The agent security gap: 54% of enterprises have already had an AI agent incident, and most still let agents share credentials

A survey of 107 enterprises reveals that AI agents are granted real access to systems and data, yet the controls to contain them lag behind. Over half of the firms have experienced a confirmed incident or near‑miss, only about a third assign each agent its own scoped identity, and most agents continue to share credentials. Furthermore, just three in ten isolate their highest‑risk agents, and the security stack is largely borrowed from model providers and hyperscalers rather than purpose‑built for agents.

Summary written by Kernelia from the original article by VentureBeat AI. The story and its rights belong to its author.