Home Cybersecurity Article

69% Of Enterprises Deploy AI Agents With Shared Credentials

TL;DR

Survey reveals critical agent security gap: majority of enterprises share API keys across agents, eliminating audit trails and multiplying breach impact.

Key Points

  • 69% of enterprises run agents with credential sharing; 54% have experienced agent security incidents or near-misses
  • Only 32% give each AI agent its own scoped identity; 30% sandbox high-risk agents despite 63% incident rate at large companies
  • Incident rate jumps to 63% at enterprises with 1,000+ employees, but sandbox isolation drops to 20%—creating 60-point exposure gap
  • Palo Alto Networks ($21.1B CyberArk acquisition), CrowdStrike ($740M SGNL acquisition), and Cisco ($400M Astrix acquisition) targeting this security layer

Why It Matters

Shared credentials eliminate attribution when agents are compromised, converting single-agent breaches into multi-system incidents. Large enterprises face the worst risk-to-containment ratio, making agent identity management and isolation critical infrastructure decisions that most organizations have deferred.
Full VentureBeat Q2 Agentic Security Report

Source: venturebeat.com