Fannie Mae and Freddie Mac are requiring AI governance regardless of origination channel, third-party service provider, or other mortgage operations. Aletyx turns that guidance into explainable decision services and AI guardrails, with approval, versioning, and retrievable audit trails built in.
Fannie Mae, Freddie Mac, servicing policy, and internal controls.
Reviewable decision tables, workflows, and test cases.
Approvals, versions, environments, and release history.
Applications, agents, vendor workflows, and guardrails call the same governed decision.
The new guidance turns AI from an innovation program into an accountable operating control. The lender or servicer of record needs evidence across origination, servicing, third-party service providers, and every loan origination channel where AI is used.
Effective March 3, 2026
Requires Seller/Servicers to govern AI and machine learning used in origination and servicing, including senior management approval, annual policy review, prompt disclosure, monitoring, audits, bias/security checks, segregation of duties, and oversight across all channels of loan origination.
Published April 8, 2026. Effective August 6, 2026
Requires seller/servicers using AI or machine learning in origination or servicing to maintain policies and procedures, manage AI risk, address trustworthy and ethical use, govern subcontractor and vendor AI, and disclose AI use on request.
AI decisions may happen inside the lender, at a broker, through a correspondent, with a contract underwriter, or inside a vendor platform. The obligation still rolls back to the lender or servicer of record.
Loan officers and operations teams use AI for pricing, income, fraud, conditions, or exception routing.
Broker submissions can apply lender guidance differently before underwriting or post-close review detects the issue.
The lender remains accountable for loans acquired through correspondent books, including repurchase exposure.
External underwriters make credit decisions that define the lender's regulatory position and exception patterns.
Title, appraisal, income, flood, and other services may embed AI the lender does not directly control.
Aletyx turns distributed AI activity into a defensible record: channel, inputs, policy version, decision result, owner, approval path, and monitoring evidence.
Mortgage AI governance is not proven by a policy document alone. Lenders need decision-level evidence before a GSE, investor, examiner, or internal risk team asks for it.
AI governance does not stop at a model inventory or inside the lender's four walls. The hard part is connecting each channel's decision, workflow, approval, vendor, and production record to policy the lender can defend.
Document how AI and machine learning are developed, implemented, used, maintained, reviewed, and owned across mortgage operations.
Map, measure, and manage AI risk based on legal requirements, risk tolerance, security expectations, and fair lending obligations.
Prove governance for AI used by retail teams, brokers, correspondents, contract underwriters, and vendors, not only systems the lender directly controls.
Keep a record of the inputs, rule version, decision result, approval path, and effective date behind each governed decision.
Separate the people who author, approve, deploy, and monitor decision logic so control is visible and enforceable.
Review AI-assisted systems for performance, security, bias, vulnerabilities, and deviations from approved policy.
When third-party AI tools are called through governed workflows, record the interaction, apply controls to the outcome, and maintain an audit trail of what was sent and returned.
The first project may start in origination, but the same accountability pattern applies anywhere AI or automation influences a loan-level outcome, regardless of channel, vendor, exception, communication, or handoff.
Aletyx is not a loan origination system or servicing platform. It is the governed decision and workflow layer that sits beside loan origination, servicing, agent, and vendor workflows to make policy execution traceable, testable, and auditable.
Aletyx AI Assistant helps translate investor guidance, lender overlays, servicing policies, and compliance rules into decision models and workflows that humans can review.
Policy owners, compliance teams, and engineers can inspect the logic as decision tables, test cases, workflows, and versioned changes before anything is published.
Approved models run as governed services for loan origination, servicing, vendor oversight, quality control, pricing, or compliance systems, with environment controls and release history.
The same approved decision can be invoked by agents through skills, or used as a guardrail pattern in large language model pipelines, including NVIDIA NeMo based architectures.
QCon AI Boston 2026 session
Mortgage AI works best when probabilistic AI handles ambiguity and deterministic models handle accountable decisions. That separation is what makes agentic systems usable in regulated workflows.
The evidence trail has to be created as the decision runs, not rebuilt after a review request. When a reviewer asks why a loan, borrower, file, or exception was handled a certain way, the answer should come from the system of record.
Short answers for lenders evaluating how to operationalize AI governance.
We value your privacy and believe in transparency. This panel lets you control exactly how we collect and use data during your visit.
Cookies are data stored by your browser that help websites remember preferences and understand usage patterns. You have complete control over which categories you allow.
You can manage your preferences below. Please note that disabling certain cookies may impact your site experience.
