According to a new VentureBeat Pulse Research survey of 107 enterprises, more than half have already experienced an AI agent security incident or near-miss.
The June 2026 survey found that 54% of organizations with over 100 employees experienced a confirmed AI agent security incident or a near-miss.
The credential sharing problem is worse than you think
69% of the enterprises surveyed admitted to sharing credentials across their AI agents. When agents share credentials, a single breach can cascade across systems. These are autonomous agents with real access to databases, financial systems, and operational infrastructure.
Only about a third of the surveyed organizations give each agent its own scoped identity. Meanwhile, just 30% of enterprises isolate their highest-risk AI agents.
