Ilmu Komputer & AI editorial
Open AccessOA2024
Application of Large Language Models in Cybersecurity: A Systematic Literature Review
This systematic review of 177 articles (2018โ2024) examines the application of large language models in cybersecurity, highlighting a surge in research (68% in 2023), geographic concentration (USA 30%, China 11%), and a predominant focus on defensive measures (27%). LLMs show high effectiveness in phishing simulations, administrative tasks, and security software development.
I. Hasanov; Seppo Virtanen; Antti Hakkala; J. Isoahoยท IEEE Accessยท 2024ยท DOI 10.1109/ACCESS.2024.3505983
The core problem
The rapid emergence of Large Language Models (LLMs) is driving a paradigm shift in how digital technologies are used across societies and businesses. While this disruptive effect is particularly visible in information and communication technology, there is a notable lack of holistic systematic studies on the application and impact of LLMs in cybersecurity. This article addresses that gap by presenting an exhaustive systematic literature review of 177 articles published between 2018 and 2024 on the application of LLMs and Artificial Intelligence (AI) as a defensive measure in cybersecurity. The review aims to provide an analytical compendium of recent research on LLMs in offensive and defensive cybersecurity, cyberethics, legal frameworks, and cybersecurity governance, along with a statistical summary of global research trends.
Innovation
The authors conducted a systematic literature review following a structured protocol to identify, screen, and synthesize relevant studies. The search covered articles published from 2018 to 2024, resulting in a final corpus of 177 articles. The review categorized the literature into thematic areas: offensive cybersecurity, defensive cybersecurity, cyberethics, legal frameworks, and governance. Additionally, the study performed a statistical analysis of publication trends, including temporal distribution, geographic contributions, and thematic focus. The methodology ensured a comprehensive and unbiased selection of studies, although specific databases and search strings are not detailed in the abstract.
Introduction
The rapid emergence of Large Language Models (LLMs) is driving a paradigm shift in how digital technologies are used across societies and businesses. While this disruptive effect is particularly visible in information and communication technology, there is a notable lack of holistic systematic studies on the application and impact of LLMs in cybersecurity. This article addresses that gap by presenting an exhaustive systematic literature review of 177 articles published between 2018 and 2024 on the application of LLMs and Artificial Intelligence (AI) as a defensive measure in cybersecurity. The review aims to provide an analytical compendium of recent research on LLMs in offensive and defensive cybersecurity, cyberethics, legal frameworks, and cybersecurity governance, along with a statistical summary of global research trends.
Why it matters
The findings underscore the transformative role of LLMs in cybersecurity, particularly in defensive applications. The dominance of defensive research (27%) suggests a strong focus on leveraging LLMs to protect systems, although offensive uses are also explored. The surge in 2023 publications reflects the rapid adoption of LLMs following the release of models like ChatGPT. The geographic disparity in research output highlights the need for greater international collaboration, especially from underrepresented regions. The effectiveness of LLMs in phishing simulations and administrative tasks points to their potential to augment human capabilities, but also raises ethical and legal considerations. The review calls for further research on cyberethics, legal frameworks, and governance to ensure responsible deployment. Overall, LLMs are positioned as a double-edged sword, requiring balanced scrutiny to maximize benefits while mitigating risks.
Who should read this
Pembaca Research Digest dan praktisi rumpun ini.
Opening member contentโฆ