The adoption of large language models (LLMs) as an advanced artificial intelligence (AI) system has seen significant growth, especially in the cybersecurity domain. Thus, it is imperative to understand the impact of LLMs in cybersecurity since they pose pros and cons regarding digital threats and attacks. This study systematically reviewed existing research on how LLMs impact cybersecurity, examining their benefits and challenges. Researchers conducted a systematic search across three databases, namely Scopus, IEEE and Science Direct. These databases were chosen due to their relevance to the discipline, focusing on peer-reviewed English publications from 2020–2024. A total of 47 articles were identified, 30 were deemed suitable for synthesis, and 17 were excluded as they did not meet the criteria. Using thematic analysis, the authors identified key themes related to LLMs in cybersecurity. The study’s key findings revealed several important insights, such as LLMs’ ability to assist in network anomaly detection, carry out automated threat analysis, conduct enhanced threat detection, improve incident response, carry out knowledge sharing and conduct training and awareness. In contrast, the negative findings reveal that security systems based on large language models (LLMs) may be vulnerable to cyberattacks, contribute to spreading misinformation and disinformation, raise legal, ethical, and fairness concerns, and create a false sense of security. The contribution of the study is that it provides a foundational understanding of the impact of LLMs on cybersecurity and identifies areas for future investigation. Subsequent studies should explore undiscovered impacts, challenges, and potential future developments in LLMs impacting cybersecurity in specific industries.