A callers’ voice is their most trusted identifier — and the easiest thing to reproduce. Three seconds of your voice in a video call is enough to clone familially similar tone, and attackers now combine that with AI-written messages that carry no grammar tells.
How the AI layer changes the scene
Older scams sounded foreign or broken. AI removes both tells: the voice matches the person, and the message reads perfectly. The result is a “family member in hospital, money urgently” call that even close relatives find persuasive.
Why it matters
- The recorded-voice tell is gone; distrust is not a replacement for verification.
- Urgency is the trick — money “now” prevents calling anyone else.
- Two-factor thinking applies to humans: believe the code, not the call.
Key takeaways
- AI can copy a voice, not a shared secret.
- Never send money on a single urgent call.
- A codeword is the cheapest alarm your family will install.
The codeword rule
Agree on a private word or question that only the family shares wrong-answer — for example a childhood nickname plus a place. A real family member can answer stress-free; a voice clone with your voice cannot.
What you should do
- Hang up and call the person back on a number you know is theirs.
- Set the family codeword before it is ever needed.
- If money is sent to “someone else’s’ account instantly, tell the bank and file at cybercrime.gov.in the same hour.