The Center for Data Innovation spoke with Husayn Kassai, chief executive officer and co-founder of Onfido, a company based in London that provides AI-powered ID verification. Kassai discussed how enabling identity verification with a combination of facial recognition and document reading technology can support the financial industry in securing services.
Eline Chivot: What has led you to create Onfido? What did you think was wrong with identity verification today?
Husayn Kassai: As a migrant from Iran to the UK at a young age, I saw the hardship my parents endured in gaining access to basic things like a bank account, because of the limitations associated with identity and the credit bureau system. While running the Entrepreneurs Society at Oxford University, my colleague Eamon Jubbawy and I met Ruhul Amin, with whom we discussed the inefficiency of identity verification. Together we teamed up to tackle this issue and eventually formed Onfido.
In a digital age, it’s increasingly difficult for businesses to prove their users really are who they claim to be and protect themselves from identity fraudsters. Now, with facial biometrics technology, and a hybrid machine learning, or human technology, approach, we are tackling the problem head-on.
In a world increasingly characterized by remote, online transactions, and strangers sharing commodities, human interactions without trust will only take us so far. In the past, we’ve built trust by allowing relationships to develop over time, but in a remote and instant economy, the assurance that a person is who they claim to be needs to be quickly obtained and accurate.
Chivot: Which system and technologies do you use for identity verification?
Kassai: Onfido uses many technologies that can deliver trust in digital identities at speed and scale. Identity verification solutions are becoming a necessity for businesses across the board, and new technologies are emerging to develop this even further. From machine learning solutions that can automatically recognize and classify fraudulent documents to biometric measures that provide live, ongoing identity verification, it’s an area that’s evolving rapidly. These technologies will be pivotal in ensuring trust in an economy where there is little scope for error. Plus, the more data the technology receives, the better it performs because machine learning allows the platform to continue to learn with each new data set.
Chivot: Can you explain why and how fintech firms are using Onfido’s identification technology, and the concrete benefits?
Kassai: Since the financial crash of 2008, financial services have faced dramatically increased regulation. The burden of compliance, as well as the teams dealing with it, has increased. Additionally, as traditional financial services continue migrating online and the rise of digital-first offerings accelerates, most customers interact with banks online. The legacy processes for addressing regulation and verifying identity are no longer fit for purpose.
In light of this increased demand, identity and biometric verification technology is stepping in to help verify vast numbers of users in a way that is effective, rapid, and remote. Still, regtech is seeing slow adoption from more risk-averse, traditional institutions—and as fintechs continue to close the competitive gap, traditional financial services will need to rethink their approach to stay ahead. Online identity and biometric verification will eventually become the norm—tried, tested, and trusted. By being open to these new solutions, financial services across the spectrum will be able to stay competitive.
In the next few years, we expect to see biometric solutions replace knowledge-based authentication (KBA) altogether. We also expect this to extend to other applications including e-voting and immigration systems reforms.
Chivot: It is easy to forge government documents. How efficient is your technology to address this challenge? Why is it better than human verification?
Kassai: Biometrics are increasingly providing the answers to some of identity verification’s trickiest questions. For a fraudster, forging a document or buying a social security number on the dark web is easy—but faking your face isn’t. Our technology prioritizes accuracy over speed. We take the best of both to ensure the verification process is both quick and efficient. AI allows Onfido’s verification experts (known as super recognisers) to only spend time tackling high-risk cases that require extra review and human judgment, while other users get through the verification process quickly without being blocked or delayed from executing critical transactions. As AI and machine learning technologies continue to evolve, forged documents will be spotted even quicker and without the need for additional human input.
Chivot: What distinguishes your technology and your approach from other companies which are also developing products for identity verification?
Kassai: The human role is a critical component to Onfido’s verification process and sets Onfido apart from competitors. Our hybrid approach, which incorporates both AI and human verification, lowers Onfido’s fraud exposure rate to 0.0195 percent (calculated by taking the industry average fraud rate of 1.5 percent, and multiplying it by the false acceptance rate), creating a smarter system for both users and verifiers.