KnowBe4 Strategist Urges Human-in-the-Loop Agentic AI Security

Perry Carpenter, Chief Human Risk Management Strategist for KnowBe4, is advocating for the adoption of a human-in-the-loop architecture as cybersecurity transitions toward agentic AI. These autonomous systems employ a central orchestrator and specialized agents to hunt threats, monitor networks, and simulate attacks, providing speed and scale that exceed human capacity.

Carpenter warns that excessive autonomy in these systems can trigger error propagation, produce false positives, and create automated behaviors that conflict with broader business goals. He specifically notes that adversaries may attempt to exploit inter-agent communications or inject deceptive intelligence to undermine security pipelines.

To mitigate these risks, Carpenter proposes a framework that balances autonomy with control. This approach requires the establishment of clear operational roles, the use of explainable AI for auditing purposes, and mandatory human review for all high-impact decisions.