What happened
Keeper Security announced (July 7, 2026) agentic AI governance for its Endpoint Privilege Manager, extending zero-trust policy enforcement from human users to AI agents (e.g., GitHub Copilot, Cursor, Claude Code, Amazon Q) operating on employee workstations, combining a signed agent-identity catalog with a proprietary AI-likelihood score for unknown agents and evaluating actions like process spawning and file access at the OS level.
Why it matters
Extends PAM/zero-trust identity governance to non-human AI agent 'principals' at the OS/endpoint layer rather than just the MCP/API layer, addressing a gap as enterprises scale agent deployment (Gartner predicts F500 orgs running 150,000 agents by 2028).
Applicability
IAM/endpoint security teams already using Keeper PAM should evaluate this as part of governing developer-machine AI agent sprawl; relevant now given rapid AI coding agent adoption.