Artificial intelligence is transforming the mortgage industry at a fierce speed, but regulators are on the sidelines, which increases the risk of bias and inequality.
Key Points:
- Artificial intelligence can reduce costs and reduce mortgage lending errors, but with thoughtful use while making the process more efficient.
- However, with the Consumer Financial Protection Agency being largely on the sidelines, it is unclear who is responsible for regulating the use of AI in the mortgage industry.
- Transparency and equity are key to making decisions that shape the way the industry operates today, and transparency and equity are key.
A bigger perspective is needed to think big about the success of residential real estate. Industry decoding It features industry experts who can enrich an understanding of issues affecting the entire industry.
The only views expressed in this column are those of the author.
Eight months ago, I predicted that an AI-driven mortgage approval system would dominate the industry by the end of 2025. At the time, predictions sounded bold. But today, it all feels inevitable.
What began as a modest experiment, such as using AI to trim processing times and automating fraud checks, has accelerated into something transformative. Increasingly, decisions regarding who is eligible for a mortgage are shaped not by human underwriters but by algorithms that analyze thousands of variables at the rate of lightning.
Efficiency revolution
Famous for requiring documents, repetitive verification and endless handoffs, mortgage lending has long been defined by inefficiency.
AI promises to break that slow, expensive, error-prone cycle. Instead of the processor manually combying paytabs, bank statements and credit reports, the algorithm can analyze them all in seconds. You can no longer tire you, never miss things, and you can abnormally flag them for thousands of loans.
Recent pilot and product launches have shown impact.
-
40-60% reduction in defects after closure, less expensive correction
-
For 3-5 days, we cut down our funding timeline and accelerated closures.
-
Approval speed is more than twice the industry average
For borrowers, this means faster answers and reduced headaches. For lenders, that means lower costs and higher throughput in margin-starved environments.
Expanding the net for borrowers
In addition to efficiency, AI offers inclusiveness. Self-employed workers, contractors and gig makers have often fallen by the traditional underwriting cracks in favor of W-2 employees, but AI can change it. By analyzing cash flow data, rental payment history, and even utility invoices, the algorithm can build a more subtle image of a borrower's ability to pay back.
Key examples: Fannie May and Freddie Mac are now available to use vantagesCore 4.0, a scoring model that incorporates rental and communication payment history (buy now/report later). The shift could unlock $1 trillion in lending capacity and expand homeownership to additional 5 million households.
Combined with AI's ability to process non-traditional data at scale, this could be one of the most important breakthroughs in expanding access to credit in decades.
Shadow of bias
AI models are trained on historical data, and in US housing, their history carries a weight of discrimination.
Research by Brookings and MIT suggests that automated systems can underestimate homes in most Black communities or punish applicants based on incomplete or “noisy” credit data. The model may not intend to discriminate, but it is not intended to whether the outcome perpetuates inequality.
The industry argues that AI reduces human bias. And it is in some respects, as the algorithm does not have personal bias. But they also have no human judgment or context. Although machines can apply rules faster, it is still not possible to question whether the rules are fair.
That paradox – both as equalizers and dividers – is the central tension that shapes the future of the industry.
Watch Dog Retreat
This is usually where regulators intervene. For the past decade, the Consumer Financial Protection Bureau (CFPB) has been the leading watchdog to ensure lenders comply with the Fair Lending Act.
However, in 2025, CFPB authority is under siege. That superintendent was fired earlier this year, operations were temporarily suspended, and Congress proposed cutting the budget by almost half. The ability to act decisively continues to fall.
Other regulators are monitoring development, but none have the same duties as setting clear standards of fairness in AI underwriting.
Limitations of self-regulation
Without federal guardrails, lenders would have left themselves to the police. Some people invest heavily in equity audits, bias testing, and explanability tools. Assume that others are less intentional, prioritizing speed to the market and can resolve pop-up issues later.
However, once the harm is eliminated, it is difficult to reverse it. When inequality emerges, reputational damage spreads quickly. The industry is risking a new wave of criticism if AI is sold as a force for fairness — finds it perpetuates discrimination.
Road fork
When done correctly, AI can make lending faster, cheaper, more comprehensive, and support millions of households that have historically been overlooked. The wrong thing, it could strengthen bias in the mortgage system for decades.
The outcome is shaped by decisions made by lenders, policymakers and technology providers. Transparency, explanation, and fairness require prioritization, just like cost savings and speed.
The broader credit model embrace shows that it is possible, but this is just one of the much bigger puzzles. AI doesn't just come to lend mortgages. It's already here. AI already offers measurable benefits to speed, cost savings and borrower experience. But it also arrives in a cloud of uncertainty, as regulatory frameworks are unstable and the risk of bias is still realistic.
The machine is looking at us. The real question is who is looking at them?
Coby Hakalir has been a leader in the mortgage industry for almost 30 years. He currently leads the mortgage banking and mortgage tech division of T3 Sixty, one of the most respected management consultants in real estate, and lives in Northern California. (Note: Real Estate News is an editorially independent division of T3 Sixty.)
