AI May Supplant Pen Testers, But Oversight & Trust Are Not There Yet

This report analyzes the rapid integration of AI agents into penetration testing, noting widespread adoption among testers, improving capabilities for systematic coverage and enumeration, and cost advantages demonstrated in academic research. Despite gains, current AI tools struggle with complex reasoning, GUI-driven tasks, and produce higher false positives, necessitating human oversight for validation, scoping, and accountability. Industry leaders forecast near-term feasibility for compliance-oriented scans and a long-term hybrid model where elite human experts orchestrate, validate, and continually train AI systems to maintain reliability and effectiveness.