THE CYBER INTERVIEW
Adversary builds AI efficiencies Most businesses today are harnessing AI to build and scale their own efficiencies. They are spending a large amount of time and money to understand what slows them down and how they can wield AI to overcome such restrictions.
“ That’ s exactly what we are seeing from these criminal groups. They are looking to harness AI to build efficiencies,” Zeki says, drawing a parallel between modern enterprises and evolving cybercriminal operations. This fact is backed by the threat report, which highlights“ AI is a tool, target and force multiplier for adversaries”.
“ If you look at threat actors a few years ago, they would have to create custom scripts and tools,” Zeki elaborates.“ We would frequently see these attackers drop these files, try and run them. It wouldn’ t work and they’ d disappear for a few hours and try again.”
Zeki notes that each organisation that these criminals are going after is significantly different.“ They’ ll have to build these tools, change them and customise them, and it’ ll take a huge amount of time,” he says. But apparently not anymore, as they can use AI“ to build those tools for them on the fly”.
Such tools are very comprehensive, with capabilities like error handling and testing, Zeki says. He notes they may also contain“ huge amounts of notes” that explain what it is doing at a particular time. The result? Gone are the days of adversaries that“ would disappear for two hours” or“ turn up again the next day” when
ALTERED SPIDER TARGETED
300 +
SOFTWARE DEPENDENCIES IN A SINGLE DAY
their initial scripts failed. Instead we have AI-powered adversaries with the power of rapid victimisation.
AI, dependencies and risks that leak down With AI, anyone can be a software developer. These coding tools may utilise a host of third-party libraries“ bringing in whatever tool it thinks is best for the solution or the quickest to get to it”, according to Zeki. The consideration is not usually“ libraries that are sanctioned or right for the business”.
26 October 2026