'GhostJacking' Exposes Identity Governance Gaps in AI AgentsPosted by PrivacyWireNew research shows how attackers can use security alerts and blocked events to manipulate and hijack AI agents.