AI fakes: deepfakes, cloned voices and synthetic people
Convincing is no longer evidence. Four cues that survive the polish.
What it is
AI can generate faces, voices and whole conversations. It is used for romance scams, fake investment gurus, fake support staff and family emergency calls.
How it works
A short clip of someone's voice is enough to clone it. A photo can be animated. Live video can be faked in real time with the right setup. The output is designed to pass a quick glance, not a careful check.
Warning signs
- An urgent request that avoids a callback
- Audio that is slightly flat, or missing background noise
- Video where blinking, teeth or ear shape look wrong
- A request to move to another app immediately
- Someone who can't answer a personal question only they would know
An example
A grandmother received a call from 'her grandson', sobbing, saying he was arrested and needed bail money. The voice was cloned from a social media video.
What to do
- Hang up and call the person on a number you already have.
- Use your family safe word — a cloned voice can't know it.
- Never send money based on a voice or video alone.
- Ask a question only the real person could answer.
What to teach your family
A voice or face is no longer proof. Our family rule: money requests always get a callback on a known number.
Quick quiz
What is the strongest defence against a cloned voice asking for money?
- Listening for a robot sound
- Asking them to say your name
- Hanging up and calling back on a saved number
- Recording the call
Show the answer
C. Hanging up and calling back on a saved number
Calling back on a known number bypasses the fake entirely.
