🎯 In 30 seconds: A few seconds of someone’s voice from a video online is enough for AI tools to make it say anything. So the voice, the face on a video call and the caller ID can all be fake. What still works: a family safe word, hanging up and calling back on a number you already know, and a rule that no money moves on the strength of one call.
How the scams use AI
The FTC warns that a scammer can clone a loved one’s voice from a short clip posted online. The FBI’s December 2024 alert on generative AI lists the ways criminals now use it:
- Voice clones of a child or grandchild “in trouble” who needs bail, a lawyer or hospital money.
- Fake video calls where a “boss” or “relative” appears on screen just long enough to approve a payment.
- Fake officials: AI audio and video of people posing as police, government staff or company executives.
- Fake documents and profile photos used to open accounts or to make a romance or investment profile look real. FinCEN, the Treasury’s anti-money-laundering bureau, issued an alert to banks in November 2024 about deepfake ID documents used to get past identity checks.
In February 2024 the FCC ruled that AI-generated voices count as “artificial” voices under the Telephone Consumer Protection Act, so robocalls using them without your prior consent are illegal. That helps enforcement. It doesn’t stop the calls from reaching you.
The verification habits
| Family safe word | Pick a word or phrase that isn’t online anywhere (not a pet’s name you post about). Anyone asking for money by phone must say it. The FBI recommends exactly this. |
| Hang up and call back | Call the person on the number saved in your phone, not the one that called you. If you can’t reach them, call another relative or friend who would know. |
| Ask a question only they know | Something not on social media: what you ate together last week, the nickname for a grandparent. |
| Slow it down | Real emergencies survive a ten-minute delay. A demand for secrecy (“don’t tell Dad”) is a tell. |
| Two people for big payments | For a small business: any change to payment details, or any payment above a set amount, needs a second person and a call-back on a known number. |
| Share less | Make social accounts private and limit public video of your voice, as the FBI suggests. |
What your bank will never do
- Ask you to read back a one-time verification code. The FTC says anyone who asks for that code is a scammer.
- Tell you to move money to a new account “to protect it”.
- Ask you to withdraw cash and hand it to a courier, or buy gift cards or crypto.
- Get upset when you say “I’ll call you back on the number on my card.”
A worked example
At 9:40 p.m. your phone shows your daughter’s name. It is her voice, crying: she hit a car, the other driver is hurt, a “lawyer” takes the phone and says $8,500 in cash will be picked up within the hour so she isn’t held overnight.
Here is how the habits play out. You ask for the safe word; the voice says she’s too upset to remember. You say you will call right back, hang up, and dial her saved number. She answers from her dorm, fine. Total time: three minutes. Without the habit, $8,500 leaves in a courier’s hand and cannot be traced back.
Spotting a fake on video
The FBI suggests looking for small flaws: distorted hands or feet, odd teeth or eyes, lighting that doesn’t match, and voices whose tone or word choice is off. These get harder to see every year, so don’t rely on spotting them. Rely on the call-back.
If you were targeted, report it at ReportFraud.ftc.gov and IC3.gov. Agree on your family’s safe word today; the Money Theft & Forensics page has the full family checklist.
Sources
- FBI IC3: Criminals use generative AI to facilitate financial fraud (Dec 2024)
- FTC: Scammers use AI to enhance their family emergency schemes
- FCC: AI-generated voices in robocalls are illegal
- FinCEN: Alert on fraud schemes involving deepfake media
- FTC: What’s a verification code and why would someone ask me for it?