The article discusses the potential of Large Language Models (LLMs) in automating cyber-attacks, focusing on the pre- and post-breach stages. It introduces AUTOATTACKER, a system designed to automate "hands-on-keyboard" attacks using LLMs. The system consists of a summarizer, planner, navigator, and experience manager to interact with LLMs iteratively. Extensive testing shows that GPT-4 is effective in conducting post-breach attacks with limited human involvement. The research aims to understand the risks and impacts of automated cyber-attacks using advanced LLMs.
To Another Language
from source content
arxiv.org
Key Insights Distilled From
by Jiacen Xu,Ja... at arxiv.org 03-05-2024
https://arxiv.org/pdf/2403.01038.pdfDeeper Inquiries