As artificial intelligence tools become more accessible, cybercriminals are abandoning the clumsy, typo-ridden scam emails of the past in favour of a far more convincing technique: voice cloning. Using AI tools, fraudsters can now reproduce a real person's voice with startling accuracy, and cybersecurity experts warn the tactic is only going to become harder to spot.
A changed approach to unknown calls
The threat has already altered how some professionals handle their phones. Financial journalist Rachel Rickard Straushas adopted a personal rule: when she receives a call from an unknown number, she no longer speaks first. Instead, she waits to hear the other person, since even a few seconds of her own recorded voice could be enough for someone to generate a synthetic copy.
How the scam works
The method itself is straightforward. Scammers obtain a short audio sample, taken from a phone call, a video, or a social media post, and feed it into an AI application capable of generating a synthetic voice that sounds nearly identical to the original. They then contact the victim's relatives or close friends, claiming there has been a serious accident, an urgent situation, or an immediate need for money. The combination of time pressure and emotional distress often pushes victims into hasty decisions before they can verify whether the call is genuine.
A fast-growing and low-cost threat
Experts caution that these attacks will keep becoming more effective as AI tools improve, while the falling cost of using them makes large-scale fraud easier to carry out. In the United Kingdom alone, more than four million cases of financial fraud were recorded within a single year, with total losses reaching approximately £1.3 billion.
The old warning signs are disappearing
Spelling mistakes, clumsy phrasing and obviously suspicious websites, once the clearest signs of a scam, are being replaced by polished text, convincing emails and websites that are difficult to distinguish from the real thing.
Protecting yourself and your family
Experts recommend that families agree in advance on a secret word or phrase known only to close relatives. If someone calls asking for money or help, using that word can serve as an additional layer of identity verification.
It is equally important to avoid making any financial transaction under pressure. Calls that claim urgency, demand strict secrecy, or suggest unusual payment methods should be treated with particular caution. The safest response is always to contact the person supposedly in need directly, using their known phone number, or, in the case of a bank or public service, through official communication channels.
Older advice, such as trying to keep a scammer talking for longer, is now considered counterproductive. The longer a conversation lasts, the greater the risk that additional audio is being collected for use in future attacks.
Experts' core message is simple: any suspicious contact calls for a calm response, independent verification, and avoiding sharing personal data or sending money. In the age of artificial intelligence, confirmation from a second, trusted source remains the most effective line of defence against increasingly sophisticated digital fraud.
