This is the authors' abstract. We don't add key points for this paper.
This paper presents an AI-driven approach to enhance dynamic code analysis by leveraging Large Language Models (LLMs), such as Falcon 40 B and Code Llama. The proposed framework continuously monitors key performance metrics-execution time, memory utilization, and computational overhead-while employing machine learning algorithms to detect inefficiencies and security vulnerabilities in real-time. Unlike traditional dynamic analysis methods, which rely on predefined rules, this system adapts to emerging threats and evolving vulnerabilities, streamlining code review and security assessments. By analyzing runtime data and recognizing patterns, the LLMs provide actionable recommendations to improve both performance and security. Experimental results demonstrate that this AI-assisted approach significantly improves security posture, reduces manual intervention, and enhances overall code maintainability. Despite challenges such as false positives and ethical concerns, the proposed method offers a scalable, efficient solution for secure software development. Future work will focus on refining model transparency, minimizing biases, and integrating AI-driven security assessments with penetration testing workflows.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Murugalakshmi et al. (2025) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: