- Yoti’s ‘age estimation’ system was originally developed for the age of adults
- It compares a user’s facial features with images from millions of people
- For persons aged 6-18 years, the software has an error margin of only 1.5 years
- The technology is already being used by social networks and supermarket checkouts
- Social media firms like Facebook have long fought for age verification
An artificial intelligence (AI) tool that can tell if social media users are too young to use apps like Instagram and TikTok has been developed by a UK tech start-up.
Yoti’s ‘age estimation’ system – which may soon be implemented on social media – can tell how old users between 6-18 are by an error margin of 1.5 years.
The software compares the user’s facial features with millions of other images of Yoti Digital ID app users of the age known through the device camera.
Social media firms like Facebook have long struggled to handle minimum age verification without requesting passport details, which many see as intrusive.
scroll down for video
An artificial intelligence (AI) tool that can tell if social media users are too young to use apps like Instagram and TikTok has been developed by a UK tech start-up. Image: A child uses Yoti’s ‘Age Estimation’ tool to verify his age
Yotti’s ‘age estimation’ system – which may soon be implemented on social media – can tell how old users between 6-18 are by an error of 1.5 years (right). The software works by comparing the user’s facial features with millions of other images of Yoti Digital ID app users of known ages via the device camera (left).
Who is using AI?
According to Yoti, several organizations are already using their age estimation system, including:
- Ubo, the French social network
- Smash!, the healthy eating network
- Lebara Mobile
- Estonia Retail (via Strongpoint)
- Fan Centro, an adult platform
- game payment technology
The technology can be used not only with camera-enabled smartphones, tablets and laptops – but also with checkout terminals in stores.
“The threats that children face online continue to grow, so I’m proud to estimate Yotti’s age for under 13,” said Yotti CEO Robin Tombs.
‘This technology will help businesses and regulators better protect youth with less friction while maintaining confidentiality.
‘We’ve now made it easier to design services “age appropriately” for platforms around the world.
‘The age estimation of the yoti’s face will help many different businesses comply with changing age rules such as new child codes.’
Businesses using the software – which previously only worked on adults – can set an age limit for the AI to compare to each user.
As such, the system can be used to see if children meet the legal limit of 13 to join apps like Facebook and Twitter – or to determine whether a person has reached the drinking age. has crossed.
In fact, the system is already being employed in supermarkets in Estonia for age verification at automatic checkout and by the German version of adult entertainment platform Fan Centro – and has already conducted more than 550 million age checks.
To train the AI on adult faces, Yoti engineers used images from millions of users who had downloaded the firm’s Digital ID app.
Launched in 2014, it provides a stand-alone method of proving age and identity based on a combination of ID documents and facial recognition.
To train the AI in estimating the age of young people, photographs of children were used – with parental consent – as part of a program organized by the Information Commissioner’s Office, a UK data watchdog.
According to London-based Yoti, they have managed to improve the system’s accuracy in estimating young people’s ages over the past three years.
In 2018, the AI was accurate to within a 1.5-year difference for those aged 13-24 and to within a year for those aged 16-17 – but the firm is now at 1.3 for users aged 6-12. Reports accuracy within years and up to 1.5 years for those aged 13-18.
For users on the verge of a legal limit for accessing social media sites – whose age may be incorrectly estimated – the Platform may request further forms of verification before granting access.
In addition, the firm told MailOnline, the AI is built with ‘proprietary anti-spoofing technology and passive vibrancy detection’ to ensure that a real, lively face is being presented for analysis.
As such, the company explained, the system should not be able to be submitted by younger users to a picture or video of an older person.
To train the AI on adult faces, Yoti engineers used images from millions of users who had downloaded the firm’s Digital ID app. To train the AI in estimating the age of young people, photographs of children were used – with parental consent – as part of a program organized by the Information Commissioner’s Office, a UK data watchdog. Image: An artist’s impression of the facial recognition and analysis program in operation
Yoti’s Robin Tombs said, “The dangers that children face online are sadly increasing, so I am proud to present Yoti’s Age Estimation for Under 13s.” ‘This technology will help businesses and regulators better protect youth with less friction while maintaining confidentiality.’
Research by the Children’s Commissioner for England found that 60 percent of eight-year-olds and 90 percent of 12-year-olds regularly use private messaging apps, compared to those 13 and older. are prohibited.
Similarly, a 2018 investigation by market intelligence firm Kids Insights found that nine-in-ten are illegal in 12 years on social media sites like Facebook, Instagram and Twitter.
To address this, lawmakers have proposed legislation in the draft Online Safety Bill, which would mandate social media companies that…