top of page

The AI Mortgage Landscape – The Risk Weighting of Incorporating AI into Your Company

Nov 8

3 min read

Artificial Intelligence, or AI, is the new buzzword across the world, regardless of industry. This includes the U.S. mortgage industry—one of the most highly regulated and complex sectors when it comes to consumer information, disclosure, and credit decision processes. Surely, AI can revolutionize a space burdened by a decade of continuous cost increases, resulting in per-loan manufacturing costs that exceed $12,000 per closed loan. It seems like every industry vendor is embracing AI to leverage the hype and claim efficiency, cost-cutting, and, of course, less work and more revenue.


But what about the risks of AI? How do we categorize it to assess risk? Are there regulatory implications? Potential consumer risks? How do we even apply 50-year-old regulations, such as RESPA, to emerging technology? Does my vendor management department even understand it? These are essential questions to consider when thinking about how to incorporate AI into your company's workflow.


I’ll be upfront—I’m not here to throw around buzzwords or tell you that AI is a one-size-fits-all solution that will transform your business overnight. With 20 years of experience as a mortgage banker, I've learned the importance of assessing risk and approaching new technology with skepticism. Otherwise, you might bear the consequences of regulatory penalties later.


So, what is the best way to integrate AI technologies while understanding risk? Start by cleaning house. Understand basic risks, establish an organizational structure, and then categorize your systems with a risk assessment. Big ask, right? It’s easier than you think. Let’s dive into understanding risk assessment.


Two Main Categories of AI Usage

  • Direct AI Usage – Where a company or lender directly develops and/or adopts AI internally. An example could be an internal generative AI or private GPT server or an internal rules-based guideline lookup or AI chat system for employees to check guidelines.

  • Indirect AI Usage – Where a company relies on a vendor using AI in their system. You have no control over its usage, but your business is indirectly impacted by its decisions. An example is an AI-driven credit decision-making system or a purchased CRM leveraging generative AI to convert leads.

  • Four Quadrants of AI Risk

  • Rule-Based or Machine Learning AI – Any AI based on a programmed rule waterfall or a system that adapts those rules based on input over time.

  • Generative AI – Any GPT or AI engine using predictive responses.

  • Consumer-Facing AI – AI that communicates directly with consumers or allows consumers to interact directly with it.

  • Operational AI – Any system your back office uses to automate or improve processes.


With this basic understanding, you can start mapping your AI usage into these quadrants. This

will help guide decisions on where to allocate resources and where to act cautiously.

Risk Weighting Corners of the Quadrants

  • Best Revenue Opportunities – If a system falls into the Consumer-Facing, Rule-Based quadrant, it presents lower risk and higher revenue potential.

  • Lower Risk Usage – Operational, rule-based AI is generally lower-risk compared to generative operational usage.

  • Best Cost-Savings Opportunity – Generative AI poses higher risk but offers substantial cost savings as it can replace decision-based processes over time.

  • High-Risk Usage – Consumer-facing generative AI carries the highest risk that regulators are concerned about, potentially leading to inconsistencies or misleading information, which threatens organizations.


Once you’ve categorized your AI usage, you can determine which areas to pursue and which to approach with caution.


AI has become a catch-all term for vendors and lenders, positioned as the modern solution for all issues, the "savior" of the mortgage industry. Forgive my skepticism, but we are responsible for every aspect, from consumer disclosure to data privacy. Colorado recently became the first state to pass AI legislation focused on consumer protections, and other states will follow suit. Remember, regulators are watching closely. My recommendation is to be a low-risk early adopter, not a high-risk early penalty payer. Use responsibly and always keep the consumer in mind.


Article by: Matthew VanFossen,


About the Author: Matthew VanFossen is the CEO of Absolute Home Mortgage and Mortgage Automation Technologies. A prominent figure in the mortgage regulatory community, he leverages his expertise in mortgage lending, technology, and regulatory practices to bring a common-sense approach to the mortgage landscape, making valuable contributions to the industry.

bottom of page