Where AI Labs Will and Won't Disrupt Cybersecurity

Foundation Capital's Sid Trivedi on the Three Markets AI Labs Can't Easily Enter Michael Novinson (MichaelNovinson) • March 29, 2026

Artificial intelligence labs entered cybersecurity through its most obvious door: application security. Moving from static analysis into dynamic testing was a natural extension of their code generation capabilities. But that is where the disruption has clear limits.

See Also: Taming the Rise of Shadow AI Agents

"The labs are mostly focused on horizontal value. For them, going deeply embedded into security doesn't make a lot of sense," said Sid Trivedi, partner at Foundation Capital.

Within that constraint, as AI-generated code increases, securing that output has become a natural adjacency, pushing labs deeper into development-stage security.

Trivedi identified three areas less likely to be disrupted: runtime sensors that require deep endpoint build-out, security functions built on proprietary data that can't be publicly trained on, and SOC and incident ...