Artificial Intelligence and Credit Card Payment Networks: Examining the Risks for FDIC-Supervised State-Chartered Banks

Abstract

The integration of artificial intelligence (AI) into U.S. credit card payment networks is reshaping fraud detection, transaction routing and payment processing. Visa, Mastercard, American Express, and Discover have increasingly relied on AI-driven models, yet the implications of these models for FDIC-supervised state-chartered banks remain largely unexamined. This paper explores (1) whether AI-driven fraud detection systems lead to higher false positive rates for state-chartered banks, impacting transaction approval rates and customer retention, and (2) whether AI-based payment routing algorithms disadvantage smaller issuers by deprioritizing their transactions, creating liquidity and settlement inefficiencies. This paper simulates how AI based fraud detection and payment routing may disadvantage FDIC-supervised state-chartered banks. It also examines whether the FDIC should strengthen its oversight by updating its Credit Card Activities Manual to require banks under its supervision to collect data on third-party AI decision-making and whether a broader interagency working group should be established to monitor systemic risks associated with AI in credit card payment systems.

Keywords: Payments, AI, Bias, FDIC, Credit Cards

JEL Classification: G17, G21, D81