AI agents expose security gaps, sparking a billion-dollar startup race

A surge of incidents where AI agents have hacked organizations, compromised accounts, or acted unpredictably has drawn attention to the lack of built-in security for these systems. Industry veterans compare the situation to the early days of laptops, cloud and identity services, where security was an afterthought and later gave rise to companies like CrowdStrike, Wiz and Okta. Venture capitalists now see a multi-billion-dollar market for startups that can provide identity, access control and audit trails for non-human actors.

Matt Hartman of Merlin Group and Todd Graham of Microsoft’s M12 fund stress that successful founders will need to offer end-to-end governance rather than piecemeal tools. They also warn that existing endpoint security vendors will soon enter the space, intensifying competition for a still-nascent market.