Out of the Cage: How Stochastic Parrots Win in Cyber Security Environments

TL;DR

This paper explores the innovative use of pre-trained Large Language Models as decision-making agents in cybersecurity environments, demonstrating their competitive performance and potential for complex network security tasks.

Contribution

It introduces a novel application of LLMs as attacking agents in cybersecurity, along with a new modular environment called NetSecGame for testing such agents.

Findings

LLM agents perform comparably or better than state-of-the-art agents.

LLM agents match human testers' performance without additional training.

NetSecGame mimics real network attacks and supports complex scenarios.

Abstract

Large Language Models (LLMs) have gained widespread popularity across diverse domains involving text generation, summarization, and various natural language processing tasks. Despite their inherent limitations, LLM-based designs have shown promising capabilities in planning and navigating open-world scenarios. This paper introduces a novel application of pre-trained LLMs as agents within cybersecurity network environments, focusing on their utility for sequential decision-making processes. We present an approach wherein pre-trained LLMs are leveraged as attacking agents in two reinforcement learning environments. Our proposed agents demonstrate similar or better performance against state-of-the-art agents trained for thousands of episodes in most scenarios and configurations. In addition, the best LLM agents perform similarly to human testers of the environment without any additional training process. This design highlights the potential of LLMs to efficiently address complex decision-making tasks within cybersecurity. Furthermore, we introduce a new network security environment named NetSecGame. The environment is designed to eventually support complex multi-agent scenarios within the network security domain. The proposed environment mimics real network attacks and is designed to be highly modular and adaptable for various scenarios.