Survey: 54% of Enterprises Hit by AI Agent Security Incidents as Identity Controls Lag

A VentureBeat survey of 107 enterprises reveals a critical security gap in AI agent deployments: 54% have experienced confirmed incidents or near-misses, yet only 32% assign unique scoped identities to each agent, and most still share credentials across agents. Organizations rely heavily on provider-native guardrails from OpenAI, Google, and Microsoft rather than purpose-built agent security tools, while spending under 10% of their security budgets on the problem despite high incident rates and widespread credential sharing that amplifies blast radius when agents are compromised.

Why it matters: As enterprises rapidly deploy autonomous AI agents with real system access, the gap between agent autonomy and security controls represents an urgent infrastructure gap that will reshape enterprise security spending and vendor selection in the coming year.