Australia's AI Impact Assessment Tool — Mandatory Risk Screening for Every Commonwealth AI Use Case

As Australian government agencies scaled up their use of AI, oversight remained fragmented: individual agencies assessed risk against internal frameworks of varying rigour, with no common, mandatory checkpoint aligned to the national AI Ethics Principles.

On 1 December 2025 the Digital Transformation Agency (DTA) published version 2.0 of the Policy for the Responsible Use of AI in Government, taking effect on 15 December 2025 and applying to all non-corporate Commonwealth entities. Its centrepiece is a mandatory AI impact assessment tool: a structured, 12-section questionnaire that agencies must complete for every in-scope AI use case, with a named responsible officer accountable for coordinating each assessment. The tool complements, rather than replaces, agencies' existing risk-management and legal processes, and agencies have until 15 December 2026 to bring all in-scope use cases into compliance.

Before the mandate took effect, the DTA piloted the tool with agencies: participants reported that the structured assessment surfaced fairness, transparency and explainability risks that their existing governance arrangements had not caught. The pilot also found friction -- some agencies said their legal teams would need several weeks to source external advice to complete certain sections -- feedback the DTA says it used to revise the tool before the government-wide rollout.

The evidence here is necessarily early: the tool is barely a year old, the compliance deadline for full government-wide coverage falls at the end of 2026, and unlike the UK's Algorithmic Transparency Recording Standard or Canada's Algorithmic Impact Assessment registry, there is no public register yet of completed assessments, so outside scrutiny of individual use cases is not yet possible.