ABSTRACT: Artificial Intelligence and Machine Learning are inevitable, and in many ways, beneficial. But lenders are cautioned to implement best practices in order to ensure credit decisions are made fairly and equitably, with or without the use of AI.

AI and Fair Lending Regulations

Last fall, the Consumer Financial Protection Bureau (“CFPB”) cautioned lenders that the use of artificial intelligence (“AI”) does not excuse fair lending regulations. Amid reports of error and potential biases among artificial intelligence systems used to evaluate loan applications, and an uptick in cases in other industries involving the use of AI, a new litigation trend in lending and mortgage servicing could see a rise in 2024.

From Chatbots to Credit Evaluations

AI has permeated almost every industry, and banking is no exception. No doubt those reading this post have likely encountered “chatbots” during online banking sessions or when attempting to contact their financial institution for general customer service inquiries. AI and machine learning can be used by lenders to communicate and resolve customer issues, and it can also be used in the credit application and approval process to evaluate risk and assess various factors such as credit history, income, and spending habits in order to predict loan performance.

Fiction Versus Facts

Under the Equal Credit Opportunity Act (“ECOA”), lenders are prohibited from discriminating on the basis of factors such as race, color, religion, national origin, sex, marital status, age, and receipt of public assistance. When those loans involve the purchase of a home, the Fair Housing Act (“FHA”) also comes into play. Both statutes provide for a private right of action by aggrieved applicants. Additionally, AI “hallucinations,” where fiction is presented as fact, are cited as a substantial risk among surveyed lenders.

AI Risks in Credit Decisions
The CFPB, through Director Chopra, has made it clear that “Creditors must be able to specifically explain their reasons for [loan] denial. There is no special exemption for artificial intelligence.” Furthermore, the explanation cannot just be a blanket response, but instead must provide more detail. For example, the CFPB Bulletin provides that ‘if a creditor decides to lower the limit on a consumer’s credit line based on behavioral spending data, the explanation would likely need to provide more details about the specific negative behaviors that led to the reduction beyond a general reason like ‘purchasing history.’” Thus, lenders must have a comprehensible and above-board explanation for why a loan is denied. Relying on AI or machine earning to simply check a box is insufficient.

Potential AI Litigation in Lending
A warning from the CFPB is one thing, but is litigation on the horizon? Perhaps. While not in the lending sphere, cases began to crop up in late 2023 by class plaintiffs against health insurance providers for negative consequences they allege were the result of use of flawed AI in evaluating physician determinations of “medically necessary” care and denying claims. In a case currently pending before the United States District Court in Minnesota, the plaintiffs claim a 90% error rate in evaluating medical necessity of treatment, which they claim has had a particularly detrimental effect on seniors in need of care. While this case is still in its infancy, it could be a signal of things to come.

Mitigating AI Risks in the Financial Services Sector
AI and machine learning is inevitable, but it is key that lenders implement best practices to avoid liability. Continuous monitoring and diligent testing of behaviors and outcomes of models, creating audit trails, and having human decision-making or appeal processes may help to ensure credit decisions are compliant.

Conclusion

Baker Sterchi will continue to monitor both regulatory and litigation updates pertaining to the use of AI and machine learning in lending and servicing.