Protecting Yourself and Your Family From AI Voice Cloning Scams

Your phone rings. The voice on the other end sounds exactly like your daughter, panicked, saying she has been in an accident and needs money right away. It sounds like her because, in a growing number of reported cases, it is a copy of her voice. A scammer built it from a few seconds of audio taken from her own social media account. This is the basic shape of AI voice cloning scams.

None of this is science fiction. It is a documented scam pattern. Law enforcement and banks track it across the United Kingdom, Australia, Canada, South Africa, and the United States. It also fits a situation many of our readers live. A family member is traveling, and the people who love them are waiting to hear that they are safe.

How AI voice cloning scams work

Scammers need surprisingly little to build a convincing fake. They take a short recording of someone’s voice and feed it into a voice-cloning tool. It often comes from content already online, such as a social media video, a voicemail greeting, or a podcast clip. The tool generates new speech in that person’s voice, saying whatever the scammer types.

Exactly how much audio is required is not settled to one precise figure. The Federal Trade Commission describes needing only “a short audio clip.” In follow-up guidance, it says even less than 30 seconds can be enough. A 2023 Microsoft Research paper looked at a voice-synthesis system called VALL-E. It found that three seconds of sample audio was enough to generate a convincing new voice. Starling Bank, a UK bank covered in more detail below, states the same three-second figure directly on its own consumer guidance page. Whatever the exact number, the practical point is the same. An ordinary trip video, a birthday message, or a voicemail greeting is potential raw material.

The FBI’s own guidance on this technology recommends creating “a secret word or phrase with your family to verify their identity” as the standard defense. It also warns that listeners should not expect to reliably tell a cloned voice apart from a genuine one by ear. The two “can sound nearly identical.” The defense that works is not a sharper ear. It is a verification habit, covered later in this article.

Why travelers’ families are targets for AI voice cloning scams

Family emergency scams are not new. What AI voice cloning changes is how convincing the call sounds, and how well it fits a situation your family already expects.

If you are traveling, the people at home already know you are somewhere unfamiliar. You may be in a different time zone, or hard to reach directly. Say a call opens with “I’ve been in an accident” or “I’m stuck at the airport and lost my wallet.” It does not need to overcome any doubt that you could be in that situation. You already are, or close enough. Cybersecurity researchers and several banks publish consumer warnings on AI voice cloning scams. Their warnings list a travel emergency or being stranded abroad as a recurring cover story. Arrest and car-accident scenarios also come up.

Virtual kidnapping, a related scam

The FBI has also documented a related and more frightening version, called “virtual kidnapping.” In this variant, criminals typically make first contact by text message. They send an altered photo or video as fake proof that someone has taken a family member, then demand a ransom. None of it is real. A separate but related FBI warning covers the audio-only version of the same idea. Criminals use a short cloned clip of a loved one’s voice to demand an urgent ransom or emergency payment by phone. Either way, the FBI’s advice is the same core habit this article recommends. Before doing anything else, check the person is safe through a number or contact method you already have.

There’s a second travel-specific angle here too. Posting real-time trip content, video especially, is exactly the kind of material that supplies a scammer with a voice sample. That does not mean you should stop sharing your travels. It means the protection that matters happens before you ever post anything, covered in Before You Travel below.

What the numbers show about AI voice cloning scams

The FBI’s Internet Crime Complaint Center, known as IC3, added a dedicated section on AI-related fraud to its 2025 Annual Report. It was the first time in the center’s 25-year history. In 2025, IC3 received 22,364 complaints referencing AI across every kind of fraud, totaling close to $893 million in losses.

Inside that total, the category that covers grandparent and family-distress scams recorded just over $19 million in AI-linked losses. IC3 breaks out the specific pattern this article covers even further. It calls them “distress scams,” where criminals use voice cloning to mimic a loved one in an emergency. These cost victims more than $5 million in 2025 alone. The report notes the pattern is spreading beyond grandchildren to mimic other family members and close friends.

You may see a larger figure quoted elsewhere too, $352 million, attributed to AI-related fraud against people 60 and over. That number is real, but it covers every kind of AI-assisted scam that age group reported. That includes investment and romance scams, not only the family-emergency call. The $5 million distress-scam figure is the more accurate one for the specific scam this article is about. That’s the number to keep in mind, rather than the larger total above.

What’s being reported in each market

The pattern below is consistent across SeasonedRoamer’s five markets. What varies is how much each country documents, and how it reports the figures.

United Kingdom

Starling Bank, a UK-regulated bank, runs a dedicated “Safe Phrases” campaign in partnership with the Home Office’s Stop! Think Fraud initiative. Starling’s own guidance states that it takes as little as three seconds of audio for AI to clone someone’s voice. It recommends agreeing a private safe phrase with family in advance, sharing it in person, and never saving it anywhere digital. For balance, Starling also states that, as of this writing, AI voice cloning “isn’t an issue our customers have reported.” It says this even while running an active public-awareness campaign about it.

UK media coverage has also mentioned a National Trading Standards operation. It reportedly blocked a large number of scam calls over a recent six-month period. That figure comes from a single technology blog, not a National Trading Standards announcement we could verify directly. Treat it as a reported claim rather than a confirmed one.

Australia

Scamwatch, run by the Australian Competition and Consumer Commission, confirms this risk on its own site. Scammers can clone a voice from a few seconds of recording and impersonate a friend or family member in distress. Scamwatch’s direct advice is to call the person back on their real number rather than trust the call itself. Consumer group CHOICE reported in late 2025 on Australia’s National Anti-Scam Centre. Since 2022, it had received fewer than five reports of suspected AI voice or video cloning. Even so, major banks list voice cloning among the risks they expect to grow through 2026. The warnings are running ahead of the confirmed case count, and it is fair to say so plainly.

Canada

The Canadian Anti-Fraud Centre recorded Canadians reporting close to C$3 million in losses to family-emergency and grandparent voice-cloning scams in 2024. Its standard advice matches the core protocol below: agree a family codeword in person, and verify through a number you already have before sending anything.

South Africa

SABRIC is the fraud-prevention body run by South Africa’s banking industry. It has warned publicly that criminals can now convincingly clone voices, as part of a wider set of AI-driven fraud. Its consistent advice is to verify before you trust, hanging up and calling back through an official, known number. The same rule protects you from tech support scams, where the caller claims to be from Microsoft or Apple. We could not find a South Africa-specific figure for the family-emergency voice-cloning pattern specifically. SABRIC’s published material focuses more on bank and investment impersonation. We are saying that plainly rather than filling the gap with a number borrowed from elsewhere.

New Zealand

Netsafe, New Zealand’s national online-safety body, covers AI-generated deepfakes broadly and runs a national helpline for anyone targeted by one. New Zealand has a documented business deepfake case, a fake video of a bank chief executive in investment-scam advertising. We could not find a Netsafe resource or reported case on the family-emergency phone call pattern. As with South Africa, Netsafe is still the right place to report any AI-enabled scam, even without a matching statistic for this exact pattern.

The family verification step that stops AI voice cloning scams

Every source in every market above lands on the same core habit, worded slightly differently. Put together, it looks like this.

• Agree a private family code word or phrase in person, never over text, email, or social media. That way there is no digital trace a scammer could find.
• If someone calls claiming to be a family member in trouble, ask for the code word before you do anything else. A cloned voice cannot produce a word it was never given.
• Hang up, then call the family member back on a number already saved in your phone, not a number the caller gives you.
• If you cannot reach them, try another relative or a text message, and do not act until you have confirmed they are safe.
• Treat any request for a wire transfer, gift cards, cryptocurrency, or a cash courier as a warning sign on its own. This holds no matter how convincing the call sounded.

One weakness worth knowing

A family code word only protects you if it stays private. Starling Bank’s own guidance is specific about this. Write your safe phrase down on paper rather than saving it anywhere digital. Change it if you ever think it may no longer be private. That is why every source in this article says to agree it in person or over a phone call. Never agree it in a text, an email, or a shared document.

Before you travel: protect your family from AI voice cloning scams

This is the part that makes the topic worth a dedicated article for a travel readership, rather than a general scam warning. Two small habits, done before you leave, close most of the gap.

Agree the code word with the people staying home before your trip starts, not after a worrying call has already happened. It takes a couple of minutes, and it means everyone already has the tool in place if it is ever needed.

Be thoughtful about what you post while you are away. You do not need to stop sharing your trip. It just helps to know that a long, clear video of you speaking is exactly the material a scammer wants. A private code word makes it useless to them, even if they get hold of it. This pairs well with keeping your bookings and accounts safe while you travel more generally.

If you think you’ve been targeted, or you’ve already sent money

Hang up immediately if you suspect a call is a scam. Do not argue with the caller, and do not explain your suspicions. Just end the call.

If money has already gone, contact your bank or the payment provider straight away. Wire transfers and card payments can sometimes be stopped or reversed if you act fast. Cryptocurrency and cash sent through a courier usually cannot be recovered, but report it anyway, since it helps investigators build a pattern.

Then report the call itself. Where to report depends on where you are.

• United States: report to the FBI’s Internet Crime Complaint Center at ic3.gov, and to the Federal Trade Commission at ReportFraud.ftc.gov. Mention that AI voice cloning was involved, since that detail is specifically tracked.
• United Kingdom: report to Action Fraud, or call 159, the short number that connects you directly to your bank’s fraud team.
• Australia: report to Scamwatch, run by the Australian Competition and Consumer Commission.
• Canada: report to the Canadian Anti-Fraud Centre.
• South Africa: report to your bank directly, and to the Southern African Fraud Prevention Service (SAFPS).
• New Zealand: report to Netsafe.

Do not be embarrassed. These calls are built specifically to bypass the judgment of someone who loves the person being impersonated. That is not a character flaw. It is the entire design of the scam.

For more step-by-step help with your phone and apps, see our Tech for Travelers guides.