'GhostJacking' Exposes Identity Governance Gaps in AI Agents

New research shows how attackers can use security alerts and blocked events to manipulate and hijack AI agents.