Taught by the Flawed: How Dataset Insecurity Breeds Vulnerable AI Code

Catherine Xia, Manar H. Alalfi

AI programming assistants have demonstrated a tendency to generate code containing basic security vulnerabilities. While developers are ultimately responsible for validating and re…