In the United States, ID.me’s facial-recognition system, used to process unemployment benefits, has been accused of inconsistent operation, leading to denials or suspensions of payments. As reported by Motherboard.
According to media reports, unemployment beneficiaries from many states repeatedly complained that the system failed to identify them correctly. They found it difficult to reach the company’s representatives to resolve the issue.
ID.me chief executive Blake Hall said that biometric-identification failures are an issue for people who misuse the service’s prompts during the verification process, not for the technology itself.
He said the facial-matching algorithms operate with an effectiveness of roughly 99.9%.
“There is essentially no link between skin tone and refusals by the one-to-one facial-matching verification,” Hall stressed, citing the company’s regression analysis.
He added that they were not aware of people who could not verify their identity using its software.
“The real-time video-chat session wait time is now under five minutes, and over the course of the week it did not exceed 30 minutes,” Hall noted.
ID.me’s identity-verification service cross-checks biometric data against official documents to reduce unemployment-benefits fraud. It is used by authorities in 21 states.
In April 2021, a British activist accused the government AI of racism due to facial-detection errors.
In early April, the facial-recognition algorithms of the Proctorio program, which monitors exam integrity, were accused of bias against Black users.
In early 2019, an African American spent ten days in jail due to a facial-recognition error.
Subscribe to ForkLog News on Telegram: ForkLog AI — all the news from the world of AI!