A key part of the proposal is a push for a single level of government, with officials warning that consolidation of state-level regulations “will undermine American innovation and our ability to lead the global AI race.”
“The management is looking forward to working with them Congress in the coming months so that we can turn this draft into a law that the President can sign,” the announcement said.
Diane Yu, CEO of the mortgage technology platform tidal waveyou have been told HousingWire that the state standard will accelerate the adoption of AI in the industry.
“We have lenders in all 50 states using our real estate platform today. That means every new state-level AI rule is something our engineering teams have to test, translate and build, instead of using that time to make the product better for borrowers and loan officers,” said Yu. “One level of government would allow us to invest that power in the product itself.”
Yu said he has had discussions with lenders who are ready to use AI but are stuck in legal review because their advisers cannot give them clear answers on what is required on a state basis.
“Lenders already follow one set of underwriting guidelines Fannie Mae again Freddie Mac. Adding a second, separate layer of AI-specific state laws on top of that doesn’t create clarity. It creates paralysis,” he said, adding “the way [of the framework] fair” and that the mortgage industry is a government-controlled industry at its core.
“Lenders already comply with ECOA, the Fair Housing Act, TILA, and RESPA. The GSEs set national underwriting standards through Fannie Mae and Freddie Mac. And the federal guardrails have no theory. Freddie Mac’s Bulletin 2025-16, which went into effect on March 3rd, requires every dealer/framework to study management equipment,” Yu said.
“That means ongoing monitoring, legal bias testing, compliance with NIST and ISO cybersecurity standards, senior management accountability, segregation of duties between AI development and risk oversight, and independent audits. That’s not a simple touch. That’s strong, asset-specific AI oversight, and it’s already happening at the government level.”
While Yu is optimistic about the framework’s principles, he said he wishes the framework would deal with cases of using AI that are specific to collateral.
“One thing that’s missing from a lot of the AI policy discussion is the difference between companies that are deploying AI in manufacturing and companies that are just talking about it. In mortgages, that gap is wide. We’ve been working with the country’s lenders for over a year. That’s a different conversation. [from] the company is showing a demo at the conference,” said Yu.
“I would also like to see the draft address something specific to loans: the industry is already losing about $600 in origination costs, and closing loans still takes 43 days on average,” he added. “If the goal is consumer protection, faster approvals, lower costs, and fewer errors [key to] consumer protection. A proper regulatory framework should make it easier for lenders to use technology that delivers those results, not harder.”
