← Back to incidents
AI-Powered Loan Servicing Algorithm Miscalculated Mortgage Payments Affecting Thousands
HighAI-powered mortgage servicing systems at major banks miscalculated payments for over 12,000 homeowners, causing escrow shortages and wrongful fees. The CFPB launched investigations while class action lawsuits were filed seeking damages.
Category
Financial Error
Industry
Finance
Status
Under Investigation
Date Occurred
Jan 15, 2025
Date Reported
Feb 10, 2025
Jurisdiction
US
AI Provider
Other/Unknown
Application Type
api integration
Harm Type
financial
Estimated Cost
$85,000,000
People Affected
12,500
Human Review in Place
No
Litigation Filed
Yes
Litigation Status
pending
Regulatory Body
Consumer Financial Protection Bureau
mortgage_servicingescrow_calculationcfpbbanking_aifinancial_algorithmsconsumer_protectionclass_actionforeclosure
Full Description
In January 2025, multiple major mortgage servicing companies began receiving complaints about incorrect payment calculations from their AI-powered loan servicing platforms. The errors primarily affected escrow calculations, where the AI systems failed to properly incorporate updated property tax assessments and changing insurance premiums into monthly payment calculations. Wells Fargo, JPMorgan Chase, and Rocket Mortgage were among the institutions affected, with their automated systems serving millions of mortgage accounts.
The technical failures manifested in several ways. First, the AI algorithms incorrectly processed property tax reassessments, often using outdated values or failing to account for local tax rate changes. Second, when homeowners' insurance premiums increased due to climate-related risks, the systems frequently undercalculated the required escrow contributions. Third, for adjustable-rate mortgages, the AI systems misapplied interest rate adjustments during a period of Federal Reserve policy changes, leading to payment amounts that were sometimes hundreds of dollars off the correct calculation.
The impact on affected homeowners was severe and immediate. Many received notices of escrow shortages requiring large lump-sum payments to bring accounts current. Others faced wrongful late fees when their regular payments, calculated by the same faulty systems, were insufficient to cover actual amounts due. In approximately 800 cases, homeowners received foreclosure warning notices despite making payments they believed were correct according to their servicer's statements.
The Consumer Financial Protection Bureau received over 3,000 complaints within three weeks of the first reports, prompting an immediate investigation. CFPB Director Rohit Chopra stated that the bureau was examining whether the mortgage servicers had adequate oversight of their AI systems and whether they violated federal servicing regulations. Simultaneously, class action lawsuits were filed in federal courts in California, New York, and Florida, seeking damages for affected borrowers and demanding changes to AI oversight practices.
The financial impact extended beyond individual homeowners to the mortgage servicing industry itself. The three major servicers had to establish dedicated customer service teams to handle the flood of complaints and begin manual recalculation of affected accounts. Industry analysts estimated the total cost of remediation, including refunded fees, corrected payments, and operational expenses, could exceed $85 million across all affected servicers.
Root Cause
AI algorithms failed to properly account for changing property tax assessments and insurance premium updates when recalculating escrow amounts, resulting in systematic undercalculation of required monthly payments. The systems also incorrectly applied interest rate adjustments for adjustable-rate mortgages during volatile market conditions.
Mitigation Analysis
Implementation of mandatory human review for escrow recalculations above certain thresholds, real-time validation against external property tax and insurance databases, and staged deployment with shadow testing could have detected these errors. Enhanced model monitoring with statistical anomaly detection would have flagged the systematic nature of the miscalculations before widespread impact.
Lessons Learned
This incident highlights the critical need for robust validation and human oversight in AI systems handling high-stakes financial calculations. The mortgage servicing industry's heavy reliance on automation without adequate safeguards created systemic risk affecting thousands of consumers simultaneously.
Sources
CFPB Probes AI Mortgage Servicing Errors Affecting Thousands
Wall Street Journal · Feb 10, 2025 · news
Major Mortgage Servicers Face Class Action Suits Over AI Payment Errors
Reuters · Feb 8, 2025 · news