This research provides the first comprehensive analysis of API misuse patterns in LLM-generated code, offering systematic methods to detect and mitigate these vulnerabilities.

• Identifies common API misuse patterns in code generated by large language models
• Analyzes both method selection errors and parameter usage mistakes that lead to security risks
• Proposes effective mitigation strategies for developers and LLM providers
• Demonstrates improved security outcomes through practical implementation of recommended approaches

This work addresses critical security gaps in AI-assisted software development, helping organizations prevent potential vulnerabilities while still leveraging the productivity benefits of LLMs for coding tasks.

Identifying and Mitigating API Misuse in Large Language Models