Broken by Default: A Formal Verification Study of Security Vulnerabilities in AI-Generated Code

1. 1Identify security-critical prompts and generate code artifacts using LLMs
2. 2Apply formal verification techniques, such as Z3 SMT solver, to detect security vulnerabilities
3. 3Analyze results to quantify exploitability of AI-generated code
4. 4Develop strategies to mitigate security risks in AI-generated code

Security engineers and AI researchers benefit from this study as it highlights the need for rigorous testing of AI-generated code, particularly in security-sensitive domains

💡 AI-generated code can contain security vulnerabilities that can be exploited, highlighting the need for formal verification and testing

🚨 AI-generated code may be #brokenbydefault 🚨