A Reporter Cloned Their Voice and Called Their Mum

The reporter didn't use anything exotic. No studio, no months of training data. Just widely available AI voice cloning tools and a short audio sample, and the result was convincing enough to fool someone who had known that voice since birth.

That detail matters. A mother who has heard her child's voice for decades, in every mood and every situation, still couldn't tell. If a reporter can do this on a Tuesday afternoon with free software, so can anyone else. The technology is here, it is cheap, and ordinary families are the target.

How the Call Works

Voice cloning now requires as little as 20 to 30 seconds of audio, which can come from a voicemail, a social media video, a podcast, or a clip your child posted last week. Once the clone exists, the scam follows a familiar script: the voice on the line sounds exactly like your son or daughter, they're in trouble, they need money fast, and they tell you not to call anyone else yet. The urgency is deliberate. Scammers know that if you have time to think, the con falls apart.

The scale is already large. The Federal Trade Commission recorded $2.7 billion in losses from imposter scams in 2023, before the current wave of AI voice tools became widely accessible. The World Economic Forum reported that deepfake fraud attempts increased 500% in 2024. In February 2024, engineering firm Arup lost $25 million when an employee joined a video call where every participant was a deepfake. That was a corporate target with IT resources. Ordinary families have none.

Why Your Ears Cannot Save You

Human detection accuracy for cloned voices sits at around 55 to 60 percent, barely better than a coin flip. Telling someone to "listen more carefully" or "be more skeptical" is not a realistic defense. The voice sounds real because it is real, just replayed and repurposed without permission. We recognize voices the way we recognize faces: automatically and emotionally. Scammers exploit that.

You need a different kind of signal, one that AI cannot copy.

This is where Trust Onion's three rotating codewords come in. Your family shares three simple words that rotate every few hours, calculated locally on each person's phone with no server involved. When a call arrives from someone claiming to be your child, you ask one question: "What are the words?" A scammer using a cloned voice cannot know them. There is no audio sample, no social media post, no database that hands a fraudster words that changed two hours ago and exist only on your family's phones.

If you hang up and call back on a number you know independently, a failed callback still doesn't prove the original caller was legitimate. The codewords remain the cleaner test: whoever is on the line must know your family's private Trust Onion Three Words.

The Cybernews reporter's mother heard a perfect copy of her child's voice and was fooled. If her family had been using Trust Onion, her next sentence would have been "What are the words?" and the scam would have ended right there.

A controlled voice-cloning safety demonstration

Try our tool to see just how easy voice cloning is. This controlled safety demonstration exists only to show how easy voice cloning is and to make the risk tangible. It is not a real incident or proof of anyone's identity. Your submitted sample is handled transiently to generate the fixed demonstration audio. Trust Onion does not identify you or the person from the sample. Trust Onion does not store the submitted recording, clone, or generated audio.