The speed and scope of AI security tools is desperately needed, but humans may be a limiting factor. In a roundtable in San Francisco on Wednesday, security experts agreed that the speed and capabilities of AI systems to find and fix vulnerabilities have changed the game for IT managers, but that humans might be the weakest link.
While the ability to discover flaws has drastically increased with the introduction of powerful cybersecurity AI models, the motivation to fix them is still dependent on human decision-making, which has both pluses and minuses.
“I think the human in the loop is in a complete failure mode. I think the human contributor is probably like, you want the intent, you want the experience, you want the taste, you want the judgment call, but you want to provide them with the most effective set of information that agents actually can do a very good job today of triaging,” said Ed Sim, founder of venture capital firm Boldstart Ventures, who predicted a massive increase in security spending thanks to AI.
“It's mind-boggling while we're talking about autonomous attacks in a large enterprise customer base. The number of issues found versus fixed is eight to one. Every issue fixed, eight more have been found. Why? The human in the loop is the bottom line. The human has to approve everything.”
