Australia creates national framework for government AI use

A unified approach to AI

Australia's various levels of government are currently working to figure out how to introduce artificial intelligence while simultaneously creating rules to govern it. The process is described as complex because it requires ensuring that different departments and agencies follow clear policies, processes, and procedures. Without a central plan, there is a risk of a disorganized mess in how these technologies are deployed across the country.

To address this, the Data and Digital Ministers Meeting serves as a hub for sharing problems and solutions regarding AI adoption and governance. This group liaises between the Commonwealth, state, and territory governments, as well as New Zealand. It is chaired by Minister for Finance Katy Gallagher and aims to coordinate the digital strategy of the nation.

The meeting first added AI use and regulation to its agenda in June 2023. Since then, the members have worked toward a nationally consistent approach to ensure that the use of AI by governments remains safe and ethical. This effort is driven by the expectation that the public deserves high standards of service delivery, data protection, and privacy.

Building the national framework

In February 2024, the ministers developed a general set of guidelines regarding the assurance of AI use in government. This led to the creation of a national framework for the assurance of artificial intelligence in government by June of that same year. The framework is intended to serve as a foundation for different jurisdictions to build their own aligned governance structures.

The development of this framework utilized AI ethics principles that were originally created by the Department of Industry, Science, and Resources. By using a common ethical foundation, the government hopes that the core principles of AI governance in the public sector will be similar across the country. This shared starting point is meant to provide a baseline for how technology should behave in public service.

However, the government acknowledges that the actual implementation of AI will not be perfectly uniform. Because every jurisdiction has its own unique agencies, they will still use different mandatory policies, risk-management tools, and review committees. While the underlying ethics may be the same, the practical tools for enforcement will vary depending on the specific local requirements.

Federal coordination and hurdles

At the federal level, the Digital Transformation Agency (DTA) is responsible for implementing and coordinating AI efforts. To manage this, every agency or department must appoint a specific owner for their AI projects. This structure allows the DTA to moderate the rollout of technology across the various parts of the federal government.

The DTA faces a difficult task because some departments and agencies have been found non-compliant with basic cybersecurity and information technology guidelines. This was noted by the Australian National Audit Office, which highlights the difficulty of enforcing standards in a large bureaucracy. The agency must work to bring these entities up to speed while simultaneously trying to push new technology forward.

In September 2024, the DTA issued key standards for transparency statements and the appointment of accountable officials. These documents are meant to explain how an agency intends to use AI and who is responsible for it. The DTA has found some entities struggling to finalize these statements or identify the correct officials to hold accountable for the technology.

Reviewing specific use cases

To provide deeper expertise, the federal government established a six-member artificial intelligence review committee. This committee is chaired by DTA chief executive Chris Fechner and includes representatives from various departments, including the Privacy Commissioner and the National Cyber Security Coordinator. The committee met for the first time in June 2024 to approve its terms of reference and begin providing specific advice.

On October 7, the DTA released guidance provided to two specific organizations: the Attorney-General’s Department and Services Australia. The Attorney-General’s Department looked at using AI to help staff with document analysis, matter triage, and workflow optimization. Meanwhile, Services Australia presented a proof of concept to help staff find internal operational documentation more easily.

The committee members welcomed these projects because they focus on improving staff productivity and wellbeing. However, they emphasized the need for strong human oversight, quality assurance, and monitoring. They encouraged these agencies to keep focusing on privacy, workforce capability, and planning for how these systems will scale in the future.