Code security is the degree to which software protects systems and data from unauthorized access, misuse, or compromise. AI-generated code is not automatically secure, especially when it handles sensitive data or supports safety-critical systems. section 1.5
Treat AI-generated code as untrusted until it has been checked:
- Understand the code thoroughly instead of accepting plausible output without scrutiny. section 1.5
- Use security audits and appropriate expertise when code handles user data or will be used in real-world systems. section 1.5
Sensitive and safety-critical code
Code for medical devices or code that handles sensitive user data requires more than generated output: it must be understood and subjected to security review. section 1.5
