Gender, race bias risk reported in biometrics used by US Labor Department

An alert issued by the inspector general (IG) of the U.S. Department of Labor is raising “urgent equity and security concerns” about the use of facial recognition in processing unemployment insurance (UI) claims, according to IT industry marketing firm MeriTalk.

The memo, issued March 31, emphasizes the need for “immediate attention” to address risks of racial and gender bias in facial recognition tools being provided by identity verification contractors.

“In 2019, the National Institute of Standards and Technology’s (NIST) Information Technology Laboratory reported that it found empirical evidence the algorithms used in current facial recognition technology have a racial and gender bias,” the memo reads. It also notes that more that 60 percent of the 24 states under review by the IG “did not include the privacy security measures recommended by the National Strategy for Trusted Identities in Cyberspace necessary to protect UI claimants’ biometric data.”

Contrary to the IG’s observation, NIST report author Patrick Grother told Biometric Update that the study supports the need to “know your algorithm,” rather than generalizations about the state of the industry.

“Without comprehensive guidance,” it says, “State Workforce Agencies (SWAs) are at risk of using technology that discriminates against claimants entitled to receive UI benefits and of not adequately safeguarding” personal information. The memo argues that SWAs hiring contractors that use facial recognition technology “are compelling claimants to submit biometric data to receive unemployment compensation that they are entitled to by state and federal law.”

The inspector general’s memo recommends giving people clear access to alternatives to biometric verification, requiring bias tests for facial recognition algorithms and hiring contractors who can deliver safe storage and disposal of sensitive personal data.

UI fraud spiked with COVID-19 in the United States. The IG report estimates that from April 2020 to October 2020, four states paid $9.9 billion on more than 1 million likely fraudulent claims, “with significant numbers of imposter claims being filed with stolen or synthetic identities.”

The battle against UI fraud has been on shaky ground since the beginning of the pandemic, with major service providers facing questions about security and accountability. Identification vendor ID.me has been accused of grossly inflating fraud losses suffered by COVID relief programs run by the federal and state governments, and is embroiled in lawsuits accusing it of biometric data privacy violations. Last April, the company fired half of its fraud review team for “inappropriate internal communications.”

The accumulated controversies, however, do not appear to be stunting the ID firm’s growth. This month, the company reported that 100,000,000 people in the United States have signed up for its digital wallet, on top of 14 federal departments and 35 agencies in 30 states.

Article Topics

biometric bias  |  digital government  |  face biometrics  |  identity verification  |  U.S. Government