Every Australian Cybersecurity Team Hit Problems Rolling Out AI, Securonix Survey Finds

Australian cybersecurity teams are adopting AI faster than their counterparts overseas, yet not one of them has managed a smooth rollout. A survey released on 12 November 2026 found that every Australian respondent ran into implementation problems, even as budgets for AI-driven security kept climbing.

The findings come from Securonix, a security analytics company, and its global report, The Evolution of Cybersecurity Automation: Towards Human-Governed AI Execution. The study surveyed 1,000 cybersecurity professionals in the northern summer of 2026, including 250 from Australia. The rest came from the United States, the United Kingdom and France, and all worked for organisations with between 2,000 and more than 50,000 employees.

The survey in numbers

The headline figure is hard to miss: 100 per cent of the Australian respondents said they had experienced implementation challenges when putting AI to work in security operations. Not most of them. All of them. The report notes that these challenges have stayed broadly stable over the past year, which suggests the problems are structural rather than teething troubles that fade as teams gain experience.

The enthusiasm is just as clear. Among Australian respondents, 98 per cent described cybersecurity automation as important to their organisation, and 67 per cent called it very important. Australian organisations placed greater importance on automation than their counterparts in any other surveyed region. Some 85 per cent of Australian respondents said they were using AI to some degree in their cybersecurity automation.

Spending matches the enthusiasm. Globally, 99.6 per cent of respondents said their budgets for AI-driven cybersecurity automation had increased during the previous year. In Australia, all but two of the 250 respondents reported a budget increase.

The control paradox: hands on the wheel

The report describes what it calls a control paradox at the heart of Australian security operations centres, known as SOCs. Teams want AI to move fast, but they are not ready to let it move alone.

Half of the Australian respondents said key SOC use cases will always require a human in the loop, regardless of how the technology develops. That is a higher proportion than in any other surveyed region. At the same time, 19 per cent of Australian respondents reported fully autonomous AI use in the SOC without human approval, compared with 13 per cent globally. Some 14 per cent said their alert triage operations are already fully autonomous, with no human approval step.

The two figures sit side by side without cancelling each other out. Many teams are letting AI act on its own for routine triage while insisting that a person stays in charge of the decisions that matter. The report frames this as human-governed AI execution: speed where it is safe, human judgment where the stakes are high.

Governance is the weakest link in that picture. Almost a third of Australian respondents, 31 per cent, rated their AI governance as mature and continuously improving, while 41 per cent described it as defined or managed. The report urges caution, noting that AI governance is still an emerging discipline and self-reported scores should be read with care.

Where the rollout stalls

When respondents named their biggest obstacles, three stood out. Management understanding and buy-in topped the list at 35 per cent. Budget constraints were cited by 27 per cent, and the same share pointed to incorrect or undesirable decisions made by AI.

The first figure is telling. The technology is not the main blocker. The blocker is getting managers to understand what AI in security can and cannot do, and to back the changes in process and staffing that a rollout needs. A tool that flags threats faster is of limited use if nobody in the chain of command trusts it or acts on what it finds.

The third figure, wrong decisions by AI, cuts to the trust problem directly. Security teams deal with alerts that can shut down systems, lock out users or trigger incident responses. An AI that makes the wrong call often enough will lose the confidence of the people who are supposed to rely on it, and confidence is slow to rebuild.

The survey covered senior practitioners: chief information security officers, heads of SOC, cyber threat intelligence, incident response and solutions architecture, across financial services, retail, defence, critical infrastructure and healthcare.

Budgets keep rising anyway

The most striking tension in the report is between the frustration and the spending. Every Australian team hit problems, and nearly every one increased its budget anyway. That combination suggests organisations see the problems as worth pushing through rather than as a reason to stop.

It also fits a wider pattern in Australian cybersecurity. Local organisations have been among the earliest adopters of AI-driven security tools, and the survey shows them leading the surveyed regions on automation. The challenge now is less about buying the technology and more about fitting it into organisations that were built around human analysts making judgment calls.

The report includes recommendations for organisations trying to build a human-governed, AI-enabled SOC, with an emphasis on clear rules for when AI can act alone and when a person must approve. For teams already living with the control paradox, that kind of clarity may matter more than the next tool purchase.

What it means for Australian businesses

For businesses outside the survey, the message is practical. AI in cybersecurity is now the norm rather than the exception, and the teams furthest along are still working through governance, trust and management support. A rollout that skips those questions is likely to join the 100 per cent that hit problems. The survey does not suggest slowing down. It suggests spending as much effort on the people and the rules as on the software.

Frequently Asked Questions

What did the Securonix survey find about Australian cybersecurity teams?

The survey of 250 Australian cybersecurity professionals found that 100 per cent had experienced implementation challenges when rolling out AI in security operations, while 98 per cent said cybersecurity automation was important to their organisation.

How many Australian organisations are using AI in cybersecurity?

Some 85 per cent of Australian respondents said they were using AI to some degree in their cybersecurity automation, and 19 per cent reported fully autonomous AI use in their security operations centre without human approval.

What are the biggest challenges in rolling out AI for cybersecurity?

Management understanding and buy-in was the top challenge at 35 per cent, followed by budget constraints at 27 per cent and incorrect or undesirable decisions made by AI, also at 27 per cent.