Angus Loten

More companies are turning to artificial intelligence to battle AI-powered hackers, which is creating demand for more seasoned workers as entry-level job seekers are left in the cold.

On balance, jobs in cybersecurity are holding steady. Nearly three-quarters of corporate chief information security officers said they didn't expect to reduce the size of their teams as a result of AI adoption, according to a survey by cyber-risk consulting firm IANS Research and executive recruiter Artico Search. Instead, over 80% said AI is more likely to create demand for new cyber roles and skills.

The survey, carried out between April and August, gathered responses from more than 500 CISOs in a range of industries, mostly at U.S.-based firms.

But the numbers reflect a refocusing of priorities, as more junior positions are replaced by automation-even as the overall number of jobs remains steady. Many employers are now looking for more experienced workers with higher-level skills to manage advanced AI tools.

AI is increasingly absorbing repetitive, lower-level tasks such as alert triage, log review and data enrichment. But "there is a middle tier that is reviewing AI outputs and making judgment calls that AI hasn't yet been able to replace," said Steve Martano, an IANS faculty member and a partner at Artico Search.

That means those with the chops for only operational, hands-on-keyboard jobs are in less demand these days. Instead, he said, companies are filling analyst-level positions with a stronger focus on business acumen, sound judgment and other intangible skills.

"To use AI tools effectively for cybersecurity, we need people to operate and oversee them," said Matt Fredrikson, a professor at Carnegie Mellon and chief executive and co-founder of security firm Gray Swan AI. As hackers roll out dangerous AI models, companies should "invest more in security-not make cuts," he said.

Job-market data appear to support Fredrikson's position. Last month, U.S. employers posted 13,923 cybersecurity jobs online, an uptick from 13,169 in July, according to information-technology trade group CompTIA. The gains come despite economic uncertainties and flat cybersecurity budgets.

"Many organizations are slowing hiring while they wait to see how AI agents will actually perform," said Robb Reck, chief information, trust and security officer at Pax8, a small-business cloud-computing marketplace. Rather than expanding cyber teams, he said, employers are looking to AI to augment their current workers' productivity.

Dov Yoran, chief executive of AI cyber firm Command Zero, said that unlike most white-collar jobs-where jobs are cut when automation takes over-cybersecurity threats aren't fixed and constantly require new software and platforms.

With AI and other tools, Yoran said, teams that could triage a few hundred daily security alerts can now process a few thousand. That frees up experienced cyber workers to investigate more serious alerts that require professional judgment, he said: "The work doesn't disappear; it moves up the stack."

But the loss of entry-level jobs has other industry watchers worried. Diana Kelley, CISO at AI security and governance platform Noma Security, said the emphasis on higher-expertise jobs could hollow out the pipeline of young up-and-coming professionals.

"Cybersecurity has always built senior talent through real-world experience, mentoring, labs, adjacent IT roles and on-the-job development," Kelley said. "If we narrow those pathways too much, we won't solve the talent gap. We'll turn it into a talent chasm."

Junior candidates, she said, need to show hands-on capability earlier, including cloud and identity basics, AI fluency and the ability to validate automated outputs instead of simply trusting them.

All cyber professionals need to understand how AI models behave in production environments-especially under adversarial conditions, said Sumedh Thakar, president and chief executive at Qualys, a cloud-based IT, security and compliance firm.

"The human-in-the-loop approach to AI is here to stay, and that will separate those with the expertise to guide, shape and govern AI from those who will be replaced by it," he said.