New tool to assess AI use in government
Australia's Digital Transformation Agency (DTA) has released an AI impact assessment tool to help agencies assess proposed uses of artificial intelligence against the updated AI Ethics Principles (updated 2 December). The tool is built to surface impacts and risks early and strengthen existing governance, without replacing a full risk management plan.
The tool also supports procurement where suppliers use AI to deliver goods or services. It's practical, process-focused, and built to fit alongside your current frameworks, legislation and assurance practices.
Who should complete the assessment
Each agency must appoint an officer to run the assessment end-to-end and act as the contact point. This can be a technical, data, governance or risk specialist, or a policy/project lead implementing AI in operations.
Peer review is expected. The DTA advises assessors to seek input from colleagues and relevant experts before finalising the assessment.
What information you'll need
- Who will be affected by the AI use (individuals and groups)
- Potential impacts on those people, including risks and benefits
- The input data the system uses (sources, quality, bias considerations)
- Planned or existing security, monitoring and evaluation measures
- Planned or existing transparency measures
- How decisions are recorded, logged and explained
- The extent of human oversight and how outcomes influence human decisions
Policy updates you need to act on
The government's AI policy was updated and took effect on 15 December. It strengthens requirements for safe, responsible AI use and makes AI impact assessments mandatory.
Agencies must also meet mandatory requirements for:
- Accountable officials
- Transparency statements
- Developing a strategic approach to AI adoption
- Operationalising the responsible use of AI
- AI use case accountability
- Internal use case registers
- Staff training
What this means for your agency
If you're planning, piloting or expanding an AI use case, the assessment becomes a default checkpoint. It will help you clarify decision rights, human oversight, data sources, testing plans and public transparency before you deploy.
For procurement, build the assessment into your sourcing process. Ask suppliers to provide the evidence you need to complete it, including model behavior, data lineage, security controls and explainability.
Practical next steps
- Nominate your assessment officer and set up a lightweight review panel (legal, risk, data, security, program).
- Inventory current and planned AI use cases and prioritise higher-risk services first.
- Embed the assessment into project gates and procurement templates.
- Publish or prepare transparency statements for public-facing uses.
- Plan staff training linked to the policy's mandatory requirements.
- Report any gaps or pain points in the tool or policy to the DTA so they can iterate.
Where to find the core references
Read Australia's AI Ethics Principles and guidance: industry.gov.au
Visit the Digital Transformation Agency for policy and tool updates: dta.gov.au
Building capability
Staff training is now a mandatory element. If you need structured options by role or skill, explore curated AI learning paths: Complete AI Training - Courses by Job
The DTA wants feedback to keep the policy and tool fit for purpose as new applications and risks emerge. If you spot gaps, raise them early-your input will improve the whole system.
Your membership also unlocks:
