The AAIA™ Study Guide emphasizes that the foundation of any high-performing and ethical AI system lies in the quality and integrity of its data. For high-risk AI systems, such as those used in healthcare, finance, or criminal justice, it is essential to base models on trustworthy data. This ensures reliable predictions, reduces bias, and mitigates risk.

“Trustworthy datasets are characterized by accuracy, completeness, consistency, and ethical sourcing. In high-risk AI applications, ensuring data quality at every stage is crucial to system reliability and compliance.”

While backups, user training, and MFA are important for security and operational resilience, they do not address the core challenge of ensuring model accuracy and fairness at the development and design phase. Therefore, option C is the most effective practice.