What happened
Australia's Digital Transformation Agency (DTA) mandatory AI requirements took effect on June 15, 2026 for all Commonwealth non-corporate entities. A remaining tranche of DTA requirements is scheduled for December 2026. Separately, the DTA published an 'Agentic AI Addendum' — practical policy guidance for deploying agentic AI systems in government — defining three oversight models: human-in-the-loop, human-on-the-loop, and human-out-of-the-loop, with corresponding governance requirements including kill-switch obligations. The DTA also announced an AI Review Committee to provide expert non-binding advice on high-risk AI use cases before deployment.
Why it matters
Australia's June 15 mandatory requirements represent the first binding tranche of the DTA's whole-of-government AI policy framework for Commonwealth agencies. The Agentic AI Addendum is among the most operationally specific government guidance published anywhere on autonomous AI agent deployment, setting concrete oversight models and kill-switch requirements that will influence both Australian government procurement and private-sector best practice.
Action needed
Commonwealth agencies must confirm compliance with the June 15 tranche now and prepare for the December 2026 tranche. Government AI vendors should align product offerings to the Agentic AI Addendum's three oversight models and kill-switch requirements. Private sector deployers of agentic AI should treat the Addendum as a best-practice benchmark.