Banks quietly reshape cards and credit with artificial intelligence
Major U.S. banks are quietly reshaping who gets the best perks and who loses access to credit by embedding artificial intelligence deeper into card and lending systems. Over the past six months, institutions have rolled out targeted reward programs and new underwriting models that favor predictable, high-balance customers while tightening books for more marginal borrowers. That shift is already changing incentives and balances across consumer credit.
Personalized perks for wealthy cardholders
Banks are moving from one-size-fits-all points tables to dynamic, data-driven incentives that push affluent customers toward higher-margin products. At Citi's investor presentation on May 7, 2026, executives described using AI to route spending, tailor travel benefits, and nudge customers to cards that capture more of their premium purchases. The approach treats rewards as a revenue management tool rather than a simple marketing perk.
Bank of America and other large issuers have also publicly retooled loyalty and rewards in the first quarter of 2026, explicitly linking the new programs to account balances and premium engagement. The explicit goal is to lift consumer profitability, with one bank setting a target to more than double card-related consumer profit to $20 billion by the end of the decade. Higher balances and steadier deposit behavior now map to richer benefits.
How the models work
The technologies at play are not simple scoring tweaks. Lenders and card managers are deploying machine learning that ingests real-time transaction flows, deposit patterns, travel and subscription spend, and product usage. Algorithms then rank customers by predicted lifetime value and elasticity to offers, and the bank's front end surfaces personalized bonus categories, statement credits, and exclusive experiences to the top tiers. That personalization increases revenue per active customer while concentrating perks.
Who pays the price
The same wave of automation is remaking underwriting. AI models that rely on recent cash flow, payroll deposit cadence, and behavioral signals can outperform legacy bureau scores at predicting near-term repayment. But that can also harden a two-tier market: borrowers whose accounts show volatility or low balances - often those with lower FICO scores - are more likely to see limits reduced, higher pricing, or outright declines. Recent research finds that about one third of unpaid credit card balances exist because limits were increased after account opening, a dynamic that interacts with algorithmic decisions and that rises to 60 percent among lower-score borrowers in some samples. Those mechanics can both push indebted consumers deeper into revolving debt and reduce credit access for marginal profiles.
Risk, oversight, and the business calculus
Banks argue the changes improve risk management and customer value. Regulators and consumer advocates warn the same tools can produce opaque, unequal outcomes if models optimize for profit without guardrails. Supervisory reporting and risk reviews this year flagged uneven credit conditions across borrower segments and urged firms to monitor concentration and model bias. The trade-off is stark: more efficient, personalized products for the affluent, and declining credit availability for others.
The practical implication
For consumers who bank with large national issuers, the near-term trend is clear. If your accounts show stable deposits and frequent premium spending, expect richer offers and more automation in service. If your financial profile is volatile or underbanked, expect tighter limits and heavier reliance on alternative signals that can be unforgiving. That bifurcation in credit and rewards is becoming a defining feature of the modern retail banking landscape.