Luta Security’s Katie Moussouris on Frontier Models, Bugs and Containment
Organizations that deferred patching because it was too costly or disruptive are now facing a reckoning. Artificial intelligence is generating vulnerability reports and new findings at a volume that leaves nowhere left to hide, said Katie Moussouris, founder and CEO of Luta Security.
See Also: Open-Weight Model Antares Delivers Stronger Code Security
“Everything that they left in the technical debt pile has come for collection,” Moussouris said, describing what she calls the “”repo man for technical debt arriving at an organization’s doorsteps.
The fix, she said, is prioritizing by context – identifying which bugs are most exploitable, most discoverable and tied to the most critical assets – rather than treating every AI-flagged report as equally urgent. Vulnerability disclosure programs also need stronger internal processes so companies can prevent or find more flaws themselves instead of relying heavily on external reports and bug bounties.
In this video interview with ISMG at Black Hat USA 2026, Moussouris also discussed:
- Why defenders need access to the latest frontier AI models;
- How overregulation could prevent defenders from permanently gaining the advantage over adversaries;
- Why replacing junior talent with AI could weaken the next generation of cybersecurity expertise.
Moussouris leads Luta Security, helping organizations build vulnerability disclosure and bug bounty programs. She has more than 25 years of cybersecurity experience and previously led vulnerability research and bug bounty initiatives at Microsoft and was chief policy officer at HackerOne.
