Skip to main content
Nick’s Lending
Menu

The next move / AI in Real Estate

AI in Real Estate

AI is changing mortgage fraud. Your closing checklist needs to keep up.

AI can help mortgage companies find suspicious patterns. It can also help criminals create convincing documents, messages and voices. Here is what that means for your loan file and the money you send at closing.

Jump to a section
Conceptual illustration of a magnifying glass examining two similar paper houses, one layered with transparent sheets
AI-generated conceptual illustration. Not an actual property or software product.

Imagine this hypothetical: you are a few days from getting the keys when a message arrives with revised wire instructions. The writing is polished. The sender seems familiar. Then a phone call appears to confirm everything.

Would hearing a recognizable voice make you comfortable sending the money?

That is a question worth settling before closing week. AI is changing both sides of mortgage fraud: how convincing a deception can look and how financial companies search for warning signs. For buyers, the useful response is a clear verification plan, plus a loan team that can explain requests and resolve discrepancies.

The new problem: something convincing can still be fake

In its December 3, 2024 public warning, the FBI described criminals using generative AI to improve fraudulent messages, produce false identification images and imitate voices. It also warned about fabricated video conversations. The warning concerns financial fraud broadly; it does not establish how often these methods are used in home purchases.

FBI guidance on AI-assisted financial fraud (new tab)

For a homebuyer, the implication is practical: polished language, a professional-looking image or a familiar-sounding caller should not replace independent verification. A closing plan built around spotting bad spelling leaves too much to chance.

The Treasury Department's Financial Crimes Enforcement Network, or FinCEN, documented another part of the problem in November 2024. It reported increasing suspicious-activity filings during 2023 and 2024 involving suspected deepfakes, including false identity documents designed to get past verification checks. Those are reports of suspected activity, not a count of proven mortgage crimes.

FinCEN's deepfake-fraud alert (new tab)

Mortgage fraud detection is getting an AI upgrade, too

On May 28, 2025, Fannie Mae announced an AI-powered Crime Detection Unit developed with Palantir. Its announcement described technology that monitors unusual transactions, activities and behavior to help identify suspicious activity for investigation.

Fannie Mae's launch announcement (new tab)

That points to a different use of AI from the chatbot most people picture. Instead of answering a buyer's question, a system can help investigators decide where to look across a large collection of information. The promise is finding relationships that are difficult to notice when examining one file at a time.

Keep the evidence in perspective. Fannie Mae's announcement expressed expectations about future fraud prevention and savings, and explicitly identified forward-looking statements. It was a launch announcement, not an independent performance study. It does not establish that a particular buyer will close faster, qualify more easily or receive a lower rate. Nor does it establish that your lender uses this platform.

The borrower benefit to look for is understandable follow-up: what needs verification, which documents will help and who will review them.

In 2026, oversight became part of the AI story

The change is not limited to buying new software. Fannie Mae's April 8, 2026 lender letter set governance requirements for sellers and servicers using AI or machine learning in connection with covered Fannie Mae loans. The letter took effect 120 days after publication, on August 6, 2026.

Fannie Mae's AI governance requirements (new tab)

The framework calls for documented policies, risk management, an identified owner and at least annual policy review. It also addresses vendor and subcontractor use. Buying a tool from another company does not remove the seller or servicer's oversight responsibilities.

For buyers, this suggests useful questions: Where does automation enter my application? Who can review a document-reading error? How should I provide supporting information securely? These questions are about how your loan team works; the letter does not promise you access to a lender's internal fraud tools.

A follow-up question is not a finding of fraud

FinCEN's alert makes an important distinction: a single warning sign does not necessarily establish suspicious or illegal activity. Context matters. That is a good reason to approach a document question calmly rather than assume an automated flag tells the whole story.

FinCEN's explanation of red flags (new tab)

Consider a hypothetical transfer between two of your own accounts. Someone looking only at the receiving statement sees a deposit. The full record can explain where it came from. CFPB guidance notes that buyers may need to document large deposits and recommends asking the loan officer exactly what is needed. This can arise in ordinary underwriting, regardless of whether AI is involved.

CFPB's mortgage-document guidance (new tab)

If you receive a request, ask which document, date or entry needs clarification. Provide complete copies through the verified submission channel, keep your originals and confirm receipt. Avoid editing the underlying record to make it look simpler. If the record contains an error, ask how to obtain a correction from its issuer and explain the issue separately.

Make the wire-verification plan before you need it

Your lender's fraud tools and your own payment instructions are different parts of the process. Do not assume a system checking loan information is also protecting a wire you initiate from your bank.

The CFPB's closing-fraud guidance recommends establishing trusted contacts in advance and confirming payment instructions using independently established contact information. Its original guidance predates today's AI tools, but the independent-verification principle also fits the FBI's AI-fraud advice.

CFPB's closing-fraud checklist (new tab) · FBI's verification advice (new tab)

Before closing week:

  • Save your verified settlement contact's name and phone number somewhere you can find them without opening a new email.
  • Ask how you will receive and independently confirm the final payment instructions.
  • Agree that a last-minute change means pausing to verify, even if the message sounds urgent.

Before sending funds, call the number you established earlier and confirm the receiving account information with your trusted representative. Do not use the phone number supplied in the message you are trying to authenticate. An unexpected incoming call does not substitute for that callback.

If money has already gone to a suspected fraudulent account, contact your bank or wire provider immediately, ask about stopping or recalling the transfer, and report it to the FBI through its Internet Crime Complaint Center (IC3). Recovery is not guaranteed.

IC3 (new tab) · CFPB's response steps (new tab)

Keep the people in the process

My advice is to make two plans: one for documenting your loan and another for verifying money movement. Ask your loan officer who can explain an unresolved condition. Ask your settlement team exactly how you will authenticate payment instructions.

AI can be part of the work behind the scenes. Your closing should still come with clear explanations, reachable professionals and time to verify a request before acting on it.

Research checked October 4, 2026. Sources cover different dates and parts of financial services. The opening and account-transfer examples are hypothetical. This article does not claim Nick’s Lending uses Fannie Mae’s fraud platform or any particular AI tool.

Sources and further reading

Sources checked October 4, 2026.

Keep exploring