with selected departments from late 2023 and rolled out statewide around mid-2024; iTnews reported over 14,000 users across 18 agencies in May 2025, and the model was upgraded to Azure OpenAI GPT-4o in April 2025.
In its report tabled on 24 September 2025 (Report 2: 2025–26), the Queensland Audit Office recorded 383,000 QChat conversations and 18,950 active users, of which 72,000 conversations and 3,020 users were in the Department of Transport and Main Roads (TMR). System-level safeguards (safety prompts, contextual responses, entity prompts) are managed centrally by CDSB.
The audit was critical of governance below that level: TMR had not completed an ethical risk assessment for QChat, had no monitoring controls, and staff training uptake was low. CDSB lacked whole-of-government visibility of AI use. The auditor made seven recommendations covering framework guidance, monitoring, training and risk assessment.
At The Mandarin's conference on 27 May 2026 the Premier acknowledged QChat had not become popular and that fewer than one in ten public servants used it, saying alternatives would be considered. The government has said an evaluation programme is under way, but no impact results (e.g. time saved) have been published.
The practice is therefore included as a cautionary, partly evidenced case: a secure, centrally governed platform reached tens of thousands of users, but measured public value is unproven, entity-level oversight lagged, and adoption stayed narrow.

Read the full analysis: https://www.qao.qld.gov.au/reports-resources/reports-parliament/managing-ethical-risks-artificial-intelligence