As companies large and small carry out layoffs because of AI, there are also jobs being created because of AI.

Along with FDEs (forward deployed engineers) who provide on-the-ground support so companies can properly build AI agents, hiring of cybersecurity staff is also rising as AI spreads.

ㆍThe birth of AI-era SI business... Big Tech's all-out 'FDE' push

A recent New York Times report said demand for cybersecurity talent is so high that some headhunting firms are turning down client requests because they cannot find qualified candidates.

Job search platform Glassdoor said the number of postings for cybersecurity roles rose 11 percent year on year in the first quarter.

The New York Times reported that as developers increasingly generate code using AI, bugs and vulnerabilities are emerging. It also cited growing concerns over cyber attacks that exploit advanced AI models such as Anthropic's Mythos, driving a sharp rise in hiring for cybersecurity experts.

Ria Ketsner, LinkedIn's chief information security officer, said, "Going forward, we will need people to respond to the bug-pocalypse created by AI." She added, "It will take at least a few years to find a solution for how to implement AI security in a sustainable, long-term way."

Ketsner said competition will intensify to secure engineers with technical expertise, flexible thinking to navigate uncertainty and disruption brought by the AI revolution, and an understanding of complex corporate infrastructure.

As AI models become more advanced, demand for cybersecurity personnel is rising further.

In April, AI startup Anthropic partially opened access to a new AI model, 'Mythos', to some institutions and companies. The model is seen as strong at finding and exploiting vulnerabilities. OpenAI also partially released a similar model, 'GPT-5.4-Cyber', to some users.

Michael Piacente, managing partner at Hitch Partners, an executive search consultancy specialising in security leaders, said corporate hiring of security executives has risen sharply since Anthropic introduced Mythos. "Inquiries are flooding in," he said. He added that the increase in inquiries has been fivefold since last autumn and sometimes reaches sevenfold, and that he has had to turn down quite a few hiring requests.

With AI spreading, priorities in cybersecurity also appear to be shifting. Anthropic said rapidly issuing patches is becoming increasingly important, more so than finding vulnerabilities.

In a report summarising one month of results from its Project Glasswing programme, which helps some companies find software vulnerabilities using Mythos, Anthropic stressed: "In the past, software security depended on how quickly new vulnerabilities could be found, but now it depends on how quickly vulnerabilities found in bulk by AI are verified, disclosed and patched."

Anthropic also highlighted that the security industry needs to create a process to manage vulnerabilities discovered at scale by Mythos-level models.

The industry currently follows a practice of disclosing vulnerabilities within 90 days of discovery. Anthropic pointed out that long gaps between vulnerability discovery, patch creation and patch deployment extend the period in which attackers can exploit vulnerabilities. It said Mythos-level models significantly reduce the cost and time of vulnerability discovery and exploit development, making the risks from such gaps even greater.