Attorney cautions brokers that every transcribed call and LLM chat record could show up in your next exam
Mortgage exams have always come with a document request, and brokers have generally had a sense of what to expect. AI is making that list longer in ways most brokers have not thought about yet.
AI tools are generating records that brokers may not fully realize they are creating. Transcription services capture every client’s call. Consumer LLMs log conversations that may contain non-public borrower information.
Peter Idziak (pictured top) is a principal at Polunsky Beitel Green and works closely with mortgage companies on compliance.
"If you're in a business where you are not necessarily required to save every communication you have, now you have transcripts of every call that you've ever made," Idziak told Mortgage Professional America. "I would not be surprised if you have an exam and the regulator says, ‘Well, give me all these transcripts and then I will run them through my AI and see if it thinks that you've engaged in any sort of prohibited conduct.’ In the past, it just wasn't a source of information that's available."
Widening the exam scope
The shift mirrors what happened with email in the 1990s. For years, organizations had mountains of email that they had no efficient way to review. Document review meant hiring dozens of people to sit and read. Now that the process is largely automated, the cost to find a problem in a decade of communication has dropped dramatically.
AI is doing the same thing with a new category of records, Idziak said.
"Regulators themselves really couldn't have done it sometimes," he said. "But don't be surprised going forward if it's much more of a heftier document request being asked. And then you just have a lot more documentation, a lot more information about your contacts with the borrowers."
Many brokers are also creating a data governance problem without knowing it. When a broker uses an off-the-shelf consumer LLM subscription and enters borrower information, that data is potentially moving outside any compliant framework.
"If you just have an off-the-shelf subscription to Claude and you're putting in non-public borrower information, that could be a problem for you," Idziak said.
Idziak said the same concern applies to AI used for marketing and borrower targeting. If those tools filter potential borrowers in ways that correlate with protected classes, he said, the ECOA and Reg B questions have not been answered yet.
The accountability question
Courts have already weighed in on AI accountability in related contexts. Lawyers have been sanctioned for submitting AI-generated briefs that cited cases that did not exist, and the defense that the tool was responsible got them nowhere.
"If someone tries to raise the defense of, ‘Well, I bought this AI solution, and it's not my fault that it did what it did,’" Idziak said, "I don't think regulators are going to be very accepting of that."
AI regulation is also still being written in real time. Federal and state approaches conflict, and some state laws are written so broadly that they would accidentally cover tools like automated underwriting systems. Idziak said brokers cannot wait for definitive rules before experimenting, but they need to understand the territory is unmapped.
"AI is the same as any other technology that you've had to deal with across 50 states," he said. "The issue is more that because AI is evolving so quickly, the statutes and regulations are struggling a great deal to define precisely what an artificial intelligence system is."
Idziak said the bottom line is that most brokers and lenders are trying to do the right thing. It becomes challenging when there are different regulations from state to state to know what the right thing is. While allowing states to write their own AI laws allows room for the technology to advance, looser regulations could come with their own problems as well.
"The vast majority of brokers and lenders, especially people at scale, are trying their best to comply with the law," Idziak said. "The industry gets a bad rap sometimes. But to just have a very loose regulatory framework for AI usage, my worry is that it leads to an increase in fraud and fraud at scale, and oftentimes against the most vulnerable members of society.
“And it just makes for a less trusting society. You want your borrower to have a good customer experience. You want them to think well of the industry. Public perception is very important."
Stay updated with the freshest mortgage news. Get exclusive interviews, breaking news, and industry events in your inbox, and always be the first to know by subscribing to our FREE daily newsletter.
