AI for Identity Verification

September 29th, 2017 by

Onfido, a UK-based company, has developed a system to verify people’s identities using a photo-based document, selfie, and artificial intelligence algorithms. The company initially focused on background checks, but they have been moving more towards identity checks for businesses in financial services and e-commerce. Companies and platforms in these industries have a strong requirement to verify the identities of people with whom they do business. In addition to the identity challenge is the fact that onerous checks when onboarding users often leads to a decrease in usage. To address this, Onfido uses a system that is based on standard identity documents. Their solution currently recognizes about 600 formats including drivers’ licenses from dozens of countries, passports, identity cards, and selfies. For the selfie option, the user must respond to instructions for two or three different actions such as blink, smile, or speaking a sequence of numbers. The system uses the actions to determine if the selfie is a static photo or real time. It also compares the face in the selfie to the picture on the ID card. The result is a risk report for the user. It notes if the user comes up as compatible, mostly so, or as a flag. All the data is then passed on to the company requesting the check.