AI is killing off entry-level cybersecurity jobs | Information Age

Many Australian companies have replaced entry-level cybersecurity staff with AI, according to Barracuda Research. Image: Shutterstock

AI-based cybersecurity technologies have improved so much that 45 per cent of Australian firms have replaced entry-level security workers with AI, according to new research that highlights cybersecurity’s fluid job situation at a time when threats are rapidly expanding.

Fully 64 per cent of the 200 Australian respondents to Barracuda Research’s latest Cybersecurity Skills Gap Report, which surveyed 2,000 IT and security decision-makers in 16 countries, said they had seen reductions in their professional workforce linked to AI.

Those cuts come as AI dramatically reshapes the way cybersecurity firms handle security incidents – with Arctic Wolf, for one, claiming that AI agents now resolve 60 per cent of its customers’ security investigations without human intervention.

Just 2 per cent of the company’s cybersecurity investigations are now handled exclusively by humans, with the remainder handled in a human-AI collaboration that has surged from 1 per cent of incidents last year to 38 per cent this year.

While job cuts affected entry-level and executive jobs roughly evenly at a global scale, the Barracuda figures showed that among Australian companies entry-level jobs were affected more often (45 per cent of respondents) than senior roles (33 per cent).

This suggests Australian firms are far more mercenary than their global peers when it comes to replacing workers with AI – but for all its seeming cost and efficiency benefits, the figures confirmed there’s a sting in the tail for overeager AI adopters.

With human decision-making still an important part of responding to cybersecurity incidents, fully 68 per cent admitted they now lack the skills to effectively deal with AI-related cybersecurity risks.

“AI is changing the cybersecurity landscape faster than many security teams can build the necessary skills, visibility, and governance,” Barracuda senior vice president and CIO Siroui Mushegian said in releasing the results and arguing for a hybrid human-AI model.

Cybersecurity has become almost entirely AI-led – on both sides

Concerns that an overreliance on AI may compromise the human role in cybersecurity defence come as the Australian Signals Directorate (ASD) releases its latest Annual Cyber Threat Report amidst warnings that AI has turned cybersecurity practice on its head.

“AI has been the defining force in the cyber threat landscape over the past year,” Minister for Defence Richard Marles said in introducing the report, “allowing malicious actors to find vulnerabilities faster than exploit them at unprecedented scale, speed and sophistication.”

The ASD assisted in over 1,500 cybersecurity incident responses during 2025-26, the report revealed – up 20 per cent over the previous year – received over 82,000 reports, equal to one every six minutes.

It notified businesses of potentially damaging malicious cyber activity over 4,400 times – up 80 per cent over the previous year – with firms nonetheless reporting average losses of $93,500 per incident, up 16 per cent year on year.

Over the next 12 months, the report warns, growing use of AI-driven tools “will further compress this window: In the past, defenders had months to patch [but] now they have days, and sometimes only hours.”

Breaches getting bigger as cyber teams get smaller

The consequences of this shift were clear during the first half of this year when, newly released Office of the Australian Information Commissioner (OAIC) statistics reveal, it received 667 notifiable data breach (NDB) notifications.

While this was nearly equal to the 662 notifications during the previous half-year reporting period, this year’s breaches have a much larger blast radius – with the number of breaches affecting at least 1 million people jumping to 11, from just 4 in the previous period.

This surge comes as cybercriminals refined and improved the offensive capabilities of AI platforms and AI agents, which have proven increasingly adept at circumventing access controls to compromise government and sensitive targets with impunity.

Given that government bodies have been exposed as being hopelessly slow at reporting breaches, AI’s value as a defensive tool is increasing as quickly as its popularity amongst cybercriminals – yet without the staff to use it, many businesses will struggle.