Companies Struggle to Know What Normal Looks Like for AI Agents
CIO Dive, Friday, November 13th, 2026
An Abnormal AI report finds nearly half of security teams cannot establish safe behavioral baselines for AI agents.
Companies struggle to oversee AI agents because they do not know what normal agent behavior looks like, according to a report from security firm Abnormal AI.
Some 46% of security professionals say they cannot establish safe baselines for agentic AI activity, and 53% say their organizations set no baselines for machine identities at all.
More than four in 10 businesses investing in behavioral monitoring are directing it at AI agents.
Governance is harder because agents blur the line between human and machine behavior, sometimes acting for a person and sometimes on their own initiative.