Problems with facial recognition software used to verify that unemployment beneficiaries are costing people dearly

The facial recognition program used by 22 US states to reduce unemployment fraud failed to correctly identify beneficiaries, which resulted in the denial of benefits or the suspension of their claims by dozens of people, Motherboard reports.

The unemployed shared complaints online During months on identity verification software, but outrage boiled this week in response to a Axios Newsletter on the threat of unemployment fraud based on statistics provided by

The service uses biometric data and official documents to verify users. However, several unemployed applicants told Motherboard that’s facial recognition models did not identify them correctly and that they had difficulty contacting support staff to resolve the issue. . In the meantime, their applications have been put on hold, forcing beneficiaries to wait days or weeks until they can get in touch with a “trusted referrer” to confirm their identity. Others who tried multiple times to pass’s facial recognition test were left out of the system. CEO Blake Hall said in a statement to Motherboard and other media that the algorithms behind its Face Match technology are “99.9% efficient.” In general, modern facial recognition technology has repeatedly proven to be less accurate for people of color. But Hall argued that does not suffer from this problem: a regression analysis performed by the company found “no relationship between skin tone and Face Match failure on a basis of 1. : 1, ”he said.

As to the cause of these reported issues, Hall suggested the user error could be to blame. “For example, if someone uploads a selfie that only shows half of their face,” he told Motherboard. This could apparently prevent facial recognition software from identifying them correctly.

In his statement, Hall added that the company was not aware of “eligible persons” who had not been able to verify their identity with his department. People who fail’s facial recognition test are not blocked, he continued, and can still prove their identity via video chat, which he claimed the wait time is currently “less than five minutes and has always been less than 30 minutes all week”.

However, many people have taken to twitter Say the opposite. Dozens of people complained that their emails, phone calls and chat requests went unanswered for weeks or even months. Earlier this year, benefit recipients in Colorado who had no problem confirming their identity before the state labor department started using, said they were suddenly rejected by the new system and went months without receiving payments. Similar stories made headlines Florida, North Carolina, Pennsylvania, and elsewhere.

The unemployment rate has skyrocketed amid the coronavirus pandemic, overwhelming many state offices and pushing their already dilapidated computer systems beyond breaking up point Last year. Therefore, some states reported a sharp increase in fraudulent claims last spring, and the Ministry of Labour said in February it identified $ 5.4 billion in potentially fraudulent unemployment benefits paid between March and October of last year. This is still far from Hall’s estimate, which complaints that up to half of unemployment benefits in the United States during the pandemic, amounting to some $ 400 billion, were fraudulent claims.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *