Fraudsters need just 3 seconds on a cold call to clone your voice… and scam your family

Officials at the Bank of England are so concerned about fraudsters using artificial intelligence technology for scams in the UK that it hosted an urgent summit for big banks and tech firms last week, Money Mail can reveal.

Leading banks and fraud prevention experts fear advanced technology could open the door to a wave of new scams.

Insiders say Bank of England governor Andrew Bailey opened the meeting by warning of the growing risks the cutting-edge technology poses to the financial future of British households and called for an urgent response to counter the threat.

Executives at Google and OpenAI (the firm behind ChatGPT and co-founded by Elon Musk), and social media giant Meta (which owns Facebook, Instagram and WhatsApp) were asked about their safety measures to stop the software from falling into the wrong hands.

Tech firms were warned scams will ‘get much more sophisticated’, but Britain ‘still has a window to shore up its defences’.

Fraud threat: Leading banks and fraud prevention experts are fearful new AI technology coming to our shores and opening the door to a wave of new scams

The UK is grappling with a tsunami of fraud, costing consumers more than £1.2 billion last year, according to trade body UK Finance.

But the number of attacks could soon surge out of control as criminals begin to incorporate the use of AI to make their operations far more sophisticated.

Anti-fraud ‘champion’ and MP Anthony Browne, who was at the summit, says AI scams are on their way. 

He says: ‘The Government is taking the risks associated with AI very seriously. It is clearly a threat we need to prepare for.’

Banks are beginning to deploy a host of AI-powered chatbots to combat fraudsters in a trial led by Stop Scams UK, the cross-industry group behind last Friday’s summit.

A group made up of banks, tech and telecom companies is piloting a project to gather intelligence on scammers.

An estimated 300 phone numbers and 100 email addresses will be operated by members of the Stop Scams UK group, which includes Britain’s biggest High Street and challenger banks; telecoms BT, TalkTalk and Three; and tech companies Meta and Google.

These numbers and accounts will be assigned ‘chatbots’ — computer programmes that simulate human conversation. They are often used for customer service queries as they can answer questions with great accuracy.

Because they can mimic human messages, criminals have been known to use chatbots to manipulate victims and impersonate trusted authorities or companies.

The chatbots can send messages on social media platforms that appear genuine and trick victims into handing over money or personal details by pretending to be someone they know and trust.

Paul Davis, director of fraud prevention at TSB, says these chatbots allow criminals to scale up their operations, with large numbers of bots messaging multiple victims simultaneously.

However, the banks’ chatbots will be pitted against them in the trial. ‘There’s an opportunity to delay chatbots in order to disrupt criminals,’ Mr Davis says.

‘We might end up with two AI chatbots talking to each other — one controlled by the banks, the other by criminals. They can waste each other’s time.

‘It’s a numbers game. Fraudsters can automate large parts of the process where they would have had to use a human before. That means the number of attacks will increase.’

Because they can mimic human messages, criminals have been known to use chatbots to manipulate victims and impersonate trusted authorities or companies.

Alex West, director of banking fraud and investigations at PwC, which provided research at the AI summit at the Bank of England, says the technology is providing a powerful set of tools to detect scams and track down fraudsters. However, consumers should beware of the high credibility of these scams.

Mr West says: ‘AI tools can be used by fraudsters to create highly convincing scams. AI-generated messages, cloned voices and deep-fake videos are increasingly being used to con people out of money.’ 

Natalie Kelly, chief risk officer, Visa Europe, which is also a member of Stop Scams UK, says: ‘For every AI tool available on the world wide web, there are criminal versions. Criminals have adapted quickly to adopt AI.

‘It is a constant battle and we are laser-focused on our AI always being better than the fraudsters.’

Fraudsters can use AI in a number of ways to fool their target and typically use social media platforms to do so.

Money Mail’s Stop the Social Media Scammers campaign is calling on tech firms to do more to protect users and prepare for the new threats.

One of the fastest-growing techniques employed by fraudsters is to use cheap online software to clone a voice. 

By doing this they can impersonate a loved one or family friend and lure unsuspecting victims into handing over their money. 

For example, criminals may make a phone call using a cloned voice to make it sound as if a loved one is in trouble and needs financial help.

All you need is three seconds from a voice recording or video to replicate someone’s voice, according to research by ant-virus firm McAfee.

Imposter: Criminals are making a phone call using  cloned voice to make it sound as if a loved one is in trouble and needs financial help

Anyone can clone a voice and can dictate the text that they want the voice to say in an audio clip. Money Mail took this to the test on Play.Ht, an AI-voice generative start-up, using its 5,000-words-a-month free trial.

The website allows you either to clone someone’s voice using a 30-second audio clip or to use an existing AI voice. The range of voices are AI generated, but are realistic and can be set to speak almost every language in the world.

Once you have uploaded an audio clip, you write in a text box the words you want the cloned voice to say. After a short wait, you can play and download a new audio clip of the voice saying it.

The technology is becoming so advanced that you can prompt the voice’s emotions: for example, to sound happy, sad or worried.

In one high-profile case in the U.S. a mother became distraught when scammers used AI to fake her daughter’s kidnap.

The woman received a call from an unknown number but when she picked up, she believed it was her 15-year-old daughter, who was away training for a ski race. Fraudsters had used the software to generate a clip of the teenager’s voice crying for help, before demanding a $1million ransom.

The scam has already made its way to the UK, according to research from McAfee.

Oliver Devane, a security researcher at McAfee, says: ‘In most of the cases we’re hearing of, the scammer says they have been in a bad accident, such as a car crash, and need money urgently.’

Chris Ainsley, head of fraud strategy at Santander, warns that criminals typically pose as bank staff, the police and other trusted authorities as part of impersonation scams. Action Fraud, the UK’s official fraud hotline, says it does not hold reliable data on AI voice-cloning scams, as most people do not realise AI has been used.

However, there were 82 reported scams involving AI between May 2019 and August 2023.

NatWest tells Money Mail that AI could be used in a number of ways to target customers and that it is increasingly important that consumers stop and take a moment before making any payments that might appear unusual.

Liz Edwards, of comparison site Finder.com, says it is crucial to concentrate on what is being said whenever you watch a video or receive a phone call.

Ms Edwards says: ‘Although these videos look realistic, a lot of the content will have been written by AI technology, so listen carefully and try to decipher whether it really sounds like something a human would say.’

• Have you been the victim of an AI scam? Write to [email protected]

This post first appeared on Dailymail.co.uk