An organization is implementing an AI-based credit assessment engine using internal and third-party customer data.

AAISM emphasizes that data governance over the full AI life cycle is foundational. The official content describes effective AI data management as including documented procedures for: (1) how data is sourced, (2) how lineage is tracked from origin to model, and (3) how data quality is validated and monitored. This ensures transparency, accountability, and auditability, which are especially critical in regulated areas like credit assessments. While hashing identifiers (B) and encryption/access controls (C) are important privacy and security mechanisms, they are partial controls within a broader governance framework and do not, on their own, establish end-to-end life-cycle management. Limiting training to structured data (D) is a design choice and may reduce risk but is neither sufficient nor required as a best practice. Option A directly reflects AAISM’s prescribed governance controls for AI data throughout its life cycle.