AI tools can now clone a child's voice from just a few seconds of audio — typically pulled from social media — and use it to impersonate them in phone calls to family members.
Voice cloning scams have grown rapidly. A perpetrator extracts audio from a child's TikTok, YouTube, or Instagram, generates a synthetic version of their voice using an AI tool, and then phones a grandparent or other relative claiming the child has been kidnapped, arrested, or had an accident — and demanding immediate payment.
Public-facing video and audio clips are the source material. Modern voice cloning models need very little audio to produce a convincing replica. The synthetic voice is then used in real-time over a phone call, with the scammer typing prompts that the AI speaks aloud in the cloned voice.
1. Agree a family safe word
Choose an unusual word or phrase known only to your immediate family. Use it to verify identity in any phone call involving urgency, money, or distress. The voice may sound real but the safe word will not be known to the scammer.
2. Limit public-facing audio of children's voices
Set TikTok, YouTube, and Instagram accounts to private where possible. Avoid public videos containing extended clear speech by the child. Consider whether your child's voice needs to be in any public-facing content.
3. Educate family members about the scam pattern
Grandparents, aunts, and uncles are common targets. Make sure they understand that if they get a panicked call, they should hang up and call the child directly on a known number — never act on the call alone.
ตรวจสอบล่าสุด: 2026-04-19
เนื้อหาต้นฉบับภาษาอังกฤษ: /by-risk/audio-deepfakes-voice-cloning