This study develops
and validates GUARD (Governance, Underwriting integrity, Accountability,
Resilience, Disclosure), a risk-controlled framework for the deployment of
artificial intelligence in credit origination and portfolio management by U.S.
banking organizations. It examines how lenders can convert AI-driven
underwriting into sustained credit portfolio growth while retaining defensible
control over fair lending exposure, model risk, and consumer explanation
obligations, in a supervisory environment that between April 2025 and June 2026
moved decisively from prescriptive to principles-based expectations.
