Facial verification will not fight fraud

With the U.S. The economy has just begun to recover from Covid-19, and millions of people are still unemployed. Congress authorized the expansion of unemployment benefits to supplement the state aid program.Although it is commendable to strengthen struggling Americans in the ongoing crisis, bad actors have Unemployment fraud A kind serious problemUnfortunately, many states seeking to prevent fraud through surveillance are installing Biased system This may do more harm than good.It is foreseeable that these systems are making mistakes. When they make mistakes, they will punish BIOCs, transgenders and non-gender Americans to a large extent.

Twenty-one states have turned High-tech biometric authentication Use computer vision to determine whether people are the services of who they claim to be. This is the same technology that allows users to unlock their phones with their faces-it’s a one-to-one matching process where the software will infer whether your facial features match the facial features stored in a single template. However, while facial verification is common in consumer devices, it is relatively rare for government services. It should remain in this state.

You may believe in facials confirm Is harmless, because the controversy that will not go away mainly revolves around the face recognizeWhen the police compare the image of a suspect with a database containing photos of a photo or driver’s license, they use facial recognition technology, in which an algorithm tries to find a match.Relying on facial recognition technology leads to the police Wrongly arrested At least three blacks, and there may be more.

But face verification can also be biased.When the error appeared, they were already California, They have historically been disproportionately focused on Gender and ethnic demographicsFor welfare fraud programs, the government’s reliance on facial verification increases the risk of slow or even rejection of applications for people of color, transgender, and non-binary applicants. These results can make it difficult for you to keep the lights on or keep the roof on. To make matters worse, law enforcement agencies may over-examine vulnerable groups because biased algorithms will doubt who they are. For those who simply fail the flawed algorithm test, this interaction can lead to false arrests, prosecutions, and government liens.

Sadly, government agencies are creating conditions for algorithmic injustice to persist for a long time.When Michigan launched its integrated data automation system in 2013, the plan’s features were success Five times as many unemployment fraud cases were flagged, and new fines and fees of $65 million were brought in. It turns out that the software is unreliable.The system incorrectly marked tens of thousands of Michigan people, and humans put rubber stamps on automatic judgments, resulting in Bankruptcy with worse.

Increase reliance on smartphone applications, such as identity number It also increases the risk of the digital divide.a lot of Low-income and elderly Americans There is a risk of being excluded from basic government services simply because they don’t have phones with cameras and web browsers.

As with any extension of biometric analysis, there is a second powerful threat.We are increasingly using facial verification standardization It is expected that our bodies should be used as a form of government identity. Whenever the government accepts facial verification and recognition, it creates momentum for further spread.

The good news is that biometric technology is not required to prevent fraud. Recognizing that there are alternatives requires acknowledging that the lack of a secure digital identity system for Americans is a major problem. Patching the response means that some systems will use some or all of your social security number-a result that makes the social security number an important target for hackers. Other systems use credit card transactions and credit history.These methods are error-prone and insufficiently inclusive, especially because 7.1 million American households Keep no bank account.

What is needed is a secure ID, such as a digital driver’s license or state ID with a secure encryption key. In this way, users can provide identity verification information without registering in an automated system that is subject to prejudice and subtle monitoring. Although this is not a perfect solution, we should not look for one. Any system that can finally prove your identity at any time, such as a universal ID that everyone must have, is a large-scale surveillance tool. By adopting a gradual and privacy-protecting digital ID strategy, we can reduce the risk of welfare fraud and other forms of identity theft, while protecting privacy, fairness, and civil rights.

Wired opinions Publish articles by external contributors that represent a wide range of views.Read more opinions HereAnd check our submission guidelines Here. Submit column Comments@wired.com.

More exciting connection stories