Table 4 Importance and normalized importance of independent variables (cybersecurity risks).
Cybersecurity risks in automatic code generation | Importance | Normalized importance (%) |
|---|---|---|
Injection attacks | .177 | 100.0 |
Code quality and logic errors | .053 | 29.9 |
Backdoors and malicious code | .051 | 29.0 |
Vulnerabilities in reused code (legacy dependencies) | .072 | 40.6 |
Insufficient input validation | .037 | 20.7 |
Weak authentication and authorization mechanisms | .062 | 34.9 |
Lack of encryption OR insecure data handling | .063 | 35.7 |
Reusability of vulnerable code | .091 | 51.3 |
Lack of secure code review and testing | .074 | 42.0 |
Adversarial attacks on AI models | .052 | 29.3 |
Overreliance on an AI model | .013 | 7.1 |
Privacy issues and data leakage | .126 | 71.2 |
Insecure integration with other systems | .019 | 10.8 |
Insufficient logging and monitoring | .112 | 63.3 |
