Guides
Fannie Mae LL-2026-04: what lenders need to do before August 6, 2026
What LL-2026-04 requires
Fannie Mae Lender Letter LL-2026-04 establishes AI governance expectations for sellers and servicers that use AI tools — including automated underwriting augmentation, valuations, income analysis, and fraud detection — in workflows that touch Fannie Mae-eligible loans. The letter requires lenders to: (1) maintain a documented inventory of AI tools used in covered workflows; (2) implement human review checkpoints for AI-assisted decisions, particularly for adverse outcomes; (3) retain audit records sufficient to reconstruct what AI outputs were relied upon and how they were reviewed; and (4) have a process for identifying and escalating model performance issues.
Last verified: 2026-06-16. The effective date is August 6, 2026. Requirements and guidance may be updated before or after the effective date — verify the current version of the Lender Letter directly at fanniemae.com and confirm readiness steps with qualified counsel before the effective date.
A pre-August 6 readiness checklist
Lenders should work through the following before August 6, 2026: (1) Inventory all AI tools currently in use in any workflow that touches Fannie Mae loan origination, underwriting, or servicing. Include vendor-provided models and any internally built scoring or decisioning tools. (2) For each tool, document the use case, the inputs used, the decision or recommendation the tool produces, and how a human reviewer interacts with the output. (3) Confirm that your case management or origination system captures the AI output, the reviewer's action, and a timestamp — in a record that can be produced for examination. (4) Brief your compliance and operations teams on the new requirements before the effective date.
Lenders selling loans to Fannie Mae that cannot demonstrate governance of their AI tools by August 6, 2026 face eligibility risk on covered loans. This is not a framework-for-future-planning letter — it imposes requirements with a specific effective date on active sellers and servicers.
How Hadrian supports LL-2026-04 readiness
Hadrian's evidence graph links each case decision to the AI outputs reviewed, the documents uploaded, and the human reviewer actions taken — in an immutable audit ledger. The operator-gated AI trust dial lets operators define which AI recommendations require mandatory human review before a decision can advance, creating the human oversight checkpoint the letter requires. Operators can export the full audit record for any case to support a Fannie Mae examination.
Hadrian provides the infrastructure to document AI governance; it does not constitute Fannie Mae approval of any lender's AI tools or governance program. Compliance with LL-2026-04 is the lender's obligation. Sellers and servicers should review the full text of the Lender Letter and consult with Fannie Mae relationship managers and qualified counsel to confirm their specific obligations and readiness posture.
FAQ
Fannie Mae LL-2026-04 AI governance readiness — common questions
Does LL-2026-04 apply to non-bank lenders?
Yes — it applies to all Fannie Mae-approved sellers and servicers, including non-bank mortgage companies, that use AI in covered workflows. IMBs and specialty lenders selling into the secondary market are in scope. Verify your specific coverage with Fannie Mae or qualified counsel.
What counts as an 'AI tool' for purposes of LL-2026-04?
Fannie Mae's guidance covers AI and machine-learning tools used in underwriting, valuation, income analysis, and fraud detection for covered loans. Vendor-provided tools, third-party data models, and internally developed scoring models are all potentially in scope. A conservative approach is to inventory any model or tool that produces a recommendation, score, or flag that a reviewer acts on.
The institution around the intelligence
See Hadrian run your case lifecycle — intake to close, every decision audited.
Governance-native case processing for lenders and regulated teams.