Over the past year, Comcast's cybersecurity teams have conducted structured evaluations of multiple frontier AI models across a wide range of cybersecurity activities—spanning more than 225 million lines of code and supporting enterprise infrastructure. The goal was straightforward: understand how these capabilities will impact cybersecurity at a large-scale critical infrastructure operator.

The findings, accelerated through the team's participation in Project Glasswing, challenged—and even upended—several long-held assumptions about where cyber risk resides and how security programs need to evolve as frontier AI capabilities continue to improve.

In this 30-minute RSAC Webcast, Noopur Davis, EVP, Chief Information Security & Product Privacy Officer, Comcast, will discuss the findings and how lessons learned contribute to the broader cybersecurity conversation going forward.