'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.
We haven't written up this one. Dark Reading has the full story — the link below goes straight to it.