Photo by Timur Repin on Unsplash
The call comes from an unknown number. It’s your kid’s voice, upset and talking fast: they’re in trouble, they need money now, please don’t tell anyone. Most parents would believe it. That’s the point. The voice may not be your kid at all. It can be generated from a short clip of them talking, and the tools that do it are cheap and widely available.
The scammer needs two things: a sample of someone’s voice and a person who loves them. The sample can come from a public video, a gaming stream, a voice note, or a few seconds of a kid answering a call from an unknown number. Software turns that into a voice that can say anything the scammer types.
The script is the old “family emergency” scam: an accident, an arrest, a lost phone, a kidnapping. What’s new is that it now sounds like the right person. The FBI describes criminals generating “short audio clips containing a loved one’s voice” to fake a crisis and ask for money or a ransom (IC3 public service announcement).
For the first time, the FBI’s annual internet crime report, published in April 2026, has a section on artificial intelligence. In 2025, the FBI’s Internet Crime Complaint Center received 22,364 complaints involving AI, with reported losses of more than $893 million. That covers every AI-assisted scam, not just voice calls. The report singles out “distress” scams, where cloned voices imitate a loved one in trouble: victims reported more than $5 million in losses to those alone, and the FBI notes the scam is spreading to imitations of other family members and friends (FBI IC3 2025 Annual Report, p. 39).
Those are only the cases people reported. Many families who got the call and hung up in time never file anything.
A cloned voice works best on a phone call. The call is live, so there’s no time to think. The audio is slightly compressed, so small glitches sound like a bad line. Caller ID proves nothing, because numbers can be spoofed, and the script often explains an unfamiliar number anyway (“my phone died, I’m using a friend’s”).
Kids can be targets too, not just the source of the voice. A younger child can get a call that sounds like a parent asking them to read out the code that just arrived by text. That code is often the key to an account. We covered where that leads in our posts on SIM swap attacks and smishing.
Paxio can’t detect a cloned voice, and it doesn’t listen to calls. By design, it doesn’t read texts or calls at all, and phone calls always stay available, even during a screen-time limit. The code word and the callback habit above are the real defence, and they cost nothing.
What Paxio does cover is the part before the call: where your kid’s voice ends up. With App Blocking, you decide which public video and livestreaming apps are on a younger child’s phone at all, so the decision about a public voice sample is one you make together rather than something that just happens.
Yes. The FBI describes scammers using short audio clips of a loved one's voice to fake an emergency call. Public videos, voice notes and livestreams are the usual sources, which is why keeping a child's accounts private helps.
Often you can't, just by listening. Instead, ask for your family code word, then hang up and call your child back on the number you already have. Pressure to pay now, keep it secret, or use gift cards or crypto is a strong warning sign.
Contact your bank or payment provider straight away to try to stop the transfer, then report it. In the US, file a report at ic3.gov. Elsewhere, use your national cybercrime reporting service, such as cybercrime.gov.in in India.
See why families choose Paxio if deciding which apps are on a younger child’s phone is the piece you’re missing.