AI Agents Pose Major New Insider Security Threat

AI agents are creating a new form of insider risk for enterprises, potentially surpassing threats posed by human adversaries, according to theCUBE Research analyst Krista Case. Speaking at Okta's Oktane event, Case highlighted that agents can act across applications, delegate tasks to other agents, and make changes faster than security teams can review them. The scale of the problem is stark: one financial firm discovered 13,000 agents in its environment but deemed only 1,000 valid. Case warned that revoking an agent's access does not undo prior actions or downstream effects, and that agents must not automatically inherit their human owners' full permissions.