Cybercriminals Struggling to Adopt AI in Their Work, Research Suggests

What the findings show

New research suggests that cybercriminals are experimenting with artificial intelligence, but most are not yet turning it into a major operational advantage. After examining around 100 million posts from underground and dark web cybercrime communities, researchers concluded that many offenders still lack the technical skill, knowledge, or resources needed to use AI effectively in their criminal work.

The work was conducted by researchers from the University of Edinburgh, the University of Strathclyde, and the University of Cambridge, who analyzed discussions in the CrimeBB database using a mix of machine-learning tools and manual review. They focused on conversations from November 2022 onward, a period chosen to capture how criminal communities reacted after the release of ChatGPT and the wider spread of generative AI tools.

Why AI has not transformed cybercrime overnight

A central conclusion is that AI has not automatically lowered the barrier to entry for cybercrime. Coding assistants and related tools appear to be most useful for people who already have substantial expertise, rather than newcomers with little background. In other words, AI may amplify existing skills, but it does not seem to have eliminated the need for them.

Researchers found more successful criminal uses in narrower areas, including social media bots used for misogynistic harassment, fraud, and techniques that help hide suspicious patterns from cybersecurity defenders. At the same time, they said the guardrails built into major chatbots appear to be having a meaningful effect in limiting some harmful uses.

The bigger warning for industry

The study’s warning is not that criminals have already mastered AI, but that legitimate organizations may create new vulnerabilities by deploying insecure systems of their own. According to the researchers, the more immediate threat comes from businesses and the public adopting poorly secured AI tools that can open the door to damaging attacks with relatively little effort from attackers. They also highlighted concerns about agentic AI systems that can act autonomously, as well as insecure AI-generated code in commercial products.

Dr. Ben Collier of the University of Edinburgh summed up the current moment with a short note of caution rather than panic:

"Don't panic yet."

Employment pressure inside cybercrime communities

Another notable finding is that some people in cybercrime forums appear worried that AI could threaten their conventional IT jobs. That anxiety matters because the researchers suggest displacement in mainstream technology work could push some technically skilled people toward more cybercriminal activity. The issue is therefore not only what AI lets criminals do today, but also how AI may reshape the labor market around them.

Why this matters for technology policy

The broader technology message is that AI risk may come as much from defensive weakness and careless deployment as from offensive innovation. There is no direct Central Valley focus here, but the implications extend to any region where companies, schools, hospitals, farms, or public agencies are adopting automated systems without strong security review. The findings are also notable for California because the peer-reviewed work is scheduled to be presented at the Workshop on the Economics of Information Security in Berkeley in June, underscoring how closely the cybersecurity field is watching the real-world impact of AI adoption.

Central Valley AI is produced by the CVAI Newsdesk team and developed by Kaweah Tech, a regional firm that builds, deploys, and integrates AI solutions for businesses across California's Central Valley.

Source

https://uk.finance.yahoo.com/news/cybercriminals-struggling-adopt-ai-research-230100748.html