OUTLINE: Data, AI, and Model Risk

Summary
This page covers the specific governance burden created by data quality, metadata, lineage, model risk, AI, and vendor dependence. It should help leadership understand that many banks already have more AI and model exposure than they realize, especially through vendor tools, and that governance needs to account for those risks before they become scrutiny issues.

Outline

• Why data and AI risk rise together
• Many analytics are already models
• Vendor AI is still the bank’s risk
• What changes by stage:
  • Under $10B: inventory and awareness
  • $10B–$50B: tiering, validation, approvals, monitoring
  • $50B–$250B: lifecycle governance, explainability, independent challenge
  • $250B+: industrialized AI governance at scale
• Key themes:
  • data quality and critical data elements
  • metadata and lineage
  • model inventory and oversight
  • AI use case governance
  • vendor and third-party risk
• Example visuals:
  • model lifecycle
  • analytics-to-model continuum
• Embedded assets:
  • AI/model intake form
  • model tiering example
  • vendor AI checklist