Australia joins OECD group to tackle agentic AI in government

Australia has joined a new OECD Working Group on Agentic AI in Government, as public sector agencies start grappling with AI systems that do more than generate text or analyse data.

GovTech Review reported on 29 September 2026 that the Digital Transformation Agency had announced Australia’s involvement in the group. The OECD’s own material describes the working group as a forum for public officials to share experience and shape practical guidance on using agentic AI in public processes, policies, services and administration while protecting accountability and public trust.

This is not just another AI committee with a shinier name.

Agentic AI refers to systems that can plan, reason and act with greater autonomy to achieve defined outcomes. In plain English, the software is not only answering a question. It may be deciding the steps needed to get something done, then carrying them out inside boundaries set by people.

That makes it potentially useful for governments, which run high-volume services and complex back-office processes. GovTech Review’s account said the DTA sees potential for more responsive services, help with operational needs and reduced administrative burden.

But the same shift also raises the stakes. If an AI system acts on behalf of a public agency, governments need to know where the guardrails are, how oversight works and who is responsible when something goes wrong.

The OECD.AI article, published on 25 September 2026 by the OECD Working Group on Agentic AI in Government, frames the issue in similar terms. It says AI is moving from tools that answer questions and support decisions to systems that act to pursue goals with higher levels of autonomy. It also notes that governments provide access to benefits, licences, emergency services and justice, areas where people expect speed but also fairness and reliability.

According to GovTech Review, the working group was established under the OECD Working Party of Senior Digital Government Officials. Participating countries listed in that report include Australia, Canada, Estonia, Japan, Singapore, South Korea, Israel and the United Kingdom.

The OECD Observatory of Public Sector Innovation page lists Ramsey Beydoun, AI Branch Manager at the DTA, as the Australian working group member. The OECD page says Beydoun leads AI governance, technical standards and strategic engagement to enable responsible adoption of AI across the Australian Government.

GovTech Review also reported comments from DTA chief executive Chris Fechner, who said agentic AI presents new possibilities for governments but raises the bar for AI governance. He said governments need clarity on system boundaries and where responsibility ultimately sits.

The report said the DTA released an Agentic AI addendum to the Technical standard for government’s use of AI in May 2026. That addendum builds on the existing technical standard and complements the Policy for the responsible use of AI in government, according to the GovTech Review account.

The move puts Australia inside an early international effort to define what good governance looks like before agentic AI becomes routine in public administration. The technology may be new, but the public sector problem is old: move faster, without losing control.