Freddie Mac is updating its guidelines to include explicit language requiring mortgage companies to establish a clear artificial intelligence framework governing the technology's internal use.
The new rules become effective on March 3, 2026 and mandates compliance with the policy for businesses using AI and machine-learning tools in connection with loans sold to or serviced on behalf of the government-sponsored enterprise.
Fueled by the rapid growth of artificial intelligence platforms in the public domain from the likes of OpenAI and newer competitors, AI's popularity already had some mortgage industry leaders emphasizing the need for lenders
A new industry guideline from one of the GSEs typically forges a path for the entire mortgage industry to follow. Freddie Mac's update arrives at a critical juncture when the door for both unintentional and deliberate AI misuse is already open.
Following the lead of OpenAI, similar open-source platforms, coming from both tech giants and upstarts, emerged, turning artificial intelligence into a tool many turn to willingly and regularly.
"Everybody's starting to consume that and use it in their day-to-day," said Joe Sorbello, product owner at mortgage software firm Lender Toolkit, who oversees the company's underwriting platform.
Such use of public AI tools, though, raises doubts about data quality if it were to be used for any business purpose.
"AI can do analysis, AI can feed back results," Sorbello said. "But you run the risk of AI learning something wrong."
What's included in new Freddie Mac guidelines
In early December, the GSE added a new section dedicated specifically to AI governance framework. Come March, sellers and services working with Freddie Mac will be required to demonstrate they have processes and procedures across their organizations that show "the mapping, measuring and managing of AI risks are in place, transparent and implemented effectively."
The introduction of "trustworthy" artificial intelligence features will also need to be applied to mortgage companies' practices, with businesses required to determine and employ processes to ensure they meet proper risk management levels based on their tolerance, the update said.
With an eye on safety and compliance, Freddie Mac will require mortgage sellers and servicers to monitor their AI and machine learning tools to look for threats to data integrity, as well as conduct audits to identify potential weaknesses or failures that prevent adherence to their policies.
New Freddie Mac guidance also emphasizes the separation of responsibilities at businesses to prevent conflicts of interest and guarantee accountability is in place. Iinternal rules, lines of communication and associated roles related to determining AI risk must be documented and communicated to all individuals working at the organization.
The new guidelines reinforce the point to the lending industry that ultimate responsibility for risk oversight — and potential penalties if things go awry — lay at the feet of the originators, even when they often put the onus on their tech partners.
Likewise, the update underscores the importance of proper vetting and vigilance on the part of lenders and servicers when it comes time to selecting AI partners, while also helping them determine the dividing line where automation takes place and the human in the loop takes over, Sorbello added.
"A lot of the lenders really rely on us, as the vendors, having those security protocols in place as it relates to security of data, use of data," he said.
How new regulatory clarity may drive AI growth
Freddie Mac's update arrives as mortgage businesses throughout the industry continue to welcome tech advancements but grapple with how to
In 2025 research conducted by Arizent, parent company of National Mortgage News, found that 38% of companies today are choosing to introduce AI slowly and deliberately, with 2% further limits on the introduction of such tools.
Fear of noncompliance amid the absence of clear guidelines rank
Any specific language that clearly identifies regulations and procedures, such as Freddie Mac's updated guidance, should, in the end, break down some of the current AI hesitation present in the mortgage industry today to encourage uptake, according to Sorbello.
"It's going to cause people to ask the questions, which is a good thing. From a lender's adoption perspective, from a lender's risk perspective, it's just going to give them some of that fuel to make sure that they continue to ask the right questions of their vendors," he said.
