LAS VEGAS — Securing global networks will be “rocky” over the next several years, as new artificial intelligence models grow more adept at finding and exploiting flaws in digital systems faster than defenders can fix them, U.S. and British officials warned Wednesday.
“We are discovering vulnerabilities at scale in ways that we never have before; we are piling up the vulnerabilities in need of remediation higher than they’ve ever been stacked before,” Michael Duffy, U.S. acting federal chief information security officer, said at a panel at the Black Hat cybersecurity conference in Las Vegas on Wednesday.
Jonathon Ellison, director of national resilience at the United Kingdom’s National Cyber Security Centre, said on the same panel that while these tools have the capacity to greatly improve threat detection methods in the future, “we’ve got a really, really rocky road on the way to achieving that.”
He blamed years of “underinvestment” in cybersecurity for the mounting pile of security flaws found by advanced AI models from major labs including OpenAI and Anthropic.
