Câu 38: Isaca Advanced in AI Audit

Which of the following controls would BEST minimize the risk of bias introduced into AI models from flawed training data?

1. A. Implementing enhanced access controls over AI model training data
2. B. Testing AI models using adversarial methods
3. C. Separating AI training and production data
4. D. Performing cross-validation of AI models — đáp án hiện tại