Voice Clone Scams: How They Work and the Family Code Word That Stops Them

The call sounds exactly like your daughter. She is crying, she says she has been in a car accident, and a man who says he is a police officer takes the phone and tells you money has to move now. A few years ago, the giveaway would have been a strange voice or a wrong detail. Today there may be no giveaway at all — because the voice may not be a person imitating your daughter. It may be software copying her. This guide explains how voice clone scams work and the two habits that defeat them, and it sits inside our wider guide to AI scams and deepfake protection.

How a voice clone scam actually unfolds

The FBI’s public service announcement “Criminals Use Generative Artificial Intelligence to Facilitate Financial Fraud” describes the pattern plainly: criminals generate short audio clips containing a loved one’s voice to impersonate a close relative in a crisis situation, asking for immediate financial assistance or demanding a ransom. The US Federal Trade Commission’s consumer alert, “Scammers use AI to enhance their family emergency schemes”, walks through the same scene: a panicked call, a wrecked car, a jail cell, and a voice that sounds exactly like your grandchild, son or sister — because the scammer cloned it from audio posted online. Voice notes, social media videos and recorded greetings are all raw material.

Three ingredients make the script work. The crisis (an accident, an arrest, a kidnapping claim) floods you with adrenaline. The secrecy (“don’t tell Mum, I’m so ashamed”) isolates you from the one person who could disprove it. And the payment rail is chosen so the money cannot come back: the FTC notes that scammers ask you to pay in ways that make it hard to get your money back — wire money, send cryptocurrency, or buy gift cards and hand over the card numbers and PINs. Remove any one ingredient and the scam usually falls apart. The defences below remove all three.

Defence one: the family code word

The FBI’s IC3 puts it in one line: create a secret word or phrase with your family to verify their identity. It is the same logic as a password, applied to the oldest communication channel there is. Done properly, it takes ten minutes to set up:

• Pick a word nobody can guess from social media — not a pet’s name, a street, a school or anything posted online. A random, slightly absurd phrase (“purple bicycle”, “grandma’s soup”) is easier to remember and harder to research.
• Agree who knows it. Parents, children, partners — the small circle a scammer would impersonate or target. Share it in person or on a call you initiate, never in a group chat that could leak.
• Agree the rule, out loud: any call asking for money, bail, gift cards or secrecy gets one question — “what’s the word?” A real family member answers instantly. A clone cannot, because the word exists nowhere online.
• No word, no money — no exceptions for panic. If the caller dodges, cries harder, or says there is no time, that is the answer. Hang up.

The code word works even against a perfect clone, because it tests knowledge rather than sound. Our earlier guide to AI myths and scams makes the same point from the other side: the goal is never to out-listen the fake, it is to make the fake irrelevant.

Defence two: the call-back rule

The FTC’s core instruction is short enough to memorise: don’t trust the voice. Call the person who supposedly contacted you and verify the story, using a phone number you know is theirs. If you cannot reach your loved one, the FTC says to try to get in touch with them through another family member or their friends. The FBI’s PSA gives the matching advice: verify the identity of the person calling by hanging up, researching the contact, and calling the phone number directly.

In practice: end the call politely (“I’m going to call you back on your number”), then dial the number saved in your own phone — never a number the caller gives you, and never by pressing redial on the incoming call. A genuine emergency survives a five-minute verification. Hospitals, police stations and actual grandchildren can all wait five minutes; a scammer working from a script cannot, because every minute increases the chance you talk to someone else. If money has already been requested, also read our step-by-step guide on what to do and where to report — the first hours matter most.

Starve the clone: protect your family’s voices online

The FBI also advises limiting what raw material you publish: if possible, limit online content of your image or voice, make social media accounts private, and limit followers to people you know. That shrinks the pool of audio a scammer can harvest. It does not mean disappearing from the internet — it means treating voice like other personal data. Be cautious with public voice notes, long speaking videos on open accounts, and voice messages exchanged in large group chats. For children and teenagers, who post the most audio and video, a private account and a follower list they actually recognise is a meaningful barrier, because a clone built from three seconds of poor audio is far less convincing than one built from minutes of clean recordings.

Finally, rehearse it. Tell your parents, your partner and your kids exactly what a scam call will sound like — including that it may sound like them. Families who have had the conversation once handle the call calmly, because the script is no longer a surprise. The scammer’s whole business model depends on the call being the first time you have ever imagined it.

What this means

Voice cloning changes something fundamental: for all of human history, recognising a voice was identification. A parent did not need a password to know their child. AI has quietly repealed that, and the FTC’s and FBI’s guidance amounts to rebuilding identification on a new basis — knowledge and behaviour instead of biology. That is exactly how digital security already works everywhere else: your bank does not trust your voiceprint alone, it asks for something you know and something you control. The family code word is two-factor authentication translated into kitchen-table language, and the call-back rule mirrors the oldest rule in business payments — never act on new instructions without confirming them through a channel you already trust.

Compared with the technical race — detection apps, watermarking, carrier filters — these habits win on three counts. They are free, they work today, and they do not degrade as the clones improve, because they never depended on spotting the fake. That matters for US readers in particular: impersonation is not a niche crime. The FTC reported that consumers lost $2.95 billion to imposter scams in 2024 alone, within $12.5 billion reported lost to fraud overall. Behind numbers like those are thousands of individual calls exactly like the one described at the top of this article. Each household that agrees a code word and a call-back rule removes itself from that statistic almost completely — and, just as importantly, protects the family members whose voices are most likely to be borrowed: children whose videos are public, and grandparents whose instinct is to help first and ask later. Spend the ten minutes this week. It is the cheapest insurance your family will ever buy, and unlike an app, it cannot be hacked, cloned or talked out of working.

There is a wider lesson here for every institution in your life, too. Some banks and services have flirted with voice recognition as a security check, on the logic that a voice is unique. Voice cloning breaks that logic completely — a voice is now closer to a username than a password: public, recordable, and copyable. Anything important protected by “it sounded like them” deserves a second factor, whether the “them” is your child on the phone or you, calling your own bank.

Bottom line

Identification has moved from biology to behaviour: a shared secret word and a call-back habit rebuild what cloning destroyed. Set them up this week, while it is a drill and not an emergency.

Read Next

• AI Scams and Deepfake Protection: The Complete Guide
• AI Myths and Scams: What to Believe and What to Avoid

Sources

• US Federal Trade Commission — Scammers Use AI to Enhance Their Family Emergency Schemes (accessed 7 November 2026)
• FBI Internet Crime Complaint Center — Criminals Use Generative Artificial Intelligence to Facilitate Financial Fraud (PSA I-120324-PSA, 3 December 2024) (accessed 7 November 2026)
• US Federal Trade Commission — New FTC Data Show a Big Jump in Reported Losses to Fraud to $12.5 Billion in 2024 (accessed 7 November 2026)