There was a time when if you saw a photograph or a video with your own eyes, it could be trusted.
While people previously had to be wary of phone calls impersonating their bank or doorstep scams, they at least knew the videos of public figures they saw on television or on their mobile phones were real.
But new technology means almost anyone can create videos known as deepfakes which are designed to pilfer your cash.
Fraudsters are now using AI to create convincing likenesses of famous people – and they are becoming increasingly hard to spot.
Money comparison website Finder showed 2,000 people eight five-second clips – some real, some fake – of influential celebrities and public figures including Dua Lipa, Kim Kardashian, Elon Musk and Jeff Bezos.
They asked them to identify which ones were real, and which were AI fakes.
The fake clips were so convincing that more than 96 per cent of people got at least one wrong.
You can take the test below to see how many you would get right – and then read our tips on how to identify a fake scam video.
Could YOU spot a deepfake? Finder asked 2,000 people to sort scams from genuine videos of celebrities – and it did not prove to be easy
What is a deepfake?
Deepfakes are videos, photographs or audio clips which are altered or created using artificial intelligence.
Deepfake technology uses existing audio clips, videos and photographs of a person to create videos and photographs showing them doing or saying things which they never did.
A person’s likeness is recreated convincingly in a deepfake using machine learning. This is where a computer will have been trained with hours of existing footage of that person, says Finder.
The technology can manipulate an existing video or create a new one entirely. And it’s not limited to those with expert technology skills, as they can be created using free software available online.
While this technology does have benign uses, such as creative work or making funny videos for social media, fraudsters have latched on to the technology and are using it to steal money from victims in cruel new scams.
Real or fake? Take Finder’s test yourself below to see how well you can spot an AI spoof
How do deepfake scams work?
Some one in five people have encountered a deepfake or AI-related scam, Finder says, and one in three of these people fell for the ruse.
It means as many as 3.4 million people across the country may have been lured in by these altered videos and photographs.
The scams often start on social media, where a video may show celebrity, daytime television star or other public figure endorsing a new investment which ‘could make you a lot of money’.
But the investment isn’t real – and the celebrity hasn’t endorsed it.
Instead, it’s a fake video which has been made up by a scammer. They are pulling the strings and luring viewers to ‘invest’ their money in a fake scheme.
Once you respond to the advert, you’ll be lured to invest some of your money. They will often show you fake screenshots and account details with your soaring returns.
But it’s simply a ruse designed to get you to part with even more of your hard-earned cash.
Deepfake videos may also be used in romance scams, to convince a victim the person they think they are messaging online is real and to eliminate any suspicions.
These AI scams aren’t just limited to pop stars and reality show names.
Top financial figures including Hargreaves Lansdown co-founder Peter Hargreaves and former Fidelity star fund manager Anthony Bolton have both had their likenesses recently pilfered by scammers who want to trick viewers into parting with their cash.
The fake AI-generated video showing a video which looked like and sounded like Bolton, for example, invited investors to join a stock-picking WhatsApp group.
While the video was not of Bolton, investors were lured in by the video’s likeness.
A This is Money investigation discovered that this fake video was behind a pump-and-dump scam where fraudsters used the WhatsApp group to share ‘expert’ advice on which stock to buy in order to inflate its price.
Then, once its value had soared following swathes of victims investing their money, it plunged. Shareholders around the world lost money in the blink of an eye.
But at the same time a group of anonymous scammers – likely members of organised criminal gangs – became very rich.
It’s not just well-known figures that can be copied in these deepfake videos.
A fraudster could even impersonate you to open up a new bank account in your name, or could imitate a loved one in financial or medical need to get you to part with your cash,
Would you fall victim?
It’s highly likely you have at some point seen a video that was altered or created using deepfake technology and believed it was real.
Finder presented eight video clips to thousands of adults to see if they could correctly guess which celebrity videos were real and which were deepfakes.
The results are shocking. Just 3.7 per cent of people who watched these videos correctly guessed which videos were real and which were fake.
Only 57 per cent of people were able to correctly identify more than half of the video clips, while just 18 per cent scored more than six out of eight.
The average score was 4.8 out of eight videos guessed correctly.
Let us know your score in the comment section.
Your browser does not support iframes.
Your browser does not support iframes.
Louise Bastock, personal finance expert at Finder, said: ‘Deepfake and AI fraud is here to stay. Some 6.4 per cent of the population falling prey to it is already a significant amount, and it will only continue to grow.
‘The public were generally able to guess some of the videos correctly in our quiz but virtually nobody got them all correct, and this is the danger.
‘You only need to be caught out once and it can have devastating consequences.’
Generation Z viewers – who are typically more exposed to deepfake videos on social media – scored highly as the got an average of 5.7 videos right while 74 per cent correctly guessed more than half of the videos.
Millennials also scored strongly as they correctly guessed some 4.9 videos of our eight. But those aged over 61 did poorly as they scored just 4.2.