For agency leaders, the challenge is clear: how can they harness AI quickly and cost-effectively while managing the risks that come with it?
For public sector leaders, the expectation is constant: deliver better services for less. Emerging technologies are frequently held up as the magic solution to achieving both.
Yet new technologies rarely live up to their hype - at least not immediately.
No new technology has received more hype than artificial intelligence (AI). Yet despite promising a revolution, clear, high-value use cases remain scarce, with much of the discussion focused on AI's related risks. These include its tendencies to introduce bias, leak data, hallucinate, worry the workforce, introduce operational complexity, and come at a high financial and environmental cost.
This challenge was highlighted in February 2025 by the Digital Transformation Agency's chief executive officer, Chris Fechner, who in an interview with The Mandarin stated: "The challenge ... is how safely, quickly, and affordably we can change our workforce, processes, systems, and governance to ensure that we realise the intended benefits of using AI while avoiding the downside of unintended outcomes ..."
According to Andrew Horton, managing director of the Horton Group and director of the public sector think tank the Australian Strategic Policy Institute, the rapid adoption of AI by the Australian public sector is not an optional exercise. It is an economic and strategic imperative - but one that warrants careful consideration.
"As agencies rush to leverage powerful new generative AI tools, the foundational question must shift," Horton says. "Who truly controls the intelligence driving our government?"
The requirement to improve government service delivery is unavoidable, however and was highlighted in the 2024 policy for the responsible use of AI in government, which described the benefits from adopting AI as including more efficient and accurate agency operations, better data analysis and evidence-based decisions, and improved service delivery for people and business.
Put simply, AI is too significant to ignore - even if it is not a silver bullet.
While Gartner estimated that federal agencies would spend $11.7 billion on IT and digital transformation in 2024, no figures exist to describe spending on AI specifically. But by all estimates that number is growing swiftly.
For agency leaders, the challenge is clear: how can they harness AI quickly and cost-effectively while managing the risks that come with it?
According to Nick Sone, chief customer officer at leading Australian systems integration Brennan, the company's three decades of experience in transformation programs has led it to identify several critical common factors which are equally applicable in AI implementation.
The first of these is the need to take a pragmatic approach to any technology deployment - including AI.
"While many organisations begin with the question "where can AI benefit us?", you should start by examining the underlying process itself," Sone said. "Only once opportunities for improvement are clearly defined can appropriate technology recommendations be made."
Brennan refers to this approach as Operational Innovation - a methodology focused on measurable process improvement rather than transformation for its own sake.
Sone says the second factor is the importance of real-world implementation experience. For Brennan, this means multiple AI workloads in production - not pilots - both internally within the organisation and for customers, including AI engagements with public sector agencies, including a port authority and a utility ombudsman.
The third factor is the need to work with suppliers who understand and can meet the security, privacy, and governance requirements of AI projects. According to Sone, this is best achieved through partnerships with Australian technology providers who have a deep understanding of local security considerations and regulatory obligations.
"AI isn't an independent state, answerable only to itself. It still needs to conform to the environment in which it's deployed," Sone says. "This feeds into the notion of Australia's digital sovereignty - of data integrity, homegrown infrastructure, a deep familiarity with governance and guardrails, all underpinned by skilled Australian knowledge workers to protect and manage it.",.
The importance of sovereign capability has come into sharp focus in recent months thanks to a shifting geopolitical landscape and the growing desire for public and private organisations to keep their data assets fully under their own control.
According to Horton, Australia's reliance on extraterritorial technology giants for essential services like the cloud has already introduced structural risks.
"When we add AI - the digital bloodstream of the modern nation-state - to this infrastructure, we risk a fatal loss of strategic autonomy," Horton says. "For the public sector, sovereignty is not a preference for local content; it is the precondition for secure innovation."
Horton notes that even when data is stored in an Australian facility, foreign ownership of that centre can still expose it to extra-territorial legal orders. This risk could be mitigated by using local expertise and infrastructure.
"Working with local Australian entities ensures that accountability terminates in Canberra," Horton says. "These partners are driven by the specific legal and compliance obligations of the Australian context, rather than the AI marketing hyperbole coming in from US headquarters."
This distinction is critical to ensuring operational control, rather than creating dependencies on foreign-controlled entities.
According to Sone, working with local providers ensures the development of local skills, whose value will increase as AI utilisation grew.
"Investing in Australian skills and IP, in sovereign technology - it grows local talent, drives innovation, supports jobs, keeps money onshore, and builds long-term resilience across the economy." Sone says. "Across nearly every metric worth measuring, it delivers tangible dividends to Australia," says Sone.
In a decade defined by the need for improved service delivery and greater efficiency, AI has emerged as a powerful tool, but one that requires specialised skill in its handling if those benefits are to be delivered without the associated risks.
Brennan's local experience helps public sector agencies make AI a trusted national asset, not an insecure and hype-fuelled fever dream.