Large Language Model-enabled Vulnerability Investigation: A Review

In recent years, the integration of large language models (LLMs) into cybersecurity has demonstrated significant potential in enhancing vulnerability analysis. This paper provides a comprehensive review of current literature, focusing on the applications of LLMs in vulnerability discovery, exploitation, and validation. We examine various LLM-powered frameworks that have automated aspects of vulnerability analysis, reduced the time required for vulnerability identification, and improved the precision of vulnerability assessment. In addition, we discuss LLM-driven advancements in security vulnerability exploitation and validation, which facilitate more efficient and accurate mitigation. The contributions of this review include an extensive synthesis of existing studies, a proposed framework that highlights the role of LLMs across different stages of the vulnerability lifecycle, and an outline of future research directions in LLM-based cybersecurity. Our findings aim to guide researchers and practitioners in developing robust, scalable, and automated cybersecurity solutions powered by LLMs.