Guide Family emergency
AI Scams: Voice Cloning, Deepfakes and Your Family
Scammers now use AI to copy a loved one's voice, fake video calls and write flawless messages. How AI scams work, the warning signs, and the family safe word.
By New2h Help Editorial Team Reviewed by New2h Help Editorial Team
Updated Published
Artificial intelligence has not invented new scams. It has made the old ones harder to see through. The panicked call from a grandchild now uses a voice that really sounds like your grandchild. The fake police officer can appear on a video call in uniform. The “friend” you met online has a perfect photo, and their messages no longer contain the spelling mistakes that used to give scammers away.
This guide explains how criminals are using AI, what officials say to watch for, and the simple family habits that defeat even a perfect fake.
How criminals are using AI
In a December 2024 public service announcement, the FBI described four ways criminals use generative AI to commit fraud:
- Text. AI writes believable messages quickly, translates them without grammar or spelling errors, creates large numbers of fake social media profiles, and powers chatbots on fraudulent websites.
- Images. AI creates realistic photos for fake profiles used in romance and investment scams, fake ID cards and law enforcement credentials, celebrity images promoting fake products, and pictures of disasters for fake charities.
- Audio (voice cloning). Criminals produce short clips in a loved one’s voice to pretend to be a relative in a crisis and demand money or a ransom. They also use cloned voices to try to get into bank accounts.
- Video. AI generates videos for real-time video chats with fake company executives, police and other authority figures, and creates misleading promotional videos for investment schemes.
The FBI has since reported more examples. In September 2026 it warned that scammers impersonating law enforcement and government officials use AI to appear as officials on video calls. CISA, the federal cybersecurity agency, notes that poor grammar used to be a clue to phishing — but AI can now produce flawless messages.
The new grandparent scam
The classic version still works the same way: a call from a young relative who has been in an accident or arrested, a request for secrecy, and a second voice claiming to be a lawyer or court official who needs money right now. The Federal Communications Commission says scammers mine social media for names and details, may fake the caller ID, and can now clone the voices of loved ones to make the call more believable.
Caller (sounding like a grandson): Grandma? It’s me. I’m in big trouble. I was in a car accident and the police say it was my fault.
Caller: Please don’t tell Mom and Dad, they’ll kill me. My lawyer is here — he’ll explain.
Second voice (“lawyer”): Ma’am, your grandson needs $9,000 for bail today. A courier can come to your house to collect the cash. This is a confidential court matter, so please don’t discuss it with anyone.
The FTC explains that a scammer needs only a short recording of your family member’s voice — which could come from a video posted online — and a voice-cloning program. That is why “it sounded exactly like him” is no longer proof of anything.
Deepfake video calls from “officials”
Video used to feel like proof. Not anymore. According to the FBI, AI can generate real-time video of fake executives, police officers and other authority figures. Scammers who impersonate law enforcement accuse victims of a crime, threaten arrest, and demand payment by prepaid card, courier, wire, cryptocurrency or cash inserted into a crypto kiosk.
The FBI’s message is clear: law enforcement and government authorities will never contact the public by phone or text to demand any form of payment or to ask for personal or sensitive information, and they will never ask for payment by prepaid card, cryptocurrency or courier.
Fake faces in romance and investment scams
AI-generated photos and videos also help scammers build long online relationships. The FBI says criminals send private photos and videos to convince victims that their online contact is a real person, and create fake promotional videos for investment schemes. If someone you have never met in person brings up money, investing or cryptocurrency, stop — no matter how real their pictures look. Our guides on romance scams and trading scams explain what comes next.
How much it is costing older Americans
In its 2025 annual report, the FBI’s Internet Crime Complaint Center began tagging complaints as “AI related.” That year, people 60 and older filed 3,143 such complaints and reported about $352 million in losses. Because many people never report, the real figure is likely higher.
Can you spot a fake?
Sometimes. The FBI suggests looking and listening for:
- distorted hands or feet, unrealistic teeth or eyes, or faces that look indistinct or irregular;
- strange accessories such as glasses or jewelry, shadows that do not make sense, or watermarks;
- lag time, voices that do not quite match, and unnatural movements;
- a tone or choice of words that does not sound like the person you know.
But do not rely on spotting flaws. AI keeps improving. The habits below work no matter how good the fake is.
Four habits that beat AI scams
1. Hang up and call back. Every agency gives the same advice: do not trust the voice or the face. End the call and contact the person or organization using a number you already have or look up yourself. A real emergency will still be real five minutes later.
2. Use a family safe word. The FBI and FCC both recommend choosing a secret word or phrase that only your family knows. Agree on it at the next family dinner or phone call, and never post it online.
3. Share less in public. The FBI suggests limiting how many images and voice recordings of yourself you post, making social media accounts private, and limiting followers to people you know. Less public material means less for a scammer to copy.
4. Never pay someone you have only met online or by phone. The FBI’s advice: do not send money, gift cards, cryptocurrency or other assets to people you do not know or have met only online or over the phone.
The law is catching up
The FCC says AI-generated voice calls are illegal unless you have agreed to receive them or the caller is exempt. That will not stop criminals, but it means AI robocalls can be reported. Report scam calls and texts to the FTC at ReportFraud.ftc.gov, and online fraud — including AI-assisted scams — to the FBI at IC3.gov.
If you are an adult child reading this, consider calling your parents today to set up a safe word, and share this article with them using the button below.
Red flags: how to spot this scam
- A panicked call or voice message from a "grandchild" or other relative who needs money right away — and asks you to keep it secret.
- A video call with a "police officer," "official" or "company executive" who demands payment or personal information.
- Someone you have only met online sends perfect photos or videos but always has a reason not to meet in person — and eventually brings up money or investing.
- You are asked to pay with wire transfers, gift cards, cryptocurrency, a payment app, or cash handed to a courier.
What to do
- Do not trust a voice or a face on a screen by itself. Hang up or end the video call.
- Call your relative — or the bank, agency or company — back on a number you already have or look up yourself.
- Agree on a family safe word that is never posted online, and ask for it in any emergency call.
- Never send money, gift cards or cryptocurrency to someone you have met only online or over the phone.
- Make your social media accounts private and limit public videos and voice recordings of yourself and your family.
- Report AI-assisted scams to the FBI at IC3.gov and the FTC at ReportFraud.ftc.gov.
Send this to someone you love
A quick share can stop a scam before it starts.
Where to report it
Reporting helps investigators spot patterns and warn others — even if you did not lose money. Type these addresses yourself or use the links below; do not use links from a suspicious message.
-
Federal Trade Commission (FTC)
Any scam, fraud or bad business practice — calls, texts, emails, websites and impostors.
-
FBI Internet Crime Complaint Center (IC3)
Online and computer-related fraud, including tech-support, romance, investment and crypto scams.
-
National Elder Fraud Hotline (U.S. Department of Justice)
Free help for people 60 and older: case managers help you report and plan next steps. Mon–Fri, 10 a.m.–6 p.m. Eastern.
You can also report to your state attorney general. If you sent money, call your bank or card company right away using the number on the back of your card.
Frequently asked questions
What are AI scams?
AI scams use artificial-intelligence tools to make old tricks more convincing: cloning a relative's voice, creating fake photos and ID cards, running real-time video calls with fake officials, and writing error-free messages. The FBI warned about all of these in a December 2024 public service announcement.
How can a scammer copy my grandchild's voice?
The FTC says a scammer only needs a short audio clip — for example from a video posted online — and a voice-cloning program. The result can sound very much like your loved one.
How much money do older Americans lose to AI-related scams?
In its 2025 annual report, the FBI's IC3 recorded 3,143 complaints from people 60 and older that it tagged as AI-related, with about $352 million in reported losses. Many scams are never reported, so the true total is likely higher.
What is a family safe word?
It is a word or phrase that only your family knows, chosen ahead of time and never posted online. The FBI and FCC both suggest it. If a caller in an 'emergency' cannot give the safe word, treat the call as a scam.
Are AI-voice robocalls legal?
The FCC says AI-generated voice calls are illegal unless you have agreed to receive them or the caller is exempt.
Can I spot a deepfake video?
Sometimes. The FBI suggests looking for distorted hands or feet, unrealistic teeth or eyes, irregular faces, odd shadows, lag time and voices that do not quite match the lips. But the safest test is not visual: end the call and contact the person or organization yourself.
Sources
We rely on official government sources. Read how we check facts in our editorial policy.
- FBI IC3 Public Service Announcement I-120324-PSA: Criminals Use Generative Artificial Intelligence to Facilitate Financial Fraud (Dec. 3, 2024) (ic3.gov, accessed )
- FBI Internet Crime Complaint Center: 2025 IC3 Annual Report (Elder Fraud section) (ic3.gov, accessed )
- FBI IC3 PSA I-091726-PSA: Scammers Impersonating Law Enforcement and Government Officials in Fraud Schemes (Sept. 17, 2026) (ic3.gov, accessed )
- FTC Consumer Alert: Scammers use AI to enhance their family emergency schemes (Mar. 20, 2023) (consumer.ftc.gov, accessed )
- FCC Scam Alert: 'Grandparent' Scams Get More Sophisticated (updated Apr. 24, 2026) (fcc.gov, accessed )
- FCC Consumer Guide: Stop Unwanted Robocalls and Texts (fcc.gov, accessed )
- CISA: Recognizing and Reporting Phishing (cisa.gov, accessed )
Related scams
-
Guide Email & phishing
How to Check If a Website Is Legit: 10 Checks
Before you buy, log in or pay a bill online, run these 10 quick checks to tell a real website from a fake one — from the web address to the payment page.
-
Guide Phone scams
Is This Number a Scam? How to Check a Phone Number
Caller ID can be faked, so a number alone can't prove who's calling. Use this checklist to judge a suspicious call or text and find the real number to call.
-
Guide Text message scams
How to Stop Spam Texts on iPhone and Android
Step-by-step settings to filter and block spam texts on iPhone and Android, plus the free tools from AT&T, T-Mobile and Verizon — based on the FCC's guides.
-
Guide Email & phishing
What Is Phishing? A Plain-English Guide
Phishing is when scammers pose as someone you trust to steal passwords, account numbers or your Social Security number. How phishing works and how to stay safe.