The level of expertise and the phishing attacks we're seeing now powered by AI is actually quite frightening. You can attack people on an individual basis based on their own psychological profile, using the best sales capability in the world.
AI can be an excellent ally—or a powerful adversary, using insights and personal data to launch highly sophisticated attacks. Experts are sounding the alarm, warning that much stronger protections will be needed to defend against advanced threats powered by AI.
We sat down with Peter Benson, a security expert with 25 years of experience and CEO of Neural Horizons, to discuss the potential dangers AI could expose us to in the future and what steps we can take to protect ourselves.
Master manipulator: Benson doesn’t mince words when it comes to the potential for AI to be used maliciously: "AI for offense is absolutely coming. There are no two ways around it," he says. "The level of expertise and the phishing attacks we're seeing now powered by AI is actually quite frightening." He points to the power of AI tools like Alpha Persuade, which can manipulate behaviors with alarming precision. "You can attack people on an individual basis based on their own psychological profile, using the best sales capability in the world," he explains. The result? A new era of highly targeted scams and financial fraud.
Nefarious coding: Beyond the realm of individual attacks, Benson warns of even more troubling developments in AI-driven software. "We’re going to see AI generating malicious software," he predicts. "The ability to do that now, particularly with agent-based approaches, is absolutely staggering."
Benson also highlights the dangerous potential of compromised AI libraries. These libraries, which may contain malware, can be incorporated into applications without the user’s knowledge, automatically deploying harmful code. "AI can pick up libraries that either don’t exist or that have already been tainted with malware," he warns, emphasizing the need for stronger safeguards against such threats.
There have been a number of studies recently showing that the use of AI can actually lead to a decline in cognitive thinking and analytical reasoning. And yet, the World Economic Forum says those are the very skills we need more of.
A force for good: While the dangers of AI are daunting, Benson also sees potential for AI to be used defensively. "We need to be looking at AI for defensive purposes, big time," he says. Security operations teams are often overwhelmed, and AI could help with first-level analytics and detecting behavioral anomalies. "Having AI assist in these areas is a big opportunity," he notes.
Slippery slope to laziness: For Benson, AI literacy needs to be forefront to tackle challenges to safety. "It’s not just about understanding hallucinations and bias," he says, "but also how to maintain critical thinking, analytical reasoning, and a healthy degree of skepticism."
AI, he argues, shouldn’t be our frontline problem-solving technique. Doing so puts us all at risk of becoming lazy with our thinking. "There have been a number of studies recently showing that the use of AI can actually lead to a decline in cognitive thinking and analytical reasoning," Benson explains. "And yet, the World Economic Forum says those are the very skills we need more of. So, we're stuck in this paradox where we need to shift what it means to get the most effective use out of AI. We’ll have to emphasize developing those skill sets, whether through government initiatives, organizational efforts, or skills training."
