The artificial intelligence job apocalypse has not arrived as expected.
While companies such as Meta, Facebook’s parent company, still garner attention when cutting employees to allow capital expenditures on AI research and development to continue, other enterprises that have invested in and deployed these technologies have slowly begun hiring back workers after making layoffs.
Over the past several weeks, several large enterprises and corporations have backtracked on layoffs after realizing initial AI deployments did not work as previously believed or that human workers are still needed to ensure these technologies and platforms perform as needed. Consider these examples:
- Ford reportedly hired back several hundred engineers to work on quality control issues after AI systems could not fully address them. Charles Poon, Ford’s vice president of vehicle hardware engineering, told the BBC that AI is “only as good as the information you use to train it.”
- IBM, itself a huge booster of AI, announced it would increase entry-level hiring after replacing large portions of its HR staff with virtual chatbots. The company found the technology could not solve some of the more complicated issues that HR employees had handled previously.
- Outside the U.S., the Commonwealth Bank of Australia rehired customer service representatives after virtual chatbots failed to handle heavy customer call volumes.
Although these reports are anecdotal and do not paint a full or broad picture of all hiring and layoff decisions in the U.S. over the last several years, recent research shows that enterprises and other organizations have rehired talent after eliminating jobs following AI deployments.
In a survey of 2,000 hiring managers, staffing firm Robert Half found that about 30 percent of organizations that cut jobs due to AI have had to rehire employees. A Careerminds survey of 600 HR professionals noted that approximately two-thirds of employers that cut jobs because of AI report that they were already rehiring laid-off workers. Of those organizations, 33 percent restored 25 percent to 50 percent of eliminated roles, and 36 percent restored more than half.
The data, combined with the rehiring decisions by IBM, Ford and other enterprises, indicate that AI, while a useful tool, is still not ready to automate all functions within a business and replace workers with virtual chatbots.
The broader lesson from these announcements for cybersecurity professionals is that for the moment, security roles require them to work alongside AI while providing the human judgment, oversight and decision-making these virtual chatbots and other platforms cannot consistently deliver.
Where, then, does this leave cybersecurity professionals who have witnessed AI eliminate some entry-level positions over the last year? Experts point out that while AI is not wholesale eliminating cybersecurity jobs, those professionals who have an understanding of large language models (LLMs) continue to see their value increase as organizations look to integrate AI but with human oversight and decision-making.
“We’re seeing cybersecurity companies make the same AI-driven cuts as the rest of the technology industry, but the hiring data tells a more complicated story. Demand is shifting toward people with deeper security, AI and business judgment, while entry-level roles are being reduced or redesigned,” said Mika Aalto, co-founder and CEO at cybersecurity firm Hoxhunt. “That is unfortunate, because the routine work AI absorbs is also where people build institutional knowledge and become experienced defenders.”
SUBHEAD: How AI Has Changed Cybersecurity, Tech Hiring
While large U.S. technology companies like Oracle, Meta and Amazon continue to shed workers, including developers and cybersecurity professionals, the overall unemployment rate remains about 3 percent for this sector of the workforce. Overall U.S. unemployment is higher at 4.1 percent.
While the numbers for tech and cybersecurity professionals are better than those for other workers, Aalto and other experts note that organizations still benefit from hiring entry-level workers in areas such as security operations centers (SOCs), allowing them to learn about the entire security field and build expertise. In many cases, current AI models are limited in the tasks they can perform.
It’s another reason firms like IBM are reconsidering hiring and training, including for cybersecurity positions, even as more advanced AI models are publicly released.
“Every AI agent we connect to internal systems is effectively a synthetic employee, and if it is mistaken or compromised, it can become an insider threat operating at machine speed,” Aalto told Dice. “AI can absorb tasks, but it cannot own the consequences, so the winning model in cybersecurity is machine speed under clear human judgment and accountability.”
While AI models are good at cybersecurity tasks such as data analysis and anomaly detection, these technologies have difficulty understanding intent, operational context or the physical environment, said John Gallagher, vice president at security firm Viakoo.
What organizations are learning as they invest in AI is that “human-in-the-loop” models, especially in cybersecurity, remain necessary to ensure proper oversight and reduce risks. Eliminating workers can result in less review of how these LLMs are performing their tasks.
“AI models are still prone to hallucinations and overstepping boundaries set for them. All you need to do is look at the recent spate of reports on AI autonomously launching cyber-attacks in order to achieve its objectives,” Gallagher told Dice. “Also look at the case of PocketOS, a company that put AI without human-in-the-loop in control of its software code base. Within a week, it deleted the production database and all backups, leaving PocketOS customers stuck without a solution. When asked about it, the AI agent confessed, saying: ‘I violated every principle I was given.’”
SUBHEAD: Cybersecurity Leadership Needs to Step In
While some enterprises are working to hire back employees displaced by AI, experts note that investments in the technology are still happening. At the same time, those tech and cybersecurity professionals who understand these platforms remain in the best position to thrive and move up the career ladder.
Skilled AI security practitioners are now, and will be, in high demand, with a substantial need for AI guardrails to be implemented in parallel with the adoption of AI in the enterprise, said Diana Kelley, CISO at Noma Security.
AI security and governance are also a top priority for every enterprise CISO today.
“The ubiquitous, rapid adoption of AI in the enterprise is an opportunity for every enterprise CISO to build expertise within the cybersecurity team to help secure AI, especially agentic AI, to instill confidence in AI innovation,” Kelley told Dice. “Net new hires are often seen in areas of AI that require more experience with data science and machine learning. While retraining and upscaling staff for AI tasks benefit from deep sector and business knowledge, such as helping marketing team members learn how to prompt AI to create material that fits the company brand and tone.”
Kelley added that AI security is significantly different from technologies that came before, including IT, cloud and applications, and that training and retaining staff are critical.
“The CISO's team is retraining staff to understand and address novel AI threats and risks that were never a consideration before,” Kelley noted.
