InformedIQ, a developer of AI-driven software for income and employment verification, has unveiled the results of a survey involving over 2,500 auto finance professionals. Conducted in January, the survey reveals an alarming trend: as fraud levels soar, lenders are increasingly hesitant to employ generic AI solutions, primarily due to concerns over data hallucinations and the lack of proven results.
The online survey, targeting director-level executives and above, indicates a growing sentiment of “AI and fraud fatigue” within the industry. The initial excitement surrounding AI’s potential has shifted towards a pressing demand for reliable, high-fidelity solutions capable of addressing increasingly sophisticated threats. “Lenders are no longer looking for futuristic promises; they are seeking immediate, tangible solutions to an escalating fraud crisis,” stated Jessica Gonzalez, VP of customer success and general manager of automotive at InformedIQ. “The data show a clear disconnect: fraud is becoming more sophisticated—powered by Generative AI—yet the majority of the industry is still relying on manual reviews that are slow, costly, and prone to error.”
The survey highlights the staggering impact of modern fraud, with over half of the lenders attributing between 10% and 19% of their total annual loan losses or charge-offs to documentary-based fraud, including false pay stubs and identity manipulation. This crisis is accelerating; nearly two-thirds of respondents noted that identified fraud increased by 5% to 25% over the past year, with an additional 15.5% witnessing a surge of a quarter percent or greater.
Operationally, the reliance on manual verification processes has created significant friction. Approximately 58% of lenders estimate the cost of manually reviewing a single flagged loan file to be between $50 and $100. This cumbersome approach has led to funding delays, with over half of respondents (55%) reporting that manual stipulation reviews add an additional 16 to 30 minutes to the funding process. The survey reveals that if lenders could achieve 99% confidence in automated verification, 38% believe they could redeploy more than half of their current underwriting staff to higher-value tasks.
Despite the push toward automation, confidence in existing technologies remains low. About 55% of lenders describe themselves as only “Slightly Confident” in their ability to identify sophisticated counterfeit documents generated by Generative AI or deepfakes. The InformedIQ survey also shed light on a critical barrier to AI adoption: 52% of lenders cite data hallucinations—plausible but inaccurate data produced by off-the-shelf large language models—as their foremost concern. Compounding this issue, a significant blind spot exists, as 60% of organizations rarely verify historical data to determine whether a document has been reused across different applications or lenders.
Looking toward 2026, regulatory anxiety looms large. A notable 39% of respondents anticipate stricter state-level enforcement, while 33% expect increased scrutiny from federal agencies, including the CFPB and FTC. Beyond auto lending, lenders foresee heightened scrutiny regarding document rigor in mortgage and home equity loans (32%) and personal or unsecured loans (23%).
The survey indicates a future direction for the industry, with lenders planning to prioritize AI investments in credit risk modeling (43%) and fraud detection and prevention (24%). The message is clear: the industry is ready for AI-driven modernization, but only from partners who can demonstrate accuracy, compliance alignment, and resilience against fraudulent data.
See also
Finance Ministry Alerts Public to Fake AI Video Featuring Adviser Salehuddin Ahmed
Bajaj Finance Launches 200K AI-Generated Ads with Bollywood Celebrities’ Digital Rights
Traders Seek Credit Protection as Oracle’s Bond Derivatives Costs Double Since September
BiyaPay Reveals Strategic Upgrade to Enhance Digital Finance Platform for Global Users
MVGX Tech Launches AI-Powered Green Supply Chain Finance System at SFF 2025
