AI Rewrites Lending for Borrowers FICO Scores Miss
For millions of individuals in the U.S., exclusion from credit markets has less to do with repayment risk than with how creditworthiness has traditionally been measured.
PYMNTS reported in 2025 that while the Consumer Financial Protection Bureau revised its estimate of strictly “credit invisible” consumers downward, a much larger population remains unscored. Roughly 25 million U.S. adults lack sufficient recent credit activity to generate a usable score, meaning they are often excluded from traditional underwriting despite having income and recurring financial obligations.
Static Credit Scores Leave Structural Blind Spots
Traditional credit scoring models rely heavily on static bureau data such as credit cards, installment loans and repayment histories. Those signals work well for consumers already embedded in the credit system but fail to capture financial behavior that occurs outside conventional credit products, especially for so-called “thin file” individuals who have limited information detailing their financial lives.
PYMNTS Intelligence flagged the consequences of that design in its 2023 collaboration with Sezzle, which found that nearly one-third of U.S. consumers were credit insecure. The research showed that repeated rejection often pushes consumers from being credit marginalized into becoming credit avoidant, shrinking access to safer credit products over time.
AI Brings New Credit Signals Into Focus
Much has changed in terms of the data lenders can now analyze.
Lenders are increasingly using artificial intelligence to process alternative data to identify creditworthy borrowers who would otherwise be missed by traditional models. In particular, AI systems are being used to analyze transaction-level data and behavioral patterns to simultaneously improve fraud detection and credit decisioning.
Rather than relying on a single score, AI allows lenders to evaluate consistency, volatility and resilience across time, and in real time, helping distinguish between temporary disruption and structural risk.
Cash-Flow Data Shows What Scores Miss
Among the most impactful alternative signals is cash-flow data.
PYMNTS has reported that cash-flow-based underwriting enables banks and FinTechs to see borrowers that FICO scores miss by analyzing income deposits, expense patterns and the ability to meet recurring obligations such as rent, utilities and subscriptions. These data points reflect real financial commitments, even when no formal credit line exists.
For consumers with irregular income or limited credit histories, cash-flow analysis often provides a more accurate picture of repayment capacity than relying simply on averaged bureau metrics.
Regulators Signal Support for Responsible Alternative Data Use
The expansion of alternative data in lending is occurring with increasing regulatory acknowledgment.
The Office of the Comptroller of the Currency has publicly recognized the value of alternative data in expanding credit access when used responsibly. OCC leadership emphasized that these data sources can help lenders better assess risk and responsibly serve consumers who are poorly represented by traditional credit models. That position reinforces the idea that inclusion and safety are not mutually exclusive.
Companies are already deploying AI and alternative data to enhance underwriting and expand access for thin-file or previously overlooked borrowers. Banks and FinTechs are partnering for cash-flow-based underwriting to better evaluate actual repayment capacity instead of relying solely on legacy credit scores.
Firms are tapping data such as bank account transactions, payroll deposits and spending patterns to assess real-time income stability and liquidity, enabling credit decisions for individuals whose traditional bureau footprints would have left them unscorable.
Notably, banks including Chase and digital lenders including PayPal have announced integrations with cash-flow data partners such as Nova Credit to power these underwriting capabilities … which allows lenders to fine-tune credit boxes rather than abandon them.
For credit invisibles and for thin file individuals, this approach creates an on ramp into safer, mainstream credit products for the borrowers. It replaces blunt exclusion with calibrated risk management. The result, for the lenders, is positive too, paying off not in risk-free lending, but better-aligned lending, narrowing the gap between invisible and investable.
The post AI Rewrites Lending for Borrowers FICO Scores Miss appeared first on PYMNTS.com.