Technology for anonymous age estimation

The new technology is being used to check the ages of social media users while maintaining their privacy

Spotted: Social media is a great tool to connect, but without proper supervision it can be a dangerous place for children. This is why, in the US, UK, and much of the EU, social media users must be 13 or over or have parental permission before they can sign up. In addition, online services are not allowed to collect or store the personal information of anyone under the age of 13 (in some European countries, this age limit rises to 16). Because of this legal jeopardy, social media companies are more eager than ever to find a way to verify users’ age. Startup Yoti now has a solution.

Rather than verify using ID documents, Yoti uses facial age estimation technology, which estimates a person’s age by looking at their face. Crucially, this approach allows the company to perform secure age checks from images without recognising or identifying the individual. The company’s platform is powered by a neural network, which has been trained on millions of images from Yoti users who have opted into this use of their data. The company claims that its system is accurate for 6 to 12 year olds with a mean absolute error of 1.36 years and of 1.52 years for 13 to 19 year olds.

Yoti has used ‘privacy by design’ principles in creating its system, ensuring that no individual can be identified. For example, all images are deleted instantly after age is estimated. The company also points out that, by eliminating the need for ID documents, its platform actually encourages data minimisation and privacy. Yoti points out that their technology “is a scalable solution which is quicker and more accurate than manual ID checks”.

Springwise has also spotted neural networks being used in other innovations such as background noise cancellation and prediction of psychological age.

Written By: Lisa Magloff