Your buyer is three days from closing. Their phone rings. The number on the screen matches the title company. The voice on the line is your closing officer — same warmth, same little verbal habits — calmly explaining that the wiring instructions have changed and the funds need to go to a new account today.
It is not your closing officer. It is software.
That scenario is no longer a hypothetical from a cybersecurity conference. In 2026 it is one of the fastest-growing versions of real estate wire fraud, and it is built on two cheap, widely available tools: AI voice cloning and deepfake email. If you have spent years telling clients “just call a known number to verify,” this article is the update to that advice — because the thing answering the known number can now be faked.
Why this lands on your desk, not just the title company’s
You may not touch the wire. But you are the person your client trusts most in the transaction, and you are the one they will call first when six figures vanish. Wire fraud is rarely recoverable; the FBI’s Internet Crime Complaint Center (IC3) urges victims to act within hours because once funds are gone, most buyers never see them again. The emotional fallout — a family’s down payment erased days before move-in — attaches to your name and your brand whether or not you were at fault.
So this is a fiduciary issue, a reputation issue, and a referral issue. Agents who get ahead of it look like the most prepared professional their client has ever worked with. Agents who stay quiet hope it never happens on their watch.
What actually changed: the old red flags are gone
For a decade, we taught clients to spot fraud by looking for clumsiness — broken English, odd email addresses, a sense of panic and urgency. AI erased those tells. The National Association of Realtors now publishes a consumer guide to spotting deepfake scams precisely because the warning signs we relied on stopped working.
- The typos disappeared. AI-generated emails are clean, on-brand, and written in the sender’s own tone. The grammar-mistake test no longer works.
- The voice can be cloned. Security researchers have shown a usable voice clone can be built from only a few seconds of audio — a snippet from a voicemail greeting, a podcast, a social video, or a recorded closing call.
- The caller ID can be spoofed. The number that shows up can be the real office line, which is exactly why “I called the number I had” is no longer proof of anything.
The defense that used to be solid — call the number you know — now has a hole in it, because criminals can sit on the other end of that number, sounding like someone you trust.
The numbers, briefly
The scale is why this is worth a blog post and not a shrug. The FBI’s IC3 has tracked hundreds of millions of dollars in annual losses tied to real estate and business-email-compromise fraud, and the agency considers it one of the costliest cybercrime categories. You do not need to memorize the figures to act on them. The takeaway is simple: this is common enough that assuming “it won’t happen to my client” is no longer a reasonable bet. When you share numbers with clients, point them to the FBI directly at ic3.gov rather than a secondhand headline — cite the source, not the rumor.
The protocol you can run on every single deal
Here is the part to actually use, and the reassuring news is that it is endorsed by the people who study this for a living. Both the Consumer Financial Protection Bureau and the FTC recommend a simple verification habit — and a code phrase — set up before anyone is rushed. None of it requires new software.
1. Set a verbal code word at the kickoff. At the very start of the deal, agree with your client (and ideally the title company) on a simple shared phrase. Any real call about money must include it. The CFPB explicitly suggests identifying trusted contacts and a verification method up front; a cloned voice on a spoofed number will not know your code word.
2. Treat every “change” as a red flag until proven otherwise. Legitimate wiring instructions almost never change at the last minute. A sudden change of account, urgency, or “do it today” is the single most common signature of fraud. Slow it down on purpose.
3. Verify out-of-band, not on the channel that contacted you. If the request came by email, do not call the number in that email. Look up the title company’s number from an independent source you trust — the signed contract, the company’s official website — and confirm there.
4. Never trust wiring details sent by email alone. Email is the easiest channel to compromise. The NAR wire fraud resources say it plainly: confirm the account number and routing number verbally, person to person, and have them read back to you.
5. Brief the client before they are under pressure. Hand them these rules at the start, in writing, while everyone is calm. Tell them plainly: “No one legitimate will ever rush you to wire money to a new account. If that happens, stop and call me first.”
If it happens anyway, the first few hours decide everything
Speed is the only real lever. If a client suspects they have been defrauded:
- Call the bank’s wire or fraud department immediately and request a wire recall. Within hours, recovery is sometimes possible; after that, it is rare.
- File a report with the FBI at ic3.gov and report the scam to the FTC at reportfraud.ftc.gov right away.
- Notify the title company and the receiving bank so accounts can be flagged.
Save the panic for later. The phone calls come first.
The takeaway
The tools criminals use got dramatically better; the defense did not have to get more complicated, just more deliberate. A code word, a healthy distrust of last-minute changes, and a verify-before-you-wire habit will stop the overwhelming majority of these attacks. In an AI era where a voice can be faked in seconds, the most valuable thing you offer a client is not a faster transaction — it is a safer one.
If you found this useful, save it and share it with your next buyer at the start of their deal, not the end. The best time to talk about real estate wire fraud is before there is any money in motion.
This post was written entirely by Claude.ai, Anthropic’s AI. From a single prompt by Michael George — “write me a blog post” — Claude logged into ActiveRain, researched the topic, wrote this article, the headline, the SEO title, and the meta description, and published it. We’re sharing that openly because this piece is about telling what’s real from what isn’t: the threat is real, every statistic links to a primary authority (the FBI, NAR, CFPB, and FTC), and Michael reviewed it before it went live.
Comments(6)