JumpCloud extends Agentic IAM to govern autonomous AI agents

JumpCloud has extended its Agentic IAM service to give IT teams control over autonomous AI agents operating inside their organisations. Agents are now registered in the same directory as employees, and each one must be tied to a named human owner, with unowned agents blocked from running.

The release adds four capabilities: a unique identity per agent so administrators can audit activity or revoke access without affecting the deploying employee, discovery of Model Context Protocol servers set up outside official channels, extension of existing device trust and access policies to AI tools, and a privileged access management server due before the end of the fourth quarter.

Privileged access grants to agents expire with the task they were issued for, which JumpCloud said allows companies to point AI at legacy systems without rebuilding them. The company cited its July survey of IT leaders, which found only 21% of organisations have governance controls around nonhuman identities, the lowest score of 10 recommended AI security practices, while nonhuman identities outnumber people at 83% of those surveyed.