A recent series of problems suggests that the problems of reliability of facial recognition are hurting people in a moment of need. The motherboard indicates that there are ongoing complaints on the ID.ME facial recognition system at least 21 States use to verify people looking for unemployment benefits. People went weeks or months without benefits when the facial match system does not check their identities and sometimes had no chance of getting help through a video chat system to solve these problems.
Id.me Chef Blake Hall blamed the problems of users rather than technology. The correspondence algorithms face have an “efficiency of 99.9%”, he said, and there was “no relationship” between the skin and the failures of recognition. Instead, hall suggested that people did not share the selfies properly or were not as a result of the following instructions.
The motherboard noted that at least some people have three attempts to achieve the verification of facial recognition. The outlet also pointed out that the demands of the National Unemployment Fraud Cost Corporation increased rapidly in recent months in recent months, $ 100 billion recorded at $ 400 billion. While Hall has awarded that to expand “data points”, he did not just say how his company has calculated the damage. It is not clear that the threat of real fraud is, in other words.
All that is happening with ID.ME technology, incidents highlight one of the reasons why federal governments and states hope to limit facial recognition. Although privacy and security are not problems, they do not seem to be reliable enough to avoid significant problems. This success rate of 99.9% could always leave many people without benefits, they are eligible to claim. Systems such as this may need to be considerably more trustworthy to eliminate these headaches in the future.
Update 6/20 17h and: Id.me is unsurprisingly with the part of the motherboard in a response provided to Engadget. Hall called “inaccurate and defamatory” fraud and emphasized that the company suck the National Institute of Technology Standards and Standards for Identification. He reiterated claims on the nature of the face recognition system and backup of video chat and stated that up to 2.5% of attempts were fraud that included masks, photos and videos.