Why Hybrid Offensive Security Beats Agentic AI Alone

This report summarizes an independent study comparing AI-driven agentic penetration-testing frameworks to human penetration testers, concluding that agentic AI offers superior scale, coverage, and cost-efficiency for infrastructure-level findings but lacks the creativity, GUI reasoning, and business-logic understanding of humans; the authors recommend a hybrid AI-plus-human model for continuous threat exposure management.