A recent report by Enterprise Management Associates reveals that 65% of enterprises have observed AI agents acting outside their designated scope, with 29% experiencing organizational impacts. Among the surveyed technology leaders, 46% are deploying agentic AI across departments, and nearly 79% utilize both generative and agentic AI simultaneously. Many organizations face governance challenges, with only 32.2% able to quickly detect and respond to out-of-scope actions.
The report highlights identity enforcement gaps, with 30% of AI pilot programs paused due to security risks. Recommendations include evaluating agent authorization in real-time and improving automated containment measures.