Australia - AI Policy Update

Overview

The Digital Transformation Agency announced an update to the Australian Government's AI policy framework in December 2025, delivering three interlinked products: an updated Policy for the responsible use of AI in government (v2.0), a new AI Impact Assessment Tool with supporting guidance, and Guidance on AI procurement in government. Policy v2.0 came into effect on 15 December 2025 and establishes mandatory accountability and impact assessment requirements for in-scope AI use cases within non-corporate Commonwealth entities.

Definitions

The update relies on a set of working definitions consistent with DTA resources: an "AI use case" is an activity where an AI system (or model-based automation) materially affects government decision-making, service delivery, or interactions with the public. The policy references Australia’s AI Ethics Principles and distinguishes between in-scope use cases (which require assessments and accountability) and out-of-scope activities (local, low-risk or excluded national security/intelligence contexts). The AI Impact Assessment Tool further defines key concepts (fairness, reliability, contestability, explainability, human oversight) used throughout the assessment template.

Governance and Institutional Framework

The DTA is the lead agency coordinating whole-of-government AI guidance, supported by the Department of Finance and the Australian Public Service Commission (APSC). The updated policy mandates that agencies designate accountable official(s) for AI strategy and for each in-scope use case, maintain an internal use-case register, and produce AI transparency statements under the Standard for AI accountability and Standard for AI transparency statements. The APS AI Plan (2025) coordinates additional institutional measures, including appointing Chief AI Officers, establishing an AI Review Committee, and providing shared GovAI infrastructure to reduce duplication and support secure adoption.

Key Focus Areas

The policy update and supporting materials focus on: (1) strengthened accountability (clear accountable officers for use cases, internal registers); (2) mandatory AI use-case impact assessment, using the DTA’s 12-section assessment tool aligned to Australia’s AI Ethics Principles; (3) procurement safeguards—guidance and checklists that embed AI-specific considerations into the Digital Sourcing Lifecycle; (4) transparency—requirements for publishing AI transparency statements where public-facing or materially impactful AI is used; (5) technical assurance—linkage to the DTA’s Technical Standard for AI and to existing cyber/security frameworks; and (6) capability uplift—training, Chief AI Officers, and GovAI services to support government adoption.

Implementation Framework

Implementation is staged: policy v2.0 is effective from 15 December 2025 but agencies are given structured time to operationalise new requirements (the DTA guidance clarifies implementation timetables and transitional arrangements). The AI Impact Assessment Tool (fillable Word template) and detailed guidance (PDF) are provided to help agencies incorporate assessments into existing governance and investment assurance processes. Procurement guidance maps AI considerations to each phase of the Digital Sourcing Lifecycle (Plan, Source, Manage) and recommends multidisciplinary procurement teams, supplier due diligence, contract clauses, data readiness checks and alignment with existing legislative obligations.

Monitoring and Evaluation

The DTA and Finance will monitor uptake across the APS, supported by the APS AI Plan governance structures. Agencies must maintain internal registers and transparency statements, producing evidence that assessments have been completed and accountability assigned. The DTA has piloted earlier versions of the tool (September–November 2024) and published pilot findings; ongoing monitoring will include aggregation of lessons learned, reuse via a central register on GovAI, and review by the proposed AI Review Committee for higher-risk use cases.

Penalties, Liability, and Appeals

The DTA policy is a mandatory APS policy instrument for non-corporate Commonwealth entities; enforcement relies on APS accountability, administrative remedies and oversight by agency heads, Finance and the APSC rather than new criminal penalties in the policy itself. Non-compliance can have administrative consequences (internal disciplinary or procurement ineligibility), reputational impacts, and may expose agencies to regulatory or legal risks under existing legislation (Privacy Act, protective security laws, procurement rules). The policy highlights alignment with existing legal frameworks rather than creating standalone statutory sanctions.

Relationship to Other Instruments

The policy update complements and references existing instruments: the APS AI Plan 2025, the Technical standard for government’s use of artificial intelligence, the Standard for AI accountability and the Standard for AI transparency statements. Agencies must also continue to satisfy obligations under the Privacy Act, the Protective Security Policy Framework and procurement legislation; the DTA materials are expressly designed to integrate with these frameworks and not to duplicate them.

International Alignment

The DTA materials and the APS AI Plan reference international best practice and standards (for example OECD AI principles and cross-jurisdictional technical guidance). The approach emphasises risk-based, use-case level assessment and alignment with common international themes—transparency, accountability, human oversight and safety—while preserving flexibility to adapt to domestic legal obligations and sector-specific regulatory regimes.

Implementation Timeline

Sources and References