Voice cloning tools now need only a short sample—a podcast appearance, a conference keynote, or a video posted by a family member—to produce convincing speech in real time.
Attackers use these voices to call assistants, household staff, relatives, and bankers with urgent requests: a transfer, a password reset, a change to travel plans. The emotional pressure is deliberate.
The defense is procedural. Families and offices should agree on a private verification phrase for sensitive requests, and adopt a firm rule that no financial or access change is completed on the strength of a single call, however familiar the voice.
Reducing public audio and video of principals, especially children, lowers the raw material available. Monitoring for impersonation accounts and cloned media allows takedowns before a campaign gains traction.
None of this requires living differently. It requires a few clear agreements, made in advance, so that the people closest to you know exactly what to do when something sounds right but feels wrong.
How voice cloning scams work
Modern AI tools can produce a convincing imitation of a person's voice from short samples—interviews, earnings calls, podcasts or social-media videos. Criminals pair the cloned voice with publicly available details to call an assistant, family member or finance contact with an urgent request: a payment, a password, a change of travel plans or bail money for a relative.
The FBI has warned that criminals use generative AI to make fraud more believable at scale, including synthetic audio, images and video. Executives and their families are especially exposed because so much of their voice and context is public.
Verification that works under pressure
Agree a private verification phrase within the family and with key staff, and never share it digitally. For any unusual request involving money, credentials or location, hang up and call back on a number you already know. Establish that no legitimate request will ever be harmed by a short verification delay.
For organizations, document which channels can authorize payments or sensitive changes and make sure voice or video alone is never sufficient. Combine this with phishing-resistant multifactor authentication so a cloned voice cannot talk a help desk into resetting an account.
Detecting impersonation early
Monitor for fake social profiles, cloned websites and synthetic media using an executive's name or likeness. Early detection allows takedown requests before the impersonation reaches family, staff or investors. Keep a short response plan: who confirms authenticity, who contacts platforms, and who informs people who may be targeted.
Reduce the raw material
Limit long public audio of family members where practical, and review what children and relatives post. Less available audio makes convincing clones harder to produce.
Frequently asked questions
- How much audio does it take to clone a voice?
- Some tools can produce a usable imitation from short public clips. Exact amounts vary, which is why verification procedures matter more than limiting audio alone.
- What is a family safe word?
- A private phrase agreed in person that family members can ask for during an unexpected urgent call to confirm the caller is genuine.
- How should staff handle an urgent call from an executive?
- Pause, end the call, and call back on a known number or confirm through a second approved channel before acting on any request for money, access or information.
Sources and further reading
- FTC — Voice cloning and family emergency scams
- FBI Internet Crime Complaint Center (IC3)
- FTC Consumer Advice
- FBI IC3 — Criminals use generative AI to facilitate fraud (PSA)
- FTC — Fighting back against harmful voice cloning
- FBI IC3 — Public service announcements
- CISA — More than a password (multifactor authentication)
External resources are provided for general information. This article is not legal, financial or security advice for your specific situation.



