Which elements should model governance include?

• A. Data labeling standards and model deployment controls
• B. Oversight of model lifecycle; includes policies, roles, approvals, versioning, drift monitoring, and incident response
• C. Hardware procurement for AI infrastructure
• D. End-user documentation

Answer: (B)

Model governance is the system of rules and processes that oversee the entire model lifecycle, ensuring accountability, compliance, and risk management. The best answer captures oversight of the lifecycle plus concrete elements that make governance actionable: policies that define responsibilities, roles and approvals to authorize changes, versioning to track model iterations, drift monitoring to detect performance shifts, and incident response to address failures or misuse. This holistic framework ensures that as models evolve, there is clear governance around who can make changes, how those changes are approved, how we detect problems, and how we respond when issues arise.

Data labeling standards and deployment controls are important components within the lifecycle, but they describe specific controls rather than the overarching governance structure. Hardware procurement concerns resources and infrastructure, not the governance framework, and end-user documentation supports usage and transparency rather than governing the model’s lifecycle.