Cyber security experts reveal the chilling number of images predators need to make deepfakes of children

Daily Mail​

31-05-2025

Cybersecurity experts have revealed that predators need just 20 images to create deepfake videos of children, prompting urgent warnings over the growing dangers of sharing family photos online.
Professor Carsten Maple, a leading expert from the University of Warwick and the Alan Turing Institute, said advanced AI tools can use a shockingly small number of pictures to generate realistic fake profiles and videos of minors.
The consequences, he warned, can include identity theft, blackmail, and online exploitation.
Parents are unknowingly giving criminals exactly what they need, with many doing it simply by uploading family pictures to social media and cloud storage platforms.
'It takes just 20 images for sophisticated AI tools to create a realistic profile of someone, or even a 30-second video,' said Professor Maple.
New research commissioned by privacy tech firm Proton found that UK parents share an average of 63 photos each month, most of them including children.
One in five parents post family pictures multiple times a week. Two in five do so several times a month.
The findings suggest today's children often have a digital footprint from birth, long before they understand the internet, or can give consent.
But it's not just criminals that experts are worried about.
Big tech firms are also harvesting these images for their own purposes.
Professor Maple pointed to Instagram's recent policy change, which allows the platform to use user photos to train its AI systems.
He called the move 'deeply concerning.'
He said: 'These companies use consumer data to build advertising profiles, analyse trends, train algorithms and track behaviour — often without people fully realising what's being collected.'
Over half of parents now automatically back up their family images to cloud storage. The average parent has around 185 photos of their child saved online at any given time.
Yet almost half admit they didn't know that tech companies can access and analyse those photos.
The study found four in ten parents believe tech firms only gather basic metadata, things like time, location, or device used, while 11 percent had no idea what kind of information is being collected at all.
Experts now warn that a generation of children could face serious long-term risks — including fraud, grooming, and deepfake abuse, simply because of the volume of images being shared.
'Oversharing can lead to digital records that are difficult or impossible to delete,' said Professor Maple.
'This opens the door not just to identity fraud, but also to more sinister forms of exploitation.'
Despite this, many parents remain unaware of how vulnerable their images really are. While 72 per cent say photo privacy is important to them, a staggering 94 per cent believe tech firms should be more transparent about how they use stored data.
Parental anxiety appears to be rising, with around 32 per cent of parents saying they are constantly worried about their phone or cloud accounts being hacked.
Nearly half say they worry about it from time to time.
More than half have already taken extra security steps, using Face ID, PIN codes, limiting app downloads, and keeping devices updated.
But Professor Maple says that's not enough.
With the rapid growth of AI and rising numbers of data breaches, the need to strengthen protection for children has never been more urgent.
'We are building digital profiles of children without their consent,' he said. 'The risks are real, and the damage, in many cases, irreversible.'