The Consumer Finance Podcast

Exploring the CFPB's Stance on AI in Financial Services

Episode Summary

Chris Willis delves into the Consumer Financial Protection Bureau's (CFPB) recent response to the Treasury Department's request for information on the use of artificial intelligence (AI) in financial services.

Episode Notes

In this episode of The Consumer Finance Podcast, Chris Willis delves into the Consumer Financial Protection Bureau's (CFPB) recent response to the Treasury Department's request for information on the use of artificial intelligence (AI) in financial services. Chris highlights key points from the CFPB's communication, including their views on fraud models, fair lending testing standards, and the implications for financial institutions. He also discusses the CFPB's candid remarks about its own innovation office and the future of regulatory treatment for individual firms. This episode is essential listening for anyone in the financial services industry looking to stay informed about regulatory expectations and compliance challenges.

Episode Transcription

The Consumer Finance Podcast: Exploring the CFPB's Stance on AI in Financial Services
Host: Chris Willis
Date Aired: September 26, 2024

Chris Willis:

Welcome to The Consumer Finance Podcast. I'm Chris Willis, the co-leader of Troutman Pepper's Consumer Financial Services regulatory practice. And today I'm going to be talking about some interesting things in a response that the CFPB submitted to a request for information from the Treasury Department about the use of artificial intelligence in financial services. But before I jump into that topic, let me remind you to visit and subscribe to our blogs TroutmanPepperFinancialServices.com and ConsumerFinancialServicesLawMonitor.com. And don't forget about the bushel of other podcasts that we have. We have the FCRA Focus all about credit reporting, The Crypto Exchange about all things crypto, we have Unauthorized Access, which is our privacy and data security podcast, Payments Pros, all about the payments industry and our newest podcast, Moving the Metal, all about the auto finance industry. All those podcasts are available on all popular podcast platforms. And speaking of those platforms, if you like this podcast, let us know.

Leave us a review on your podcast platform of choice and tell us how we're doing. And if you enjoy reading our blogs and listening to our podcasts, our mobile app is a great way to do it. It's available for both iOS and Android. Just look for Troutman Pepper in your app store and you'll have full access to all of our written and spoken thought leadership content right in one place. Now, as I said today I'm going to be talking about some interesting statements that were made by the CFPB in a recent response to a request for information from the US Department of Treasury about the use of artificial intelligence in consumer financial services. This wasn't a very highly publicized communication from the CFPB, but there are a number of very notable things in it that I wanted to bring to your attention and give you my impression of. The response goes on at some length and starts, of course, with a number of fairly pejorative discussions of the financial services industry.

It asserts, for example, that financial institutions sometimes behave as if there are exceptions to the Federal consumer protection laws for new technologies, which the CFPB has said on numerous occasions, but which I've certainly never found to be the case myself. And the CFPB makes other remarks about how the use of artificial intelligence in various algorithms can be discriminatory, which again, I don't feel is likely to be the case, certainly not any more than any other modeling technology. But besides the bad comments about the industry where the CFPB sets up the need for it to monitor closely the use of this new technology, there are a couple of pretty significant things in the response that I think bear mentioning and I think are important for financial institutions to keep in mind. And they really fall into two categories. One of them is the use of fraud models, and the other one has to do with the proper standards for fair lending testing of other credit models.

And so I'd like to talk to you today about both of those. The CFPB has a paragraph in its response to the Treasury that talks about the use of new technologies to engage in fraud screening. And the CFPB announces that, and I'll just quote here, “it is critical that companies offering these services,” and obviously that's pointed at the vendors who sell these fraud products, “recognize that the consumer financial laws, including the Consumer Financial Protection Act and in appropriate circumstances, the Equal Credit Opportunity Act apply to fraud screening conducted as part of a transaction for a consumer financial product or service.” Now, that statement by the CFPB is an important one because historically, at least before 5 or 10 years ago, we didn't see a lot of attention from a Fair Lending or UDAP standpoint applied to fraud screening by financial institutions. But in recent years, we've seen a lot of that from the CFPB both in supervision and enforcement.

And the RFI serves to underline that and remind everybody in the market that fraud tools, whether they're internally developed or purchased from a vendor, are equally subject to scrutiny by the CFPB, and in fact are being subjected to scrutiny by the CFPB in supervision and in other contexts as well. Anybody listening to me who thinks that fraud is off by itself and untouchable because it's fraud, it's a good reminder to read this RFI and understand that the CFPB is serious when it says it plans to apply both UDAP and Fair Lending to those products. But the other thing, and this piece almost made me fall off my chair when I read it, was the following, and let me just quote it verbatim for you again, "Moreover, because fraud screening is used to assess credit worthiness by determining who gets offered or approved for a financial product, firms that compile and provide such information are typically subject to the requirements of the Fair Credit Reporting Act."

Now, that was a statement that really got my attention because there's tons of fraud products that are in the market that are used for things like identity verification where the provider of that product thinks with very good reason, by the way, that that is not a product that's used to determine eligibility for a credit product, but in fact is used for some other purpose, like for example, identity verification. Those products are sold with the express statement that they cannot be used to determine credit eligibility and that they're therefore not subject to the Fair Credit Reporting Act. And the CFPB seems to just ignore that and say, oh, because it's part of the credit approval process, it's subject to the Fair Credit Reporting Act. I think that's an incorrect statement of the law.

I don't think that's what the FCRA says, but this statement, in addition to just being surprising and I think legally wrong, is also consistent with the very broad reading of the Fair Credit Reporting Act that the CFPB seems to be considering in connection with the rulemaking that was announced last fall at the White House event, the data broker rulemaking, the one that they released the medical debt portion of in June of this year. And the other part dealing with data brokers and credit header data and things like that is supposed to come out at some other point in time. But I do want to warn people in the market that the CFPB has this assumption of fraud screening products being subject to the FCRA that really does not agree with historic practice in the industry, nor in my view the text of their Fair Credit Reporting Act.

But the Bureau may try to apply that in either supervisory exams or enforcement, so everybody needs to be on the lookout for that. Now, the other thing that was notable to me in the response to the Treasury RFI is that the CFPB talks about the proper standards for Fair Lending testing of models. And makes public statements about things that were really known already to people who saw it in CFPB supervision, but were not really the subject of much public comment by the CFPB. Again, I'll read you another sentence from the response and talk about what it means. The CFPB says, "Robust fair lending testing of models should include regular testing for disparate treatment and disparate impact, including searches for and implementation of less discriminatory alternatives using manual or automated techniques. CFPB teams will continue to explore the use of automated debiasing methodologies to produce potential alternative models to institutions underwriting models."

Now, there's several things to unpack there that are important. The first is the requirement in the CFPB's mind that anyone using a model should include as part of its fair lending testing, a search for less discriminatory alternatives. This was something that was first announced in a speech by the then director of Fair Lending at the CFPB, Patrice Ficklin, in April of 2023, but the CFPB had never officially said anything about it. Well, now here we do see it actually in this response to the Treasury RFI. And in fact, we have seen the CFPB looking at this and examining it in the supervisory context for about the past year, I would say. Again, it's something that happened without a lot of public fanfare, but the CFPB does talk about it in the quote that I just read you, and it is very real that they do have that expectation in terms of compliance with the Equal Credit Opportunity Act.

Second thing that I want to highlight in that sentence is it talks about the use of either manual or automated techniques to find a less discriminatory alternative. And there, I think what the CFPB is signaling is it has not settled on any particular method that it wants financial institutions to use to find a less discriminatory alternative, if one exists to a particular model. They're not at the point of prescribing, "Hey, you have to do it X way." They're merely saying you need to do it and leaving flexibility for the industry about how that actually is done. I think that's an important thing for industry to know. But then this final sentence is also important. “CFPB teams will continue to explore the use of automated debiasing methodologies to produce potential alternative models to institutions underwriting models.” The CFPB is saying in this document that it has developed the capability to do less discriminatory alternative searches using a creditor's underwriting model.

And presumably it intends to do this probably as a part of supervisory exams. Maybe in enforcement too, but in supervisory exams is where it's most likely to come up first. The CFPB might find itself in a position of redoing someone's underwriting model and saying, "Hey, we came up with this less discriminatory alternative and you need to use it because otherwise it's a violation of the Equal Credit Opportunity Act." I have not seen them actually do this in supervision yet, but the sentence that I just read to you seems to suggest that they're contemplating doing that. And it'll be very interesting if they do that, if in fact their model is just as predictive from a credit performance standpoint as the one that it supposedly is there to replace as a less discriminatory alternative.

And if the CFPB mandates something like that on a financial institution and the financial institution needs to tune or redo the model, as is typical in the industry every so often, will that put a straitjacket on the creditor in terms of its ability to change the model or even develop a new model in the future? All those are unanswered questions, but I think it's definitely a newsworthy thing that the CFPB may be saying, "We may redevelop your model for you and find you a less discriminatory alternative that you didn't find yourself." I think that's very material for the industry. The final thing that I thought was interesting about the response to the Treasury RFI was the CFPB made some very candid statements about the activities of its own innovation office under the prior administrations. And you may remember that there were a number of instances where the innovation office granted no action letters or other things for particular creditors, sometimes on things that had fair lending sensitivity, like the attributes and models or underwriting strategies or things like that.

One of the things that we observed in this administration is that the CFPB seemed to want to back away from and disown those previous no action letters, and in fact, it terminated one of them during this administration very loudly, I might add. And so it's relatively clear by implication that the CFPB didn't really want to do that anymore because we certainly didn't see any more coming out that were like that. But here in this response to the Treasury RFI, the Bureau speaks very candidly about that. And let me again read you a quote from it, "The CFPB has previously experimented with offering sandboxes and other forms of special regulatory treatment to individual firms to foster innovation, including issuing no action letters. The CFPB has learned important lessons from these programs, which fell short of their intended purpose of encouraging pro-consumer innovation in financial markets." And the CFPB then goes on at length to detail some of its other complaints about how those no action letters and other sandbox approvals played out. But I think it's relatively clear that the CFPB doesn't want to do that anymore.

And in fact, one of the themes running through this letter is we don't think the market is aided by giving special permission to one market participant that all market participants should be on the same footing. And so we're not going to essentially give someone a leg up anymore, presumably through a no-action letter or something like that. Of course, we already knew that there were no no-action letters being issued recently by the CFPB, but here I think you have a very clear statement as to why that is and the fact that it's not likely to change, at least under this administration. Those were some of the nuggets that I thought were important to mention from this RFI response that the CFPB submitted in August of this year.

I think they serve as important reminders or notices to the industry about the CFPB's expectations about model development, model fair lending testing, less discriminatory alternatives. And there's certainly a warning sign there for companies that offer fraud scores to make sure that they are buttoned up in their position about why they think their product is not governed by the Fair Credit Reporting Act, given the broad statement that I shared with you earlier in the podcast, so that's something for everybody to look out for. I appreciate you listening to today's episode. Don't forget to visit and subscribe to our blogs TroutmanPepperFinancialServices.com and ConsumerFinancialServicesLawMonitor.com. And while you're at it, why not head over to the Troutman website Troutman.com and add yourself to our Consumer Financial Services email list.

That way we can send you copies of our industry-only webinar invites, as well as the alerts and advisories that we send out from time to time. And as I said before, don't forget to check out our mobile app. Just type in Troutman Pepper in your app store and give it a try. And of course, stay tuned for a great new episode of this podcast every Thursday afternoon. Thank you all for listening.

Copyright, Troutman Pepper Hamilton Sanders LLP. These recorded materials are designed for educational purposes only. This podcast is not legal advice and does not create an attorney-client relationship. The views and opinions expressed in this podcast are solely those of the individual participants. Troutman Pepper does not make any representations or warranties, express or implied, regarding the contents of this podcast. Information on previous case results does not guarantee a similar future result. Users of this podcast may save and use the podcast only for personal or other non-commercial, educational purposes. No other use, including, without limitation, reproduction, retransmission or editing of this podcast may be made without the prior written permission of Troutman Pepper. If you have any questions, please contact us at troutman.com.