Computer Science

PENTESTGPT: Evaluating and Harnessing Large Language Models for Automated Penetration Testing

G. Deng, Y. Liu, et al.

LLMs promise to transform penetration testing—this study builds a real-world benchmark and shows they excel at sub-tasks but struggle with whole-context reasoning. Introducing PENTESTGPT, a three-module, LLM-driven framework that boosts task completion by 228.6% over GPT-3.5 and succeeds on real-world targets and CTFs. The research was conducted by the authors present in <Authors> tag and PENTESTGPT is open-sourced with strong community uptake.