Australian Government expands AI adoption across public services with new framework

The Australian Government is expanding AI adoption across public services with a new framework that establishes governance requirements, procurement guidelines, and ethical principles for AI use in government agencies. The framework, published by the Digital Transformation Agency in July 2026, applies to all federal agencies that are developing, procuring, or operating AI systems, and it establishes a risk-tiered approach to AI governance that aligns with the AI regulation framework being developed by the Department of Industry Australia. The framework is designed to enable responsible AI adoption while protecting the rights and interests of citizens who interact with government AI systems.

The framework categorises government AI systems into four risk tiers: minimal risk, low risk, medium risk, and high risk. Minimal risk systems, including spam filters and productivity tools, are exempt from most governance requirements. Low-risk systems, including internal workflow automation and document classification, require basic documentation and human oversight. Medium-risk systems, including customer service chatbots and benefits eligibility assessment, require impact assessments and transparency reporting. High-risk systems, including law enforcement decision support and immigration risk assessment, require full conformity assessment, independent audit, and ongoing monitoring.

Procurement guidelines and vendor selection

The framework includes procurement guidelines that require agencies to conduct due diligence on AI vendors before entering into contracts for AI system supply. The due diligence includes assessment of the vendor's AI safety practices, data handling policies, and compliance with applicable standards and regulations. The guidelines require agencies to include specific AI provisions in contracts, including representations and warranties about model behaviour, data residency guarantees, and audit rights that allow the agency to inspect the vendor's AI systems and data processing practices.

The procurement guidelines also address the use of open source AI models and fine-tuned models that agencies develop internally, requiring the same governance controls to be applied regardless of the source of the model. The requirement is designed to ensure that agencies do not circumvent governance requirements by using open source models or in-house development, and it establishes a level playing field between proprietary and open source AI procurement. The guidelines are consistent with the government's open source procurement policy, which encourages agencies to consider open source alternatives while maintaining appropriate governance standards.

Ethical principles and human rights considerations

The framework includes ethical principles that require government AI systems to be lawful, fair, transparent, accountable, and respectful of human rights. The principles are designed to ensure that AI systems do not discriminate against individuals on the basis of protected attributes including age, disability, race, and gender, and that individuals have the right to understand and challenge AI decisions that affect them. The principles are aligned with the Australian Human Rights Commission's guidance on AI and human rights, and they provide a basis for agencies to assess the ethical implications of AI systems before deployment.

The human rights considerations include requirements for agencies to conduct human rights impact assessments for high-risk AI systems, and to consult with affected communities before deploying systems that have significant impacts on individual rights. The consultation requirement is designed to ensure that AI systems are developed with input from the people who will be affected by them, rather than being designed solely by technical teams without community engagement. The requirement is a significant shift from previous government AI procurement practices, and it reflects the growing recognition that AI systems can have profound effects on individual rights that must be considered in the design and deployment process.

Services Australia and agency implementation

Services Australia is the lead agency for AI adoption in the public services sector, with active AI deployments in customer service, fraud detection, and document processing. The agency is using AI to automate the processing of government payment claims, with the AI system extracting relevant information from application forms and supporting documents and routing the claims to the appropriate processing team. The automation has reduced processing time for routine claims from days to hours, and it has freed up human caseworkers to focus on complex claims that require judgment and empathy.

The agency's AI deployments are subject to the new governance framework, and Services Australia has established an AI governance committee that oversees the development and deployment of AI systems across the agency. The committee includes representatives from legal, privacy, ethics, and operational teams, and it is responsible for ensuring that AI systems comply with the framework requirements and the agency's own AI ethics policy. The governance structure is designed to ensure that AI adoption is accountable to both internal stakeholders and external regulators, and it provides a model for other agencies that are expanding their AI capabilities. Explore more government technology analysis at The City hub

For the Digital Transformation Agency's AI framework, see DTA AI. The AI Safety Institute's government programme is at AI Safety Institute. Services Australia's AI strategy is published at Services Australia.