Voice cloning scams: what they sound like and how to protect your family

The scam, in one paragraph

Your phone rings. It is your daughter's voice, or your father's, panicked: an accident, an arrest, a kidnapping. Someone takes the phone and demands money fast, by wire, crypto, or gift cards. The voice was cloned from a few seconds of audio scraped from social media or a voicemail, the caller ID may be spoofed, and the entire design of the call is to make you act before you think.

This is not an emerging threat; it is an operating business. The defenses are cheap and they work, which is why this page exists on a voice platform's own docs.

The call that reached Congress

In January 2023, Jennifer DeStefano of Arizona answered a call that opened with her 15-year-old daughter's voice crying for help, followed by a man staging a kidnapping and demanding a ransom that started at one million dollars. Her daughter was safe the whole time; the voice was synthetic. DeStefano told the story under oath in written testimony to the US Senate Judiciary Committee in June 2023.

Her case is the reference point for this whole category, and it is one reason children's voices do not exist on Cantari at all.

What the regulators say

The FTC's consumer alert on AI family-emergency schemes (March 2023) describes the playbook above and the two-step defense: call the person back on a number you know is theirs, and treat wire transfers, crypto, and gift cards as the signature of a scam.

The FBI's Internet Crime Complaint Center followed with a public service announcement (December 2024) on criminals using generative AI for fraud, including short cloned clips of a loved one's voice in crisis and ransom calls. Its standout advice is the cheapest security control in this entire field: agree on a secret word or phrase with your family to verify identity.

The FTC also ran an open Voice Cloning Challenge, announcing four winning approaches in April 2024 spanning clone detection, watermark-style protections, and provenance checks. Worth knowing about, and not yet something to rely on: detection remains a moving target.

The defenses that actually work

None of this requires technology. That is the point: the countermeasure to a synthetic voice is a verification habit, not a better ear. Studies and the regulators agree that humans cannot reliably distinguish good clones by listening.

• Set a family passphrase this week. A word no stranger could guess, known to the people who might get the call. Asking for it ends the scam instantly.
• Hang up and call back on the number you already have for that person. Not the number that called you.
• Let urgency raise suspicion instead of lowering it. Manufactured time pressure is the scam's engine; real emergencies survive a two-minute verification.
• Treat payment by wire, crypto, or gift cards as a tell. No hospital, police department, or kidnapper-of-record bills that way.
• Talk about this with the most-targeted people in your family before the call comes. Grandparents are the named audience of half these advisories.

What this means on Cantari

Our side of the fence: there is no anonymous cloning here. Creating a clone requires a signed-in account, the server refuses the create call without the consent attestation, and the time you gave it is stored with the voice. Using Cantari for fraud or impersonation breaches the terms, community content runs through a moderation pipeline that fails closed, and clips published here always credit their engine and voice rather than posing as recordings.

And the honest limit: no voice platform, ours included, can promise its technology will never touch a scam, because scammers do not need any particular platform and detection tools are imperfect. What we can promise is a design where misuse requires lying to us in writing, plus this page, which is the part that protects your family regardless of whose tools the scammer uses.