Fannie Mae’s artificial intelligence and machine-learning governance requirements take effect Thursday, Aug. 6, giving approved seller/servicers a final deadline to formalize how the technology is used across mortgage origination and servicing.
The requirements apply when a seller/servicer uses AI or machine learning in connection with loans sold to or guaranteed by Fannie Mae or loans serviced on its behalf. They cover internally operated systems as well as AI used by vendors and subcontractors.
Fannie Mae established the framework in a letter published April 8 with an effective date 120 days later.
The lender letter requires seller/servicers to maintain policies and procedures governing the development, implementation, use, and maintenance of AI and machine-learning systems. Those policies must address risk management, applicable laws, trustworthy and ethical use, employee communication, and responsibility for reviewing the program at least annually.
Companies must also manage vendor and subcontractor AI under protections that are no less stringent than those applied to their own systems.
Why The Deadline Matters
The requirements fall on Fannie Mae seller/servicers, not individual loan originators. But they could directly affect which AI tools LOs are permitted to use and how borrower information is handled.
Originators should understand their company’s policies before entering borrower data, loan documents, financial information, or internal company material into an AI platform. Consumer-facing uses—including AI-generated emails, document analysis, underwriting support, and borrower communications—may also require approved systems, human review, and documentation.
An AI feature embedded inside a larger mortgage platform should not automatically be assumed to fall outside the framework. Seller/servicers remain responsible for governing AI used by their technology providers.
Fannie Mae may request prompt disclosure of the types of AI or machine learning a seller/servicer uses, how the systems are used, their purpose, and the safeguards in place to mitigate risk.
NMP recently reported that mortgage technology provider TRUE released an AI governance guide focused on tracing borrower information from its source document through AI processing, validation, human review, and entry into the loan origination system. TRUE’s specific architecture is not required by Fannie Mae, but the guide illustrates the operational questions lenders are beginning to ask of their systems and vendors.
Freddie Mac’s separate, more detailed AI and machine-learning governance requirements have been in effect since March 3. Companies approved by both GSEs must evaluate their programs against each organization’s requirements.
The Aug. 6 deadline does not require lenders to adopt AI. It requires seller/servicers that use the technology in covered origination or servicing activities to demonstrate that its use is governed, monitored, and accountable.
