Australia AI adoption outpaces governance, says study

Insight Enterprises has published research showing Australian businesses are adopting artificial intelligence faster than their governance frameworks are evolving. The findings point to a gap between AI use and organisational controls.

The study surveyed 318 business decision-makers in Australia and 220 in Singapore. It found that most Australian organisations remain at an early stage of AI adoption despite broader operational use.

Only 21% of Australian organisations said they were scaling AI across functions, while 42% were still experimenting and 30% remained in pilot programmes. Just 8% said AI was fully embedded in operations.

Singaporean organisations reported higher maturity. There, 37% said they were scaling AI across functions and 14% said AI was fully embedded in operations.

The data reflects what Insight called an "autonomy paradox", in which companies give AI systems more decision-making responsibility while governance, trust and control structures remain underdeveloped.

Nearly half of Australian leaders, or 47%, described their organisations as being in the early exploration stage for autonomous AI, while only 10% considered themselves highly prepared. In Singapore, almost 40% of leaders said they were moderately prepared and 20% said they were highly prepared for autonomous AI deployment.

The pressure appears strongest at senior levels. Half of Australian founders, Chief Executive Officers and C-suite leaders said they were prepared to delegate more to AI than their current governance frameworks allow, compared with 39% of leaders in the layer below.

Mike Morgan, Senior Vice President & Managing Director, APAC, at Insight, said the results showed a widening disconnect between implementation and oversight.

"What we're seeing is an autonomy paradox emerging across Australian organisations. Businesses are increasing their reliance on AI faster than they are developing the governance, trust and control capabilities needed to support it," Morgan said.

He said governance needed to be developed alongside deployment, not after it.

"Having partnered with multiple clients across their AI journey, we see first-hand that governance cannot follow implementation; it has to be built alongside it," Morgan said.

Scaling barriers

The research also identified practical obstacles slowing broader rollout. Integration with legacy systems was the biggest barrier to scaling AI, underscoring the difficulty many organisations face in fitting newer tools into older technology estates.

Data readiness was another weak point. More than half of Australian organisations, or 53%, said their data was only somewhat ready for AI, while another 20% said it was not ready at all.

That matters because data quality and accessibility are central to whether AI systems can be used consistently across functions rather than in isolated trials. The figures suggest many companies are trying to expand AI use before key operational foundations are in place.

Skills shortages are also weighing on progress, especially in the mid-market. Among mid-sized organisations, 24% of leaders said talent shortages were a barrier, compared with 9% of enterprise organisations.

The gap points to a resource divide between larger companies and smaller peers. Bigger organisations are often better placed to recruit specialist staff or absorb the cost of external support, while mid-market firms may find it harder to build teams that can manage deployment, oversight and integration at the same time.

Regional gap

The comparison with Singapore adds to concerns about Australia's pace of execution. While both markets are investing in AI, the survey suggests Singaporean businesses are moving more quickly from experimentation to scaled and embedded use.

The difference is not only about adoption levels but also about readiness for more autonomous systems. The higher preparedness reported by Singaporean respondents indicates stronger alignment between ambition and operating structures.

Morgan said many organisations were not waiting for full readiness before expanding AI decision-making.

"Organisations are not waiting until they are fully ready to delegate to AI; they are doing it now, under pressure to deliver outcomes," Morgan said.

He said that tension sits at the heart of the findings.

"That's what makes this a paradox: autonomy is increasing, but confidence, governance and preparedness are still catching up," Morgan said.