In 2026, credit underwriting no longer operates as a bounded score-and-policy function. It sits at the intersection of fraud, credit, compliance, and operations: under rising examiner scrutiny and within a regulatory environment that is fragmenting faster than most institutions can adapt. Developed in partnership with Mastercard, this playbook examines where legacy decisioning breaks down, what regulators now require from AI-enabled credit systems, and how to deploy underwriting that is fast, explainable, and defensible. It's a practical guide to building decisioning infrastructure that holds up when it's tested.

What you'll learn:

• How the regulatory landscape is shifting, and what it demands of your systems. From ECOA and SR 11-7 to Colorado's AI Act, the EU AI Act, and expanding BNPL oversight, requirements are diverging across jurisdictions while converging on one expectation: every decision must be explainable, reproducible, and defensible after the fact.

• How modern fraud targets the underwriting process directly. Synthetic identities, first-party misrepresentation, AI-generated documents, and cross-lender velocity attacks are designed to pass underwriting as legitimate—and why evaluating fraud and credit in separate systems guarantees these losses get misclassified.

• What AI-powered credit decisioning looks like in production. Real-time data orchestration, explainable ML, unified fraud-credit signals, and the role of AI agents, plus a framework for evaluating whether a decisioning platform is built for control as well as speed.