AI agent identity security demands layered defenses, Omdia says

AI agent identity security remains uncertain, as enterprises grapple with where attacks may originate and which security layers should respond. Omdia's Todd Thiemann, principal analyst for identity and access management and data security, suggests that consolidated platforms are needed to protect the growing class of machine identities.

Palo Alto Networks' acquisition of CyberArk Software Ltd. reflects this consolidation in the identity security market. Okta recently added an AI agent runtime gateway and formed the Blueprint Alliance with Amazon Web Services and CrowdStrike. However, enterprises are still unsure about which security layers should respond to agent attacks.

Thiemann's survey of 400 security leaders shows that the number one answer was "my data security platform," indicating a preference for Palo Alto Networks, which now includes CyberArk or Microsoft. The second most popular answer was "I want to use my identity platform." The Blueprint Alliance defines gateway capabilities, such as authentication and authorization, but data security and other risks may require controls from multiple providers, increasing policy coordination and potential gaps.

Thiemann emphasizes the need for defense in depth, as the landscape surrounding AI agents is still evolving.