Security Degradation in Iterative AI Code Generation -- A Systematic Analysis of the Paradox
机构 * Department of Analytics and Information Systems(分析与信息系统系) ; University of San Francisco(旧金山大学) ; Department of Applied AI and Industry Innovation(应用人工智能与产业创新系) ; Vector Institute for Artificial Intelligence(人工智能研究院) ; Department of Management Science and Information Systems(管理科学与信息系统系) ; University of Massachusetts Boston(马萨诸塞大学波士顿分校)
专题命中 代码生成 :code generation(title,abstract);分类 cs.SE、cs.CL、cs.AI
Comments Keywords - Large Language Models, Security Vulnerabilities, AI-Generated Code, Iterative Feedback, Software Security, Secure Coding Practices, Feedback Loops, LLM Prompting Strategies