The terrifying rise of 'voice cloning' scams: How hackers can use AI to replicate your voice before placing fake panicked calls to friends or family pleading them to send money

2 months ago
Please Share to your Social Media
Please Follow Naijamerit on Social Media

Our voices are just about as unique as our fingerprints - so how would you feel if your voice was cloned?

A new type of deepfake known as voice cloning has emerged in recent months, in which hackers use artificial intelligence (AI) to simulate your voice.

Famous faces including Stephen Fry, Sadiq Khan and Joe Biden have all already been victims of voice cloning, while one unnamed CEO was even tricked into transferring $243,000 to a scammer after receiving a fake phone call. 

But how does it work, and just how convincing is it?

To find out, I let a professional hacker clone my voice - with terrifying results.

A new type of deepfake known as voice cloning has emerged in recent months, in which hackers use artificial intelligence (AI) to simulate your voice. But how does it work, and how convincing is it? To find out, I let a professional hacker clone my voice - with terrifying results

Voice cloning is an AI technique that allows hackers to take an audio recording of someone, train an AI tool on their voice, and recreate it. 

Speaking to MailOnline, Dane Sherrets, a Solutions Architect at HackerOne, explained: 'This was originally used to create audiobooks and to help people who have lost their voice for medical reasons. 

'But today, it's increasingly used by Hollywood, and unfortunately scammers.'

When the technology first emerged back in the late 1990s, its use was limited to experts with an in-depth knowledge of AI. 

However, over the years the technology has become more accessible and more affordable, to the point where almost anyone can use it, according to Mr Sherrets. 

'Someone with very limited experience can clone a voice,' he said. 

'It takes maybe less than five minutes with some of the tools that are out there which are free and open source.'

When the technology first emerged back in the late 1990s, its use was limited to experts with an in-depth knowledge of AI. However, over the years the technology has become more accessible and more affordable, to the point where almost anyone can use it, according to Mr Sherrets (stock image) 

How to protect yourself from voice cloning scams

Listen for key signs in the audio including pauses, unnatural phrasing and noise in the backgroundDon't be afraid to ask questions that only the real person would actually knowSet-up a safe word with your friends and family 

To clone my voice, all Mr Sherrets needed was a five-minute clip of me talking. 

I opted to record myself reading out a Daily Mail story, although Mr Sherrets says that most hackers could simply lift the audio from a quick phone call, or even a video posted on social media. 

'It's possible to do it during a call, if there's something shared on social media, or even if someone's on a podcast. Really just things that we're uploading or recording every day,' he said. 

Once I'd sent Mr Sherrets the clip, he simply uploaded it into a tool (which he has chosen not to name), which could then be 'trained' on my voice. 

'Once that was done, I was able to type in, or even speak directly into the tool, and have it output whatever I wanted the message to be in your voice,' he said. 

'What's really crazy with the tools that are out there now, is I'm able to add extra inflections, pauses, or other things that make the speech sound more natural, which makes it a lot more convincing in a scam scenario.'

Despite not featuring any pauses or added inflections, the first clip of my voice clone that Mr Sherrets created was startlingly convincing. 

The robot voice perfectly nailed my American-Scottish hybrid accent, as it said: 'Hey Mum, it's Shivali. I've lost my bank card and need to transfer some money. Can you please send some to the account that just texted you?'

However, the creepiness was taken up a notch in the next clip, in which Mr Sherrets added pauses. 

'Towards the end you can hear a long pause, and then a breath, and that makes it sound a lot more natural,' the professional hacker explained. 

While my voice cloning experience was thankfully just a demonstartion, Mr Sherrets highlights some of the serious dangers of the technology. 

'Some people have had fake kidnapping calls, where their "child" has called them, saying "I've been kidnapped, I need millions of dollars or they won't release me," and the kid sounds very distressed,' he said. 

'What we're seeing today, increasingly, are folks trying to do more targeted social engineering attempts against companies and organisations.

'I used the same technology to clone my CEO's voice. 

'CEOs often have public appearances, so it's very easy to get high-quality audio of their voice and clone it.

The robot voice perfectly nailed my American-Scottish hybrid accent, as it said: 'Hey Mum, it's Shivali. I've lost my bank card and need to transfer some money. Can you please send some to the account that just texted you?'

Voice cloning is an AI technique that allows hackers to take an audio recording of someone, train an AI tool on their voice, and recreate it

'Having a CEO's voice makes it a lot easier to get a quick password, or access to a system. Companies and organisations need to be aware of that risk.'

Thankfully, Mr Sherrets says there are several key signs that indicate a voice is a clone. 

'There are key signs,' he told MailOnline. 

'There are the pauses, the issues where it doesn't sound as natural, and there might be what you call "artefacts" in the background.

'For example, if a voice was cloned in a crowded room and there's a lot of other people chatting, then when that voice clone is used, you're going to hear some garbage in the background.'

However, as the technology continues to evolve, these signs will become trickier to spot. 

'People need to be aware of this technology, and constantly be suspicious of anything asking them to act urgently - that's often a red flag,' he explained. 

'They should be quick to ask questions that maybe only the real person would actually know, and not be afraid to try and verify things before they take any action.'

Mr Sherrets recommends having a 'safe word' with your family and friends. 

'If you really are in an urgent situation, you can say that safe word and they'll instantly know that this is really you,' he said. 

Finally, the expert advises being aware of your digital footprint, and keeping an eye on the amount you upload online. 

'Every time I upload now, it expands my audio attack surface and could be used to train AI later,' he added. 

'There's trade-offs to that that everyone will need to make, but it's something to be aware of - audio of yourself that's floating out there can be used against you.'

Read full article
Please Follow Naijamerit on Social Media
< Back | News content