MISMO’s Brian Vieaux on AI governance and state regulation

HousingWire

On today’s episode, Editor in Chief Sarah Wheeler talks with MISMO President Brian Vieaux about the organization’s work on AI governance, state regulatory audits, credit score models and more.  Related to this episode:  MISMO board adds VantageScore, Guild and FICO leaders amid credit modernization push https://www.housingwire.com/articles/mismo-adds-vantagescore-guild-fico-executives-to-board/  HousingWire | YouTube⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ https://www.youtube.com/channel/UCXDD_3y3LvU60vac7eki-6Q  Buy one, get one FREE tickets to the Mortgage Banking Summit on October 1st https://events.housingwire.com/mortgage-banking-summit-2026  More info about HousingWire https://lnk.bio/housingwire  Want more from Sarah? Don’t forget to subscribe!  The HousingWire Daily podcast brings the full picture of the most compelling stories in the housing market reported across HousingWire. Each morning, listen to editor in chief Sarah Wheeler talk to leading industry voices and get a deeper look behind the scenes of the top mortgage and real estate.

Transcript

00:10Welcome, everyone. My guest today is MISMO President Brian Vue to talk about the organization's

00:16work on AI governance, state regulatory audits, credit score models, and more. I'm excited

00:22to jump in. Brian, welcome back to the podcast.

00:25Sarah, I'm excited to be here.

00:27Excited to have you here. We were talking before we started recording, and wow, are you guys

00:31busy. You have a lot going on.

00:33Yeah, I'm on. Literally, tomorrow is the completion of month 11 for me, so I'm beginning my final

00:40stretch into the first year in the role at MISMO.

00:43You know, it's interesting because if you were to describe this to someone who has no idea

00:47what MISMO is or whatever, it's like a standards organization. It doesn't sound that exciting,

00:52but actually, you guys are working on some really interesting things, which I'd love to

00:55dive into today. Let's start with AI governance, right? Got to be the big topic of the day,

01:02AI, and how are you trying to help mortgage lenders as they're trying to figure this out?

01:08Yeah, great question. It is literally the thing that I spend probably the most time on right

01:14now. So back in the spring, MISMO published this product called FRAME, F-R-A-M-E. It's an acronym

01:24for Framework for Responsible Use of AI in the Mortgage Ecosystem. So much easier to say FRAME,

01:31just like it's easier to say MISMO. FRAME came to MISMO as a work suggestion from the Mortgage Bankers

01:40Association's Residential Board of Governors, MBA Resbog. And the MBA Resbog chair for fiscal 26,

01:50Dan Sugg, made it his top priority or one of his top priorities for Resbog to deliver to members,

01:58so principally MBA members. But this is where MISMO is not just an MBA member-only organization. We are for

02:05the entire universe of mortgage lending in the United States. And so Resbog brought this work

02:11idea to MISMO to create, call it a set of guardrails. It was loosely defined at the beginning,

02:18right? And this is the beauty of the MISMO community. I'm not, like, I was not the guy that

02:24anybody came to to say, Brian, go create an AI governance framework. No, no, no, no. We have

02:29smart people across the industry that come together in our community and we leverage their experience

02:36and expertise. And FRAME is a great example of that. So what we published in the spring of this

02:41year is version one of FRAME, which is essentially it's a toolkit for lenders and lenders of all sizes

02:49can use it. I'll be candid and tell you that when we developed this, we really thought it was going

02:54to

02:54be ideal for the mid to small size mortgage lender. And I'll say this, and I say this a lot,

03:00it's the organizations that may not have that single throat to choke as it relates to risk

03:05governance. And they certainly don't have it for AI governance. And so what FRAME is, is a,

03:11it's a toolkit with three artifacts in version one. One is a templated policy document. It's like

03:1810, 12, 14 pages. It's meant to be editable. So yeah, not edible, but editable. And lenders can

03:27download it, modify it for their business practice, and then publish it as their AI governance policy.

03:35And then the other two, two artifacts tie together with the policy document. The one is an AI use case

03:42inventory tool. And it's just what it sounds like. It's a place where lenders can collect and document

03:50all of the AI use cases in their business, both those that they use directly, the AI use cases that

03:57they're consuming indirectly through vendors and tech platforms, et cetera. And we even go as far as to

04:03suggest that lenders, you know, give their employees an amnesty day and bring all the shadow AI use cases

04:10to the forefront and get those documented. And the idea here, Sarah, is nobody wants to be the lender

04:17when, not if, but when Fannie or Freddie or the state regulator comes knocking and says, Sarah,

04:25can you tell us where you're using AI in your business? You don't want to be in a position that

04:29says, let me get back to you in 90 days. So, you know, the, the frame inventory arms lenders with

04:35the

04:35ability today to document all those use cases. And then the third tool in the toolkit is a risk

04:42assessment tool that is meant to be applied to each of those use cases. So if I've identified 50 AI

04:50use

04:50cases, presumably I'll do a risk assessment on each of those use cases. And then based on the tiering of

04:56the risk, high, medium, low, I'll have, you know, the, the requisite mediation if needed to mitigate

05:03those risks. That's in a nutshell, what we've developed is this I'll call it a DIY, but it's much

05:10more, it's, it's more than that, but it's, it's, it's the ability for the average lender, the small to

05:15average size lender to, you know, spend a few days with this toolkit and they can literally stand up

05:22an operational AI governance practice within their organization.

05:27I think that what you just talked about is a great illustration of MISMO, what, what you do,

05:33and also the strength of MISMO. So first off, it's not just lenders in this community, it is about

05:38mortgage lending, but you have so many stakeholders. So you have a lot of different, I mean, you have

05:43all these committees, you have all these working groups, you have these, you know, different, what do

05:48you call them, action chairs or something like that. Communities of practice, yeah.

05:50Yeah, there's just so many people involved here, coming up with something, and then it is, you know,

05:57helpful for people throughout the industry, specifically, like you said, small and midsize

06:02and small lenders who may not have their, you know, dedicated person. So I guess one of my, one of

06:07my

06:08questions is, if someone is listening to this, they don't know that much about MISMO, like how, how does

06:13the government look at MISMO standards? So like, is the MISMO standard, does it become the de facto

06:19regulatory standard? Or as you know, if regulators come, if auditors come, if they see that, and they

06:25see that someone is, is complying with MISMO, that means something to them, like, hey, this is a

06:29company that's really trying. Great. That's a great question. And there's, there's a couple paths

06:33there. I'll start by, by, by explaining that. So there's somewhere in the magnitude of 100 plus

06:41data specifications and or standards, you could call a data spec, a standard, I guess, that have

06:49been created over the 26 plus years, now 27 years that MISMO has been in existence. There are maybe

06:56a dozen, 13, maybe 15 at the most of those data specs that roll up to the Uniform Mortgage Data

07:06Program, UMDP, which is the mandated standard created by Fannie and Freddie for loan delivery.

07:16So I don't, I think it was about a decade ago, the FHFA, the Fannie and Freddie's regulator mandated

07:22that they come together to create a more fungible asset. And part of that process was to ensure that

07:31the same data used for loan delivery to Fannie could execute a loan sale to Freddie Mac and vice

07:37versa. MISMO, the MISMO reference model, you'll hear us talk about the reference model. That's

07:43essentially the architecture that houses all of the data specifications. The MISMO reference model is

07:50where Fannie and Freddie house their dozen or so data standards. And so the first part of your

07:58question was, are there standards that are mandated? Yes, that's an example of mandated

08:04standards. If you want to sell a loan to Fannie or Freddie, you must follow their UMDP, which is part

08:10of the MISMO reference model. Beyond that, most other standards are like, I'll call them more best

08:16practice type standards, things like a standard for loan boarding, right? So there's been a standard

08:23developed that's adopted at a different scale for loan boarding. So taking a loan from origination and

08:30boarding it to loan servicing as an example. And then there's an emerging standard that we're really

08:38excited about that we believe over time will be mandated by state regulators. And it's the mortgage

08:45compliance data set. And MCD, another acronym, MCD, it has a really cool story in terms of how it came

08:54to be. But it's an example of the power of MISMO and why everybody in our ecosystem should care about

09:02MISMO. So about three years ago, the Conference of State Bank Supervisors, which is the state's,

09:09you know, basically it's the trade association for the state regulators. So CSBS came to MISMO with a

09:16problem statement. And their problem statement said something like, our 50 state regulators have requested

09:24through CSBS that we develop a data standard so that they can, when they're asking lenders for loan file

09:32data for audits, it's in the same format. And the problem they're trying to solve for is,

09:39nobody from a technology platform perspective is going to spend a lot of time trying to build a

09:44solution for the state of New Mexico in their loan audit process. So if they could come together on a

09:49standard, there's much more likelihood or a higher likelihood that technology can come in and actually

09:54change and help the states with their audits. Literally at the same time that CSBS brought that

10:01to MISMO. Rocket, large 50 state regulated licensee, brought the exact same problem statement

10:09worded differently to MISMO. And so the fact that the problem came from industry, Rocket, and regulators

10:17through CSBS, we then formed a work group that consisted of lenders like Rocket and others, technology

10:25platform providers like loan origination system platforms, compliance tech platforms, and then

10:31the regulators all had a seat at the table. And Sarah, this is, I don't want to brag about this

10:37because it's not worth bragging about, but it took three years for those parties to collaborate and

10:44create what today is the MCD. We want to move much faster than that going forward, but you can imagine

10:51there was a lot of push and pull and give and take. And the work process that exists within these

10:56MISMO work groups is a collaborative process. We talk about the participants come in, they take off

11:03their company hat and they put on industry hat and they get to work together to solve these problems

11:09through standardization. So we're in the adoption phase of MCD. It would be great at this time next year,

11:17you and I are talking and maybe we're now at a stage then where states are mandating that lenders

11:23use this data set. And we do see this as a tremendous opportunity to improve the quality

11:31of the audits that get done, improve efficiency. The states will tell you that the rework of data,

11:38the back and forth that they do with their licensees could double the cost of the state audits,

11:44which the lender pays for. And then lenders, especially large lenders, they have full-time

11:49staff members that literally most of their job is working and reworking data. So there's a huge cost

11:56opportunity and a quality opportunity here as we move from this adoption phase, hopefully into a

12:02mandated state. You know, anytime you're doing collaboration, especially across 50 states, regulators,

12:09industry, all parts of industry, no wonder it took three years. I mean, actually, I mean,

12:13when you look at the Road to Housing Act, how long is that? How long did that take? So anytime

12:17you're

12:17doing collaboration, it's going to take a while. But I think it's such a great example of the power

12:24of this, because, you know, what we hear all the time from companies is like, certainty is so helpful,

12:29right? And standards are in some ways, it's certainty. It's like, hey, this is what you need to do. And

12:36someone else's, all the work going into this, again, it's just like, the first thing we talked about,

12:41I just feel like this is such a benefit to small and midsize, but anybody, large lenders, like

12:46Rocket or other people who have the 50 states, I mean, I can't even imagine audits from 50 states, and

12:53the

12:53different data requirements that they have.

12:56Yeah, what one of the things that one of the byproducts that CSBS is excited about, and by extension, its

13:04members, the state regulators is there's already this concept of one company, one exam. And so if you're a

13:11multi state lender, you can opt in to a process where your lead state will do the exam, and then

13:19other

13:19companies can accept that. But there's still a little bit of friction in that because the data requirements from

13:25state to state are differ. And so once MCD is is adopted, you know, by by most of if not

13:33all the states, then it

13:34totally changes the dynamic of this one company, one exam concept. And again, I, there's a there's a massive

13:41opportunity to take cost out of the equation for both the regulators and the licensees, and ultimately, who benefits the

13:50consumers that we all serve.

13:53Absolutely. Okay, let's talk about another hot topic credit score models. Okay, so this is something that, you

14:00know, I feel like there's not a lot of consensus around you have, you have different people feel like

14:04different things. I think that what people are trying to solve for here, what lenders would like to solve for

14:09is is the

14:11cost right now, right. But then you also have quite a bit of talk about likes, standards, but, but especially

14:19like,

14:19the risk, right, is this opening up if we change this? What does this mean to lending standards? What is

14:25the risk

14:26involved? How does some how does MISMO come into this kind of conversation?

14:31Yeah, it's, it's a great question, Sarah. And it's one where so first of all, we were on the heels

14:36of our fall summit and

14:39Clayton Collins moderated a super credit super panel, we called it. Somebody told me afterwards, that may never happen again

14:48to

14:48have those five companies share the stage. So I bet we were, we were very fortunate. But I think it

14:55speaks

14:56to, I think it speaks to the what MISMO means to the industry, and it is this collaborative place. And

15:03so

15:03our, you know, one could argue that there's no, there's no more contentious part of our industry

15:09right now than the credit part of the industry. You know, there's a lot of kind of stuff going on

15:14there. And so where, where MISMO sits, and I'm a certified mortgage banker, you know, I, I have

15:22a personal opinion, but I have to set that aside. I set my personal opinion aside as it relates to,

15:30you know, FICO classic versus 10 T versus vantage score 4.0 versus try merge versus buy merge versus

15:38single merge. I do have an opinion. Nobody cares. It doesn't matter. My role at MISMO is it does not,

15:45my opinion doesn't come into my role here. And what we, what we see specific as it relates to credit

15:51is a focus within our credit reporting work group. And trust me, there's people in there that have

15:58opinions one way or the other on, on how this shakes out. But again, they, they kind of, they check

16:04that

16:04opinion at the door when they step into the work group and whatever the problem is that they happen

16:10to be working on, which in this case, it's really it's really modernizing the, the, all of the data

16:17that makes up these credit files that get moved from, you know, point A to point B to point C.

16:22And so we were, we were fortunate to, to have, you know, Clayton in the, in the, in the five

16:29companies

16:30do that session. And I'll, I'll tell you, there was a, there was some tough questions asked, but I

16:35thought all the participants replied with respect for the industry, for each other. And there was,

16:44you know, it was truly, it came across as a, a positive collaborative session. But it's, it's,

16:52it is, you know, we're not going to solve in MISMO. We're not going to solve for

16:56the, the tri merge, buy merge, single merge. We're not going to solve for what's better,

17:01which models better. What we're going to solve for is what, what does the data set need to look like

17:07across the different credit products and credit models such that we, we have these fun, you know,

17:13we have a fungible process and we can, you know, the technology can talk to each other and things like

17:18that. Yeah. You're not going to solve it for us, Brian. Come on now. We, we need somebody to solve

17:24this.

17:25Not my job. No, that it's a great, a great way to bring everything together. That, that panel,

17:32of course, we had a reporter there, actually an editor covering that. Clayton thought it was

17:38an incredible panel. Your whole fall summit was great. You have a winter summit coming up.

17:43Yeah. And, you know, would love to, to promote that for you guys, because it is such a gathering of,

17:49of, of what the community does. What else do you guys work? Okay. That kind of hit some high topics.

17:53Anything else that are your priorities for you guys right now? So we're, yeah, we're kind of,

18:00I'm going to sneak this in here. I'm going to get in trouble, but stay tuned guys. First part of

18:05October, we're launching a new brand. So we're refreshing the brand of, of Mismo. So we're

18:11excited about that. We, you know, there's, there's 30 plus active work streams and we just talked about

18:19three or four of kind of the big ones. But there's work going on in loan servicing from,

18:25from loan boarding to investor reporting and all the things in between. There's, there's an exciting,

18:31exciting work being done. I think it's exciting, but I'm kind of a mortgage nerd, exciting work being done.

18:38Um, and we're standing up a brand new work group. It's, it's going to be, uh, titled the tri or

18:43the, uh,

18:44tri party agreement. So today, if you, if you think back to February and you all reported on this,

18:51our industry hit the 15% mark for e-note adoption, right? That was like the high water mark.

18:57Well, it's 2026. Don't you think we should be like well beyond 15% adoption?

19:04There's things there's, there's friction still in the process. One of those is the requirement for

19:09a tri party, uh, agreement on every note between the warehouse provider, the ultimate investor buyer

19:17of a loan, and then the originator of the loan. And we're told that, you know, for an originator

19:22to add an investor, another investor that maybe buys, uh, purchases loans with e-notes,

19:29it could take three months or six months to negotiate these tri party agreements.

19:34So again, MISMO became the place and we kicked this off at our, uh, our recent summit where we had

19:40somewhere around 50 or 60 representatives in the work group representing lenders, warehouse providers,

19:47technology platform providers, um, and investors, right? The penny max, the Ameri homes of the world

19:54were in their plan at home loan, et cetera. And so we're excited that that group is now getting formed

19:59formally. It'll be a brand new work group. And the goal, the objective is to, again, check our

20:06corporate hats at the door, work collaboratively, collaboratively to create templated legal language

20:11that the industry agrees upon. And that takes us out of this three months, six months, nine months

20:17contract negotiation. Cause we do believe that if we can advance, um, if we can get more buyers

20:23of these e-notes that opens up the liquidity. Uh, and then we, we, we think, you know, getting from

20:2915

20:29to 25% in, in 2027 is within reach. So we're kind of informally calling it our drive to 25,

20:37uh, for

20:37e-note adoption. I love this again, another example of where standardization doesn't sound very sexy,

20:45but it actually could make a huge difference in people's bottom line and their business,

20:49which is the whole goal. And those, you know, that e-note, if you're, there's a couple studies

20:54out there. I saw, uh, Veterans United, um, did a great study, like the, the impact of their e-note

21:01adoption. Now they're at the very, very, very far end of the spectrum. Like they've really leaned into

21:07e-note adoption, like they're North of 75% probably on their, on the Ginnie Mae for sure.

21:13You can go check the study, but you know, I think they pegged like an $8 million a year

21:19incremental savings to their operation because of their, uh, ability to adopt a fully, you know,

21:25e, e-mortgage, e-note process. And so, you know, even the smaller lenders, if they can get

21:31half of their production in that category, it does make a meaningful difference in their

21:37operational cost to produce. And ultimately the, you know, the rate and fees that they pass along

21:42to consumers.

21:44Okay. You, you're, you mentioned you're coming up on a year. What's the funnest part about your job?

21:48Um, it's, it's actually doing this. It's, it's talking, it's, it's sharing

21:53the, the, the work that gets done by the community. And like myself and my team, there's six of us

21:58on,

21:59on the MISMO team and people are like amazed. There's only six people in MISMO. Yeah, that's it.

22:04Six of us. None of us have propeller hats. We're, you know, we're not in the,

22:08we're not in the work groups doing the work. The, the most gratifying part of, of my first 11 months

22:15is getting to see the hundreds of humans that come from every part of our industry, uh, that sit in

22:23these work groups, uh, and selflessly give their time, experience, expertise to truly solve problems

22:30that are bigger than their company, right. That are truly industry problems. And, and yeah, you know,

22:37fall summit was a great example. We had our, a record attendance, uh, at our summit, you know,

22:43it could have been because we had Clayton Collins on stage. I don't know, maybe, uh, but you know,

22:47we had record a number of people attend the summit and I popped into as many of those work groups

22:53as

22:53possible. And just, it's just cool to see the energy, uh, that people have to solve these problems.

22:59And so for me, I've, I've gotten the most enjoyment, um, out of like uncovering those people, like,

23:06you know, almost being like the, the inside eyes and ears of what's happening. And then

23:11sharing that with the industry because there's a lot of people that they still don't know exactly

23:16what we, what we do at MISMO. And I, I thank you for giving me this opportunity because it's,

23:21it takes these kinds of conversations to really, you know, uncover the, the power and value of,

23:26of this MISMO community. We love helping, uh, shine the spotlight on people doing such great work.

23:33Brian, thank you so much for being with us and we will talk again soon. For sure. Thanks, Sarah.

Category

Transcript

Recommended