A number of recent problems showed the reliability problem of facial recognition hurt people at the time needed. The motherboard reports that there are complaints that are ongoing about ID. Face recognition system at least 21 countries are used to verify people seeking unemployment benefits. People have gone weeks or months without benefit when the face match system does not verify their identity, and sometimes it is not lucky to get help through the video chat system intended to solve this problem.
- Me Chief Blake Hall blames the problem with the user than the technology. The face match algorithm has “99.9% efficacy,” he said, and there was “no relationship” between skin tones and recognition failure. Hall instead suggested that people not share selfies correctly or not follow instructions.
The motherboard notes that at least some people have three attempts to continue the facial recognition examination. The outlet also shows that the company’s claim about the cost of national unemployment fraud has grown rapidly in the past few months, of the $ 100 billion reported to $ 400 billion. While Hall connects it to expand “data point,” he did not say only how the company counted damage. It is not clear what the threat of real fraud is actually, in other words.
Whatever happens with Id.me technology, the incident highlights one reason the federal and state government hopes to limit facial recognition. Even if privacy and security are not a problem, they don’t seem to be quite reliable to avoid significant problems. The success rate of 99.9 percent can still leave many people without the benefits of qualifying to claim. This system may need to be far more reliable to eliminate this headache in the future.
Update 6/20 5pm et: ID.me is not surprising to take problems with the motherboard pieces in the response given to Engadget. Hall is called fraudulent claims “inaccurate and defamatory,” and emphasizes that the company sucks to the National Institute of Standards and technology standards for identification. He reaffirmed the claims made about the nature of the face recognition system and backup video chat, and claimed that up to 2.5 percent of fraud efforts included masks, photos and videos.