Australia joins new OECD working group on agentic AI in government

Agentic AI has significant potential to reshape how governments operate, deliver services and support public servants. As these systems become capable of planning and acting with greater autonomy, governments are considering how to make use of these capabilities while maintaining appropriate oversight, accountability and control.

Established under the OECD Working Party of Senior Digital Government Officials - known as E-Leaders - the working group brings together digital government leaders from participating countries, including Australia, Canada, Estonia, Japan, Singapore, South Korea, Israel and the United Kingdom. 

Members will share experience and lessons learned across different operating environments, as they explore how agentic AI can be safely used and governed across the public sector.

Navigating new AI capabilities

Unlike traditional AI tools that primarily analyse information or generate content, agentic AI systems can plan, reason and act with greater autonomy to achieve defined outcomes. While people continue to set goals and guardrails, these systems can determine and carry out actions with less direct human involvement.

For governments, this could support more responsive services, help address complex operational needs and reduce administrative burden. However, it also raises important questions about how these systems should be governed as their capabilities develop.

As autonomy increases, so does the importance of effective oversight and control. Governments need to establish clear boundaries and appropriate safeguards for how these systems operate, while ensuring people remain accountable for the outcomes they enable.

“Agentic AI presents governments with new possibilities, but it also raises the bar for how we govern AI. As systems take on greater autonomy, we need to be clear about the boundaries within which they operate and where responsibility ultimately sits,” emphasises Chris Fechner, Chief Executive Officer, DTA.

“Strong governance gives us the confidence to explore and innovate with these capabilities while maintaining meaningful human oversight, clear accountability and the trust people expect from government.”

Bringing Australia’s experience to the table

Australia is continuing to strengthen how AI is governed and used responsibly across government.

In May 2026, the DTA released an Agentic AI addendum to the Technical standard for government’s use of AI. It builds on the existing standard and complements the Policy for the responsible use of AI in government.

The addendum provides additional guidance for agencies designing, deploying and operating agentic AI systems. It recognises that increased autonomy requires appropriate safeguards, with people retaining oversight and control as systems take on more complex tasks.

This work provides a strong foundation for Australia’s contribution to the international discussion, while offering opportunities to learn from emerging approaches in other governments. 

“The decisions governments make now will help shape how agentic AI is used in the years ahead. Working together at this early stage gives us an opportunity to build a stronger collective understanding of what good governance looks like in practice,” continues Mr Fechner.

“Australia has valuable experience to contribute, but no one country has all the answers when it comes to agentic AI. Through the OECD working group, we can draw on different perspectives and help shape a common understanding of how these technologies can be adopted responsibly across government.”

Building a shared path forward

Over the coming months, the working group will compare international experiences and build a common understanding of key public sector concepts. This will inform practical guidance on the foundations and guardrails needed for the trustworthy use of agentic AI in government.

With our international counterparts approaching agentic AI from different starting points, sharing lessons early will help governments draw on emerging approaches, avoid duplicating effort and strengthen their governance arrangements as the technology continues to evolve.

The DTA is committed to continuing this work with the OECD and international digital government leaders, contributing Australia’s experience and supporting the responsible adoption of agentic AI across government.

We’ll continue to share updates and insights here as the working group progresses.

To learn more about the working group, visit Governing with agentic AI: when machines act on government’s behalf - OECD.AI

To learn more about the Australian Government’s suite of AI policies, guidance and frameworks, please visit digital.gov.au.