The federal government in December took the first steps toward steamrolling any potential bottlenecks that states may create when it comes to moving the mortgage industry forward with
The risk is real. If individual states move forward with piecemeal AI regulation — especially in ways that raise compliance costs or punish experimentation — banks and
I've watched this dynamic up close. When I served at the Federal Housing Finance Agency, it was obvious that the largest banking and mortgage institutions were still operating with post-financial crisis scar tissue. Dodd-Frank and the long regulatory hangover didn't just shape risk management — it shaped psychology. In mortgage technology, fear has become the operating system. Even when leadership teams are adopting AI elsewhere in their businesses, the mortgage stack has remained underinvested, understaffed and culturally resistant to change.
In 2024, the FHFA hosted a "
And here's what surprised me: Major banks declined to participate — not because the work wasn't serious, but because they feared the optics of being associated with generative AI. That is how deep the chill still runs.
At the corporate level, many large financial institutions make significant investments in AI. But in a space as regulated as the mortgage market, it's difficult to portray themselves as forward-leaning. The result is an uneven distribution of investment and thought put into generative AI that's adversely affecting the mortgage industry.
This is why federal leadership matters. The Trump administration wants to be an accelerator for AI to make our country competitive in all industries, including financial services, and it can be by committing to enforce President Donald Trump's executive order from Dec. 11, 2025, "Ensuring a National Policy Framework for Artificial Intelligence." States like New York, California and Colorado need to be held back from creating a further chill on AI innovation in housing finance.
Fraudsters and modestly dishonest employees can use generative AI to quickly create convincing fake utility bills, pay stubs, passports and other documents banks rely on.
Let's be clear about what's at stake. The cost to originate a mortgage has climbed sharply — reaching $11,800 in the second quarter of 2025, up from under $8,000 just five years earlier. That is a massive cost burden embedded in the American dream.
At the same time, there's little competition in the space. When incumbents control distribution and workflow, the system becomes expensive by default — and any extra friction (including regulatory friction) makes it harder for challengers to enter, innovate and reduce costs.
Mortgage lending is, at its core, a data verification business — collecting, checking, reconciling and documenting borrower information. That is exactly the kind of work where AI, deployed with proper controls, can dramatically increase productivity and reduce the reliance on slow, manual processes. Done right, AI doesn't replace judgment, it replaces repetitive operational drag. And when operational drag decreases, labor hours fall, cycle times shrink, defects drop and unit costs come down.
A
But banks and mortgage lenders also play an important role here and can help make homeownership affordable for everyday Americans by reducing costs with AI. Why shouldn't they be able to pass some of those cost savings onto consumers?
Right now, the biggest barrier isn't technology. It's fear — fear of regulatory backlash, fear of headlines, fear of being early. Federal regulators can change that by making the rules of the road clear: Encourage responsible experimentation, set outcome-based guardrails and prevent state-by-state fragmentation from becoming a de facto ban.
If we want a housing finance system that is more efficient, more competitive and more affordable, we need to stop treating innovation like a threat. AI is not a magic wand, but in a market this constrained, it may be one of the only levers left that can actually move the numbers in the right direction.
