Making AI Work for Cyber Defenders

In this episode of Cache Me If You Can, we explore how advances in artificial intelligence are reshaping the balance between cyber attackers and defenders. As increasingly capable AI models discover vulnerabilities, navigate complex systems, and accelerate sophisticated cyber operations, organizations face a critical question: can defenders harness these same capabilities quickly and safely enough to gain the advantage?

Our guest, Carol Kuntz, adjunct fellow with the CSIS Strategic Technologies Program and former senior Department of Defense official, joins us to discuss Making AI Work for Cyber Defenders: A Strategy for Strengthening U.S. Cybersecurity. Drawing on the report, Kuntz explains why powerful AI models alone are not enough—and why secure infrastructure, effective workflows, human oversight, and the ability to translate vulnerability discoveries into deployable fixes are essential to realizing AI’s defensive potential.

The conversation examines what AI-enabled cyber defense looks like in practice, from persistent monitoring and vulnerability discovery to patching and incident response. We discuss the challenges of getting defensive tools to small businesses, open-source maintainers, and critical infrastructure operators; emerging federal efforts to coordinate vulnerability remediation; and how policymakers and developers can govern increasingly powerful frontier AI capabilities without undermining legitimate cybersecurity research or slowing innovation.