What US Banks and Fintech Lenders Disclose About AI: Evaluation, Safeguards and Sourcing
Year: 2026
Type: article
Abstract: Public statements about artificial intelligence (AI) give outsiders uneven information about what an application does, how it is controlled and who supplies it. This study codes the public statements of a purposive frame of 37 US regional banks, large national banks and digital lenders into 194 AI applications. Performance figures accompany 55 applications; 8 report a baseline or control group, and 1 states costs. Oversight or testing is mentioned for 20, and neither internal work nor an external provider is named for 109. Only 24 applications appear in an SEC filing; 138 appear only outside filings. Applications in use (136) and those stating internal work more often state a performance figure, but not a baseline or control group; claim origin and, within organizations, channel show no detectable difference. Of nine comparisons among institution types, only the stating of model class survives correction for multiple testing. None of the US federal securities and consumer credit rules examined requires application-level public disclosure of AI use by these institutions, and most of the retrieved statements appear outside SEC filings. The study argues for a targeted public record for consequential applications and proposes an internal evidence check; both remain untested.
Source: FinTech
Author: Trond Arne Undheim
Institution: Stanford University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,207
OpenAlex ID: W7236795225
