During the pandemic, Black applicants for unemployment assistance in Wisconsin and North Dakota received benefits at approximately half the rate of their white counterparts. That's according to a June 2022 report by the Government Accountability Office (GAO), which looked at the impact of pandemic relief across four different states.

And this isn't an isolated case. It is an indication that, across the board, digital identity verification that relies on legacy data, systems, and documents don't treat people fairly, equitably, and with respect. Instead, they lock in the problems of the past. And unless we rethink systems from top to bottom and use advanced data science as the foundation for verifying identity, we risk making the same mistakes over and over again.

Today, digital identity verification is a fundamental part of the way government serves people. Government has to be able to distinguish between bad actors and those who are eligible for benefits. If we don't get this right, fraud will rise and people who are owed benefits won't get them. That's a true lose-lose.

Identity Verification Shortcomings

This dynamic became crystal clear in the pandemic. We know that individual scam artists and international fraud rings pocketed billions in relief money using synthetic fraud techniques, combining fake names with real Social Security numbers, and even donning wigs to successfully fool legacy systems.

And who suffers most? People who live in communities of color, new-to-country individuals, and others who have been historically underrepresented. Let's be clear — government identity infrastructure as we know it today is fragmented and broken. The pandemic shined a light on the extent to which legacy systems marginalize people who might have thin credit files, are underbanked, have limited broadband access, are rightfully hesitant about facial recognition, or lack government identification. When trying to verify their identities online, these people are met with roadblocks, friction, and long delays — a longstanding issue that has only recently gained mainstream attention.

Identity verification is the gatekeeper to accessing and engaging in an online world, whether you're a homeowner with a lengthy credit history, an elderly individual applying for Medicare, a young entrepreneur looking to start their dream business, or an individual applying for unemployment benefits after getting laid off unexpectedly.

Of course, economic status, race, geography, and access to broadband have long dictated — and still do — the ease by which an individual can prove who they are or otherwise engage in modern society. According to the Consumer Financial Protection Bureau, approximately 45 million Americans, mostly Blacks and Hispanics, lack the traceable credit histories often used to prove identity. For the approximately 57 million people living in rural areas, having to take hours off of work and drive to an in-person facility to verify their identities is not an option. And an estimated 21 million Americans still lack broadband access, making it harder to complete applications that rely on video chat verification.

3 Steps to Improve Digital Identity Verification

These problems can't be solved overnight. But there are steps we can take in the near term to move toward accurate digital identity verification for all.

First, Congress should take steps — such as the Improving Digital Identity Act, now under consideration in both the House and Senate — to digitize common identity verification documents, including driver's licenses, birth certificates, and Social Security cards. We should move to ensure that government-issued data (numbers for Social Security, taxpayer identification, passport numbers, etc.) are made available to ease the burden on the public and service providers when verifying an individual's identity. Doing so will move our system away from a reliance on physical documents that are often lost, misplaced, or stolen and will offer more avenues to verify identity online.

Second, we need to design a system that combines machine learning, artificial intelligence, and data analytics to enable more efficient, more equitable verification. Historically underrepresented people often lack the government-issued IDs that current systems require. By combing through all information available — from applicant name to the device being used and more — we can verify good actors faster, flag fraudulent identities across systems, increase auto-approvals, and minimize fraud.

Third, we must protect public trust through transparency. The erosion of public trust hamstrings the use of innovations that can better combat ever-evolving fraud. An overreliance on intrusive (and error-prone) technologies like selfies can weaken this trust and cause significant backlash and public outcry over valid concerns like privacy and bias. Vendors partnering with the public sector must be transparent with their practices.

Without a real focus on equity, the fragmentation of identity across the US and the widening disparities across different populations will only grow. To move forward, we must think comprehensively in our approach when it comes to digital identity verification so that no one gets left behind.