Industry & Regulatory Update: Where Compliance Risk Is Heading: Part 2
As federal consumer finance enforcement pulls back, state regulators are stepping in. In final episode of this two-part episode of COMPLY, Bryce McNitt, director of state government affairs at PayActiv and former chief of staff at the CFPB’s Office of Markets, joins PerformLine’s Ashley Cianci. They discuss how states are filling the gap and what that means for compliance teams planning for 2027.
Key Takeaways
- The three types of state AI laws emerging, and how Colorado’s automated decision-making technology (ADMT) law applies to financial services
- The open question of liability when an LLM steers a consumer toward a product
- An early theory of liability for agentic AI from California’s Rohit Chopra
- What the Nov. 3 midterm results could mean for state and federal regulators
- What to expect from the CFPB, and why look-back periods should be on every compliance leader’s radar
- Audience Q&A on buy now, pay later, AI steering in mortgage and what CCOs should document now
Show Notes:
- Listen to Part 1: https://performline.com/blog-post/industry-regulatory-update-where-compliance-risk-is-heading-part-1/
- Access the full webinar: https://events.performline.com/industry-regulatory-update-for-compliance-risk
- Connect with Bryce McNitt on LinkedIn: https://www.linkedin.com/in/brycemcnitt/
- Connect with Ashley Cianci on LinkedIn: https://www.linkedin.com/in/ashley-cianci/
- Subscribe to PerformLine to stay connected to resources and updates: https://lp.performline.com/subscribe-to-performline
Subscribe to COMPLY: The Marketing Compliance Podcast
About COMPLY: The Marketing Compliance Podcast
The state of marketing compliance and regulation is evolving faster than ever. On the COMPLY Podcast, we sit down with the biggest names in marketing, compliance, regulations, and innovation as they share their playbooks to help you take your compliance practice to the next level.
This podcast is for informational and educational purposes only and does not constitute legal advice; consult your own compliance or legal counsel.
Episode Transcript:
Ashley Cianci:
Welcome back, COMPLY Podcast listeners. I’m your host, Ashley Cianci. This is part two of our conversation with Bryce McNitt, former chief of staff for the CFPB’s Office of Markets. In part one, Bryce walked through how enforcement is shifting from Washington to the states. And if you missed it, go back and start there. Special note that this webinar was recorded on Nov. 10, so just about a week after the midterm elections took place. Today we’re getting into AI. The federal government has stayed out of it by design, and the states aren’t waiting.
Bryce breaks down the three types of AI laws emerging across the country, the open question of who’s liable when AI steers a consumer, and an early theory of liability for agentic AI. And then he looks ahead: what the midterms mean for state and federal regulators, what to expect from the CFPB, and why look-back periods should be on every compliance leader’s mind. We’ll close with audience questions on buy now, pay later, AI steering and what CCOs should be documenting now. Let’s pick it back up with Bryce.
Bryce McNitt:
OK, so I’m gonna switch gears a little bit and talk about regulation of AI. Obviously it’s the gorilla in the economy, it’s the gorilla in the room for regulation and financial services. We all know federal government has not done anything here yet, and that is by design. The president’s been very vocal saying that he doesn’t want any regulation at the federal level or the state level for AI until it’s fully built out.
But I think there was a task force created recently, and there’s supposed to be a report issued in the next couple of months, but would be surprised if too much comes from that report, if or when it’s issued. And so the states have not hesitated to pick up the slack here, and even red states, you know, Utah and Texas, are big leaders in AI laws being passed.
And so it’s this great example of federalism at its best, where you have lots of different experiments running in these petri dishes around the country simultaneously, and we’ll kind of see where things go. So there’s a few buckets of laws that have begun to be passed, and I’ll talk about a few of them here briefly. And I’ll say that there’s a lot of overlap between these. There’s these are not mutually exclusive categories.
There will be, I think eventually over time they’ll kind of morph into one roughly similar style of regulation across the country. But the first is systemic and consequential risk. The best example of this type of law is in Colorado. They passed their automated decision-making technology law. They passed one actually in 2024, I believe, but it was overly broad and it was probably a little bit ahead of itself in terms of really knowing what AI was going to be. And so re-they tailored the rule more, and it’s going to come into enforcement in early 2027, but I believe. So the law defines covered ADMT as technology that materially influences consequential decisions, which are those affecting a consumer’s access to education, employment, financial services, insurance, health care and essential government services.
The law puts obligations on both developers and deployers. Deployers here are the businesses utilizing large language models for the categories just mentioned above. And the obligations include providing clear disclosures to consumers at the point of interaction with a covered ADMT, disclosing the technology’s role to consumers within 30 days after an adverse outcome, and offering consumers the right to correction and meaningful human review.
The second category of law is use case and consumer protection regulation. This regulates specific uses of AI rather than AI generally, and focuses so the two of the biggest categories are health care and government so far in some of the states. Utah is the best example of this. It really focuses on user safety, child safety in particular, and heavily disclosure based, as I mentioned.
The last category is authenticity and transparency regulation. This is very broad based. It protects against things like deepfakes and scams. Again, a big emphasis on child protection. It’s very straightforward. I won’t spend additional time on that category.
So drilling down a little bit more, what does this mean for financial services? So again, in Colorado, it specifically mentions financial services as a category for its ADMT law. And so under that law, financial institutions would be considered a deployer because they’re implementing an LLM tool for consumer loan approvals, risk pricing and other heavily regulated activities.
And I would say that because these activities are already heavily regulated for banks and fintechs, I don’t think that complying with this should be too heavy of a lift. Originators have to issue adverse action outcome notices already required by law. And so, you know, issuing a companion notice under an ADMT-style law and many different states shouldn’t be too big of a leap, I think.
I think, though, one thing that’s missing from these laws so far is, what about the consumer use of AI? So what are the liabilities and what are the rules of the road for when a consumer uses an LLM? So an example would be when a consumer could be misled toward a specific financial product because the provider paid the LLM for priority, or when an LLM leads a consumer into a product flow that utilizes dark patterns or digital interference, to borrow a regulatory term. Someone is kind of sort of trapped or guided, steered into a product that is not good for them and creates some kind of harm. So who’s liable there? So we don’t have any laws passed that address this right now, or any regulations specific to AI that I’m aware of, although it’s changing kind of by the day, it feels like.
But I would say that you should keep the Biden-era abusiveness statement issued by Rohit Chopra’s CFPB in 2023. You should keep that close. You should read it at night before bed and keep it on your nightstand, because it is a really well-reasoned and very detailed overview of abusiveness, and I think it has been a guide for other regulators at the state level and will continue to be. It’s probably the most restrictive interpretation of abusiveness that you’re gonna find, but because things are likely to change, you know, quickly in the future, it’s good to keep that one close by.
So even further out on the frontier of regulation is agentic AI. And we are very early in these stages. We’re really just in the legal theory stage of regulating agentic AI. You know, I think we’ve been hearing for the last 18 to 24 months that agentic AI is the future and it’s gonna kind of be take over everything. But at least for me, it wasn’t really until Meta released its Muse, it’s an app-based Muse agent product that it really started to go into the mainstream. And you started to hear about banks like JPMorgan Chase and retailers like Amazon banning agents from their sites. And so that’s kind of bringing all these issues to a head here in terms of, like, who’s liable for the use of agents. And we don’t really have any rules on the books that cover this.
So I wouldn’t have included this at all, except that Rohit Chopra, who’s now the Secretary of Business and Consumer Services Agency in California, has just recently shared his theory of what this liability might look like. And I know some of the big financial services trade associations have worked on this too. Consumer Bankers Association had a really nice white paper that kind of evaluated how agentic liability might play out in, like, Reg E for payments.
And so anyway, Mr. Chopra, his thesis here was essentially that if an agent strays from consumer instructions, so like a rogue agent, then the developer would be liable. And then if a financial institution accepts instructions that don’t reasonably look like the consumer’s, the financial institution would be liable. I think the first one’s pretty clear. We can all agree that if you have a technology that goes rogue on you, then that might be the fault of the creator of that technology. But the second one is much fuzzier, and it’ll be interesting to see how this plays out, because it almost seems to create, like, a fiduciary-style obligation on the part of a financial institution to use their best guess of whether or not this agent that is trying to make a transaction on their website or on their product stack is acting in the consumer’s best interest. I would say this is just a theory. And so it’s certainly not the final word, but it’s a very interesting conversation and certainly something for a nerd like me to watch closely over the next few years.
And so I would also look for more theories to be developed from influential states like Utah and Texas. You know, Texas was a red state for almost my whole lifetime, but now it’s a purple state. So it’ll be very interesting, just because they’re, you know, the second-biggest economy in the United States and very influential. And also, I think it’s interesting here, and I might talk about this again when we get to the midterms, is that federal agencies, you know, they haven’t done anything in AI, they don’t want to, but if the momentum really picks up at the state level with the increased number of statehouses that are blue, federal agencies might have to jump into the fray to try to assert some sort of preemption or just establish some sort of guiding theory that’s overarching. And so we’ll see how that plays out as well.
All right, so the midterms happened last week. We all stayed up late watching. It was very exciting. It was very interesting for me. I follow elections, you know, as closely as everyone else, and I’ve always felt like the actual results end up being a few points redder than the polls suggested. And this is the first time I’ve seen the results just blow out to the blue side, like, dramatically. So that’s what was very interesting.
But, and we all know, Democrats swept the House and the Senate federally, and that’ll create a lot of big changes at the federal level. At the state level, there was a lot of change as well. So six governorships flipped blue. Democrats will now have, I think, 30 states. So this is a pretty big deal. Thirty, 30 is a lot. And you can see, you know, we talked earlier about states banding together. That’s a very big, influential bloc of states.
Especially with Texas being having a blue a Democratic governor now, Texas and California and New York, those three states are a huge portion of the U.S. economy. Four, probably five state AGs also flip blue in Texas, Kansas, Iowa, Georgia and probably Ohio. I don’t think that’s been called yet. So big, very influential red states for a long time, now being led by Democratic AGs.
At the state legislative level, we had 12 state chambers turn blue and none of them turn red. And I believe five states gained trifectas, as they’re called, governor, house or assembly, and senate, all blue, and Republicans lost five.
And 43 of 99 chamber leadership positions also changed hands. So it’s just a massive amount of change happening in one election, and there will be a lot of change coming here, I think, next year. A lot more states that were purple will now have an ability to pass legislation much more easily. So I would think the trend of, you know, consumer protection at the state level will just increase. I don’t see it slowing down.
So if there’s anything sort of, like, interesting here, I mean, there’s a tension, obviously, with now the president’s administration, the executive branch, being sort of isolated now among other centers of power. But there has been this growing trend of bipartisan state action. And my hope would be, and I like to be an optimist, so my hope is that this election will increase that trend, and that sort of red state leaders will kind of understand that consumer protection and affordability issues really are a big deal to voters and they need to pay attention to it. And so this example of the Block case with 46 states participating, and then another example recently of 39 states plus D.C. urging the Supreme Court to get involved and intervene in prediction market litigation, with the aim of trying to establish some sort of stronger regulation around this market, which has just exploded in size and popularity recently. I think these trends could point to more cooperation at the state level, potentially, but, you know, you never know.
And also note here, too, at the midterms. I mentioned junk fees earlier coming back at the state level. Maxine Waters is running for the head of House Financial Services Committee. Again, I think she has been the head of that in the past. I don’t know if she’ll win or not. I don’t think that vote has happened yet, but she issued a memo last week with her top five priorities, and junk fees was one of them, as well as shoring up the CFPB and ensuring that there’s no further attempts to defund it. So very interesting, like, immediate consequences for some of the trends we’ve been discussing here, and then sort of the reestablishment of junk fees and affordability at the federal level again. So I think that kind of foreshadows what we’ll see here in the next two years.
What will happen with CFPB? Brian Johnson, I think, will be confirmed this month. I don’t know for sure. If he’s not confirmed this month, then he might not be confirmed at all. So I think there’s probably going to be a push to get him in there. What will his tenure look like? It could be a very short tenure. He might only have two years there if the party of the president changes hands in 2029. And so, you know, passing final rules takes a really long time. And if they aren’t in effect before the next person comes in and they they’re never gonna go into effect. That’s what we saw with open banking, the 1033 rule. It never got over the finish line. It got finalized, but it never went into effect, and then it was withdrawn. So will he be able to make any long-lasting impact? Will he try to address AI in financial services for fintechs? I would say that’s a hard thing to pass out that opportunity, but we’ll see.
And so I think for industry, you need to watch what Brian Johnson does with his time, but also think about, I would plan for reverting back to a Chopra-style CFPB in 2029, because it’s just, you know, at least a 50-50 chance that that happens. And then looking to the next administration and the pace of change again, kind of this theme of, like, things changing faster than they used to.
The Trump v. Slaughter decision, which allows the president, any president, to fire prudential regulator board members at will, means that these prudentials that used to move at a glacial pace… I remember working with OCC counterparts during my time at OCC, and we asked them a question about changing something, and they said, “Well, we move, we move glacially.” CFPB is not quite like that. And so that era could be kind of coming to an end, which was a very, very long era. I mean, stretching back many, many decades. If leadership is just gonna be one party for these regulators, it’s gonna create more uncertainty, and I mean that things are gonna change much more quickly.
Another thing to watch here over the next two years is, you know, we’ve had this huge era of innovation. Tons of new bank charters being issued for national banks, industrial banks, narrow banks or trust banks for crypto products. That’s really change the landscape, and these businesses are just kind of getting established or, in their with their new charters, or how they’re gonna operate is just getting established. Do we have two more years of that? The window gonna stay open, or is it gonna close under increased scrutiny from the House and Senate? And then what will happen after 2029? I mean, I think there’s a decent chance here, again, maybe 50-50 at least, that everything we get in terms of new charters and everything, and new businesses operating in the U.S., will just close at 2029. And then that’s the environment we have for quite a period of time, because it might just be too difficult to open the floodgates again.
So it’s a really critical time. And I think for businesses, have to make a risk calculation and how much innovation and growth do you want to pursue now against what the consequences might be under a rapidly increasing enforcement and regulation in 2029. And so that kind of… To wrap it all up is think about look-back periods, you know, 2027 look-back periods are at least two years for disparate impact and ECOA, Equal Credit Opportunity Act, you know, fair lending. It’s even longer, four, five, six years. And so everything happening right now is gonna be up for review in 2029. And so it’s a balance here. It’s a very interesting time.
That is what I’ve got today. Again, thank you so much, Ashley. Really exciting times, and it’s been fun for me to be kind of living through it. You know, I went through the whole transition at CFPB, and now I’m out in industry and still having fun and watching all this. And I think it’ll be interesting to see what happens the next couple of years here as well.
Ashley Cianci:
Awesome. Thank you so much, Bryce. I think Bryce has this ability to make all of these one very like meaty conversations, so fun and engaging, but then also so informative. So thank you so much, Bryce. Awesome. We do have some minutes left, so we are gonna take some questions. I’m gonna start with one, Bryce, that was submitted from one of our registrants during the registration process. So it’s about buy now, pay later. Support is growing for more regulatory oversight of buy now, pay later services. What is the CFPB’s timeline for deciding on a course of action, and how should companies prepare in the meantime?
Bryce McNitt:
Yeah, it’s a great question. My take, my read on where the CFPB is on BNPL is that I wouldn’t expect much on that in the next couple of years in terms of actual regulation. There might be more reports issued about its usage and growth. It obviously continues to grow very rapidly. But unless there is, I think, really substantial evidence of widespread, like, tangible harm… So Russell Vought talked about this. He’s only gonna go after, like, really clear instances of consumer harm, tangible consumer harm. And so if that’s gonna continue to be the ethos under Brian Johnson, which I would assume that it would be, then I wouldn’t expect anything in terms of regulation in the next couple of years from CFPB.
But the states have started to do more here. So New York passed the BNPL regulation, the first in the nation. And then Illinois followed in 2026. Early, I think this summer they passed a law. And so it’s a licensing regime and some regulation. But I would not be surprised if you saw buy now, pay later regulation in more states over the next couple of years. So it’ll continue to work on that channel, but I would be surprised the CFPB does anything right now.
Ashley Cianci:
Makes sense. All right, we’ve got two questions, both pertaining to AI, so I want to take one that just came in. Will there be any liability for steering by AI as it pertains specifically to the mortgage firms?
Bryce McNitt:
Yeah, I mean, I think that’s, it could be steering in for any financial services product. I’m not an expert in mortgage, but I don’t see that it’s, would be really different for any product. And I think again, like I said, you could look at UDAP precedent, you could look at, you know, the abusiveness statement goes through things like digital interference that guides people or makes it very difficult for people to understand what they’re doing.
It talks about the ability for the consumer to understand what the product is actually doing and make an educated choice about what the product offers. It talks about the reasonable reliance on a third party to act on the consumer’s behalf. So that might be, like, a marketplace-style product. So it could be an LLM, like, recommend a mortgage to me, and the consumer’s probably wanting the best rate, right? And so if the LLM guides them to something where they’re getting a kickback or a, you know, affiliate marketing lead fee and it’s not the best product, then yeah, you’re probably square within the sights there of a regulator or an enforcement an AG in that case. And I think that’s universal across all products.
Ashley Cianci:
Makes sense. And you mentioned that no state has clearly addressed consumers using AI yet. And until one does, what is your advice on how a compliance team decides whether an AI-driven customer experience creates UDAP or fair lending risk?
Bryce McNitt:
Yeah, I mean, I would just think really carefully, again, about, yeah, UDAAP. It’s the same principles, and they could be used right now by a regulator or an AG in a state. And so I would think about those UDAAP principles you already use in your marketing materials and sales. We think really carefully about how the consumer’s data is being used, and watch the space of use of consumer data.
I think the states are a little bit slower on FCRA-style laws and regulations to grapple with the consumer data market. Rohit Chopra tried to do that at CFPB with his full. It never got over the finish line. And I know that the current CFPB has been very adamant that federal preemption is very strong here. I think there’s actually more, like, statutory language in statutes saying there’s more preemption. So I think that’s made it harder for states. But yeah, think about the uses of the consumer data, and then think about those abusiveness and deceptive and unfair practices principles.
Ashley Cianci:
Super helpful. All right, I think this might be our last one. This is one that we keep coming back to kind of in our conversations, Bryce, but you’ve warned the decisions made during this kind of deregulatory window obviously could be reviewed later by a different set of regulators in that look-back period. And I think that comes even more apparent with the midterms that happened last week. So what are one or two things you would tell a CCO to document now so that they are protected if that time comes in 2029?
Bryce McNitt:
Yeah, I think explaining the rationale for product design and sort of the best understanding… I think that what’s interesting about AI is that how it actually works is still a little bit of a mystery to folks. So documenting what the product design was intended to be and what the understanding of the company was when they deployed it would potentially be helpful if the actual implementation of the product, through, like, especially through AI channels, does not result in the outcomes the company had intended. But yeah, I think that’s a really good question. It’s hard, it’s hard to say, because we just still don’t know what the rules will be. So it’s a risk tolerance question. And I’d say anyone playing in this space would should be well prepared to document as much as possible when someone knocks at the door here in a couple of years.
Ashley Cianci:
Absolutely. Just something to keep in mind. Keep at the forefront in the meantime. Awesome. All right, well, thank you, Bryce, so much. That was so incredibly helpful. I think it was exactly the conversation that the compliance teams that have joined us today needed, kind of heading into 2027. Thank you all who’ve joined us today and sent in questions.
So like I said, if you want to see how PerformLine can help compliance teams stay ahead of kind of these shifting state requirements and possibly a look-back period, you can also reply to the email that we will be sending, or you can visit performline.com, and you can book time with our team to talk more.
Thank you, Bryce. Again, we so much appreciate it, and we hope everyone has a great day.
Bryce McNitt:
Thank you, Ashley. Thanks, everyone.
Ashley Cianci:
And that wraps our two-part conversation with Bryce McNitt, again, which was recorded on Nov. 10, 2026. A huge thank-you to Bryce for sharing his perspective that I think is so valuable, especially given the current environment. If you remember one thing, make it this: The rules are being written right now, state by state, and the decisions you make today will be judged by whoever is in charge in 2029. Document your reasoning, watch the states, and don’t wait for Washington to tell you what’s coming.
Watch the full webinar on demand at performline.com. If you wanna see how PerformLine helps compliance teams stay ahead of shifting state requirements across web, social, email and AI responses, book time with our team at performline.com as well. Thank you guys so much for listening to the COMPLY Podcast, and we’ll see you next time.
You can follow PerformLine on LinkedIn for news and content in between episodes. And if this episode sparks something, a question, a disagreement, a topic you want us to dig into. Reach out at performline.com, or you can connect with me personally on LinkedIn.