AI Is Already in Your Mortgage Company. Do You Know Where?

I had the opportunity to join Greg Sher on his One on One LinkedIn Show alongside Dan Sugg, Chairman of the Mortgage Bankers Association’s Residential Board of Governors (RESBOG), to discuss two topics that are quickly becoming inseparable: artificial intelligence and governance.

The conversation was prompted by the recent release of the MBA White Paper, Examining AI-Powered Mortgage Through the Lens of Federal Law, a document every mortgage executive, compliance professional, technology leader, and vendor should download and read.

I realize that not everyone has time to consume an 18-page legal and regulatory analysis. That is exactly why I wanted to use this week’s column to highlight what I believe are the most important takeaways.

First, let’s address the biggest misconception in mortgage lending today. Many lenders still believe AI governance is something they can worry about later. Others believe it does not apply to them because they are not building AI models. Both assumptions are wrong.

The reality is that AI is already embedded throughout the mortgage ecosystem. It exists in loan origination systems, document processing platforms, fraud detection tools, customer relationship management systems, servicing technologies, quality control platforms, marketing systems, and borrower-facing applications. Many lenders are already using AI whether they realize it or not. That point came up repeatedly during my discussion with Greg and Dan.

The question is no longer whether AI will arrive in mortgage lending. The question is whether organizations understand where AI exists inside their businesses and whether they are prepared to govern it responsibly. That is the central message of the MBA White Paper.

One of the most important conclusions from the paper is that mortgage lenders do not need to wait for Congress, regulators, or state legislatures to create entirely new AI laws. The mortgage industry is already regulated. The Equal Credit Opportunity Act. The Fair Housing Act. The Fair Credit Reporting Act. The Gramm-Leach-Bliley Act. RESPA. HMDA. UDAAP. All of these existing requirements continue to apply when AI is involved. The legal question is not whether AI is regulated.

The legal question is how lenders demonstrate compliance when AI is involved in a decision, recommendation, communication, or workflow. That distinction matters. A lot. One area that generated significant discussion during the interview was the concept of AI-driven mortgage origination.

The MBA White Paper explores whether AI systems can originate mortgage loans without human involvement and whether AI tools require licensing under the SAFE Act. The paper reaches a conclusion that many in the industry may find reassuring. Current federal law contemplates a human mortgage loan originator.

The SAFE Act repeatedly references individuals. Regulation Z requires disclosure of a specific loan originator and NMLS identifier. Existing rules assume that a human remains accountable to the borrower throughout the process. Could technology continue advancing? Absolutely. Will policymakers eventually revisit these requirements? Perhaps. But under today’s framework, human accountability remains a foundational principle. That is not simply a compliance issue. It is also a consumer trust issue.

For most borrowers, purchasing a home remains one of the largest financial decisions they will ever make. Technology can accelerate the process. Technology can improve accuracy. Technology can improve efficiency. But consumers still deserve transparency regarding who they are dealing with and who is ultimately accountable for the outcome.

The White Paper also identifies five major risk areas lenders should be thinking about today. The first is fair lending and bias. AI systems learn from historical data. If historical patterns contain bias, AI systems can unintentionally reproduce those same patterns at scale. The second is explainability. When a borrower receives an adverse action notice, lenders must explain why. A black-box decision is not an acceptable answer. The third is steering. Consumer-facing AI tools cannot be allowed to direct borrowers toward products that benefit the lender at the expense of the consumer. The fourth is privacy and data security. Mortgage lenders handle some of the most sensitive consumer information in financial services. Organizations must understand how AI systems collect, use, store, and protect that information. The fifth is third-party risk. This may be the most important takeaway for many lenders. Buying AI from a vendor does not transfer responsibility to the vendor. If a vendor’s AI system contributes to a fair lending issue, privacy issue, servicing issue, or compliance issue, regulators are not calling the vendor first. They are calling the lender.

As I said during the interview, even if the lender does not own the model, the lender still owns the outcome. That concept is becoming increasingly important because both Fannie Mae and Freddie Mac have now made their expectations clear. Freddie Mac’s Seller/Servicer Guide requires organizations utilizing AI and machine learning to maintain governance frameworks and risk management controls. Fannie Mae has issued similar guidance regarding policies, procedures, governance, and oversight. This is a significant development. AI governance is no longer simply a best practice. It is increasingly becoming a counterparty expectation.

That reality was one of the primary reasons RESBOG identified AI governance as a top priority and ultimately asked MISMO to lead development of a mortgage-specific framework. That framework became FRAME, which stands for Framework for Responsible AI in the Mortgage Ecosystem. FRAME was developed through MISMO’s Artificial Intelligence Community of Practice and became available to MISMO member companies this month. The objective was straightforward. FRAME is a practical toolkit designed to help organizations answer a simple question: How do we responsibly govern AI? The framework begins with visibility. You cannot govern what you cannot identify. That is why the AI System Inventory is the first artifact within FRAME. Organizations start by finding AI. Then inventorying it. Then assigning ownership. Then assessing risk. Then monitoring performance. Then updating governance as systems evolve. That operational rhythm became one of the central themes of our discussion with Greg.

FRAME also recognizes that not every AI system deserves the same level of scrutiny. An internal productivity tool should not require the same oversight as an underwriting engine making credit decisions. FRAME uses a risk-based approach that categorizes systems as Critical, High, Medium, or Low risk based on consumer impact, regulatory exposure, and business risk. That approach aligns closely with recommendations found throughout the MBA White Paper.

Perhaps most importantly, FRAME was intentionally built for the broader mortgage industry. Large institutions often have mature risk management departments, model governance teams, compliance officers, and dedicated resources. Many independent mortgage banks do not. Many community lenders do not. Many credit unions do not. And certainly Mortgage Brokers do not. FRAME was designed to help those organizations establish a defensible governance process without requiring an army of consultants.

As a proud Certified Mortgage Banker, I continue to be impressed by the leadership being shown across MBA, MISMO, and RESBOG on this issue. Dan Sugg and the RESBOG leadership team recognized this challenge early. MBA commissioned the White Paper to help the industry better understand the legal and regulatory landscape. MISMO’s volunteer community transformed that understanding into practical tools lenders can implement today. That is the industry working exactly as it should. By and for the industry.

The mortgage industry will absolutely continue adopting AI. That much is certain. The organizations that succeed will not necessarily be the ones deploying the most AI. The winners will be the organizations that can demonstrate governance, accountability, fairness, explainability, and oversight. Those are the organizations that will earn the confidence of regulators, investors, the GSEs, and most importantly, consumers.

If you take one action after reading this article, make it this: Download the MBA White Paper. Read it carefully. Then take a hard look at your own organization and ask a simple question: Do we know where AI is being used in our business? If the answer is anything less than a confident “yes,” now is the time to start.

#VieauxPoint