Exploring the Potential and Obstacles of Large Language Models in Enhancing Code Security

Large Language Models (LLMs) have significantly transformed the field of software engineering by providing novel approaches to improve code security. This systematic literature review (SLR) consolidates recent findings to investigate the potential benefits and obstacles associated with the use of LLMs in this area. We pinpoint major advantages, including enhanced vulnerability identification, automated code evaluations, and secure code creation, while also addressing challenges such as model inaccuracies, biases in training data, high computational demands, and ethical concerns. Based on an analysis of ۳۵ studies published from ۲۰۲۰ to March ۲۰۲۵, this review emphasizes the groundbreaking possibilities offered by LLMs while highlighting the necessity for strategies to mitigate their drawbacks. The insights gained serve as a guide for researchers and practitioners looking to utilize LLMs in software development where security is paramount.

کلیدواژه ها:

Software Engineering ، Large Language Model ، Ethical Artificial Intelligence ، Secure Code