The Consumer Financial Protection Bureau has once again made it clear that new AI- and machine learning-based technologies are held to the same consumer protection regulations as established technologies.  

Lenders must test and monitor new technology to ensure that the tools provide correct information to consumers and are compliant with fair lending requirements, John Redding, partner at law firm Alston & Bird, told Bank Automation News’ sister publication Auto Finance News. 

“I don’t think the CFPB is saying you shouldn’t be using these new technologies; what they’re saying is that they understand there is a lot of new technology out there and that people are looking for ways to use technologies to be more efficient and provide consumers information more quickly,” he said. “Before you start using it, you need to make sure that it’s used correctly and be careful when using it.” 

The CFPB on Aug. 12 published a letter to Treasury Secretary Janet Yellen addressing the “rapid adoption of new technologies in the consumer financial marketplace” by third-party technology providers and established financial institutions.  

The bureau made clear that companies using AI and other tech must adhere to laws governing discrimination and unfair practices, and that exceptions will not be made for new technologies. 

CFPB Director Rohit Chopra said in remarks published in November 2023 that “there is no fancy new technology carveout to existing laws.” 

The CFPB in the Aug. 12 letter highlighted several examples in which institutions must maintain compliance as they employ new technologies to make processes more efficient. These include: 

• Automated customer service and technologies built on large language models: The CFPB is monitoring compliance of these tools with the Equal Credit Opportunity Act (ECOA) and urges businesses to ensure that the tools provide the right information to consumers. Businesses must also ensure that any AI, machine learning (ML) or automated decision-making tools do not have an inherent bias.  

• Fraud screening: Companies using AI- and ML-based third-party technology providers for fraud prevention must comply with fair lending requirements. 

• Lending and underwriting decisions: Regardless of how new a technology is, institutions must ensure that credit decisions comply with the ECOA and should monitor for unlawful discrimination, including testing for potential disparate impact violations.  

If the technology provides inaccurate information to consumers, that can put the lender at risk of CFPB violations, Redding said.  

In a stark example of the risks of chatbot technology, an AI-based chatbot used by Chevrolet offered to sell a vehicle to a consumer for $1 in December 2023, according to published reports.  

“AI in underwriting models …  we see the need for care in automated underwriting, whether it’s decisioning or pricing,” Redding said. “Customer service, loss mitigation, if you’re using these tools, you need to be sure that however they’re being used, the information they’re providing is correct. That is the core message that should come through from the bureau.” 

However, one of the complaints by companies is that the CFPB’s rules and guidance are vague. Non-specific rules surrounding AI- and ML-based technologies are in line with the bureau’s broad approach to guidance that may allow it to maintain flexibility in deciding what actions are permitted, Marci Kawski, partner at law firm Husch Blackwell, told AFN. 

“There’s nothing new in this pronouncement; some of the frustration is that the CFPB has not given any clear rules of the road,” she said. “It’s true that the same laws apply, regardless of the technology, so it’s not helpful for the CFPB to keep reiterating that.” 

Editor’s note: This article first appeared on Auto Finance News, a sister publication to Bank Automation News.