Security Brief Asia reports that Teleport has launched an agentic identity framework designed to secure autonomous and semi-autonomous AI agents operating across cloud and on-premises infrastructure.The company asserts that traditional identity and access controls are inadequate for AI agents, which operate at scale with minimal human oversight, leading to risks like secrets sprawl and fragmented visibility. Based on a survey of over 200 infrastructure leaders, 69% believe AI adoption requires significant identity management changes. The framework treats agents as "first-class identities," utilizing cryptographic controls, ephemeral credentials, and open standards like MCP and SPIFFE to establish trust. CEO Ev Kontsevoy stated a unified identity layer is a prerequisite for enterprise AI, warning that deploying on "fragmented credentials... is a recipe for secrets and data leakage."The approach contrasts with security methods focused on model safety or runtime monitoring, positioning identity as the foundational control. Teleport's underlying platform provides a hardware root of trust and real-time behavior visibility to enforce zero-trust principles for agentic systems.
You can skip this ad in 5 seconds