The agent security gap: 54% of enterprises have already had an AI agent incident, and most still let agents share credentials
VentureBeat
Most enterprises face AI agent security risks due to insufficient identity management and a reliance on basic provider-native guardrails.Summary
Research from VentureBeat indicates that 54% of enterprises have experienced AI agent security incidents or near-misses. A critical 'agent security gap' exists because most organizations fail to assign unique identities to agents, leading to widespread credential sharing that increases blast radii during breaches. While companies express high satisfaction with provider-native security tools from model vendors, they remain underprepared, with few utilizing advanced isolation techniques like sandboxing. Consequently, a majority of enterprises plan to replace their security tooling within the year as they struggle to keep pace with AI-enabled threats.
(Source：VentureBeat)