July 9, 2019
Facial recognition technologies have increasingly been at the heart of fierce debate among lawmakers, privacy advocates and technology innovators. While facial recognition technologies have the potential to make identity and authentication not only more efficient but also more secure, there are some who would like to see them banned.
In large part, facial recognition isn’t widely accepted as a viable form of biometrics because there is so much confusion that exists around the technology. “There is a heightened level of narratives and stories about facial recognition and other technologies that do not convey a full understanding of how the technology actually works,” said Jake Parker, director of government relations at Security Industry Association (SIA). emphasizing the increasing need to educate both the public and stakeholders.
In a recent report, Face Facts: Dispelling Common Myths Associated with Facial Recognition Technology, SIA wrote, “Advanced facial recognition technology has benefited Americans in countless but underpublicized ways—helping to find missing children, fight human trafficking, secure the border, find dangerous criminals, bring sexual predators to justice and thwart identity thieves.”
Lumping Uses Together in One Bad Bucket
Despite its ability to mitigate threats and prevent crimes, facial recognition feels creepy to the wider public. As a result, Parker said there is a lot of confusion about the different uses of facial recognition as a tool for access control in traditional security and the way that it is used in law enforcement. “It’s being lumped together and confused,” he said.
Guided by that fear, lawmakers in San Francisco recently voted to ban the use of facial recognition technologies among city departments. It’s a movement that is spreading not only throughout cities in California but across other states as well. These efforts, said Parker, are driven by fear and concern about privacy issues because of the lack of transparency around who has access to the data.
The use of facial recognition by the U.S. Customs and Border Protection (CBP) is another type of use, but it’s different from using facial recognition for access control. What’s important to remember if you are a traveler is that you are already in a situation where you are required to provide identification. “Facial recognition makes that more secure and more convenient, so there’s not much of a privacy implication,” Parker said.
Several years ago, activists tried to get a ban on facial recognition at the federal level, but it failed. The campaign was followed by efforts at the state level, which also failed. “Now they are going after the local level,” Parker said.
For many years, there has been a well-coordinated and ideologically-driven campaign targeted at CBP, according to Parker. The technology has been criticized at every step throughout the policy development process. “It’s no surprise they are ramping up efforts now,” said Parker.
“More accountability needs to be in place, but we don’t think there needs to be a ban before moving forward. If we start banning technology because somebody says its creepy or dangerous, that’s not good for communities or for safety.”
Perception vs. Reality
Parker said that it’s widely understood among the security industry that the irresponsible use of facial recognition needs to be avoided. Still, the American Civil Liberties Union of Massachusetts is fighting back against what it calls the secret use of government facial surveillance. There are many similar claims attempting to sway the public’s perceptions of facial recognition technologies.
Though there are many claims of misuse or abuse, Parker said they are mostly theoretical. “I don’t know of any examples in the U.S. where facial recognition has been misused. You can’t just say something is dangerous without showing that there is a harm.”
Any technology tool can be used for ill, which is why it is important to be able to trust that the government is using the technology the way they say they are. Parker said, “There is a way to do audit-abile searches to ensure that people are using it as they should be. Build audit-ability into the system, and explain to the community that we are using this for making sure there is not fraud in the driver’s license system.”
As of the writing of this story, though, several publications including the Boston Globe reported that the Federal Bureau of Investigation (FBI) and Immigration and Customs Enforcement (ICE) have been exploiting state DMV records for facial recognition data without the knowledge or permission of drivers.
“Georgetown Law researchers, together with the Washington Post, have obtained facial recognition requests, documents, and emails which have revealed a project that uses vehicle ownership and driver license databases for surveillance purposes,” wrote ZDNet on July 8.
Adam Levin, founder of CyberScout and author of “Swiped” said “This is a clear violation of consumer privacy when government agencies are allowed to access personal data from driver’s licenses without legal consent.
“Combining this with facial recognition technology only makes this more dangerous since the technology itself is not always accurate, has been criticized for racial bias and is another tool that adds to our surveillance economy.”
In attempting to elucidate some of the many myths around facial recognition, the SIA report stated, “Criminals use aliases and fraudulent identities every day, harming public safety by slowing time-critical investigations and wasting taxpayer resources. Additionally, searching for a common name (e.g., John Smith) could yield hundreds of results that must be narrowed down using traditional methods. Facial recognition technology simply automates and improves the first step in these processes to identify potential matches.”
Different Types of Bias
Bias is one of those words that evokes upset, particularly when used in common speech. But biases in technology are different from racial biases, Parker said. The performance of facial recognition algorithms has been tested by NIST for many years. “Bias in this type of technical research means something different from how we would use it in everyday speech. There are all different types of bias—selection bias and other types of bias that describe research results.”
If a person is identified as a potential match, that person has to then be verified by other more traditional means such as passports and other types of identification. “It doesn’t mean that person is singled out for abuse or harm,” Parker said.
What’s important to understand in these use cases is that the investigative use of facial recognition in law enforcement does not identify someone. “All it does is aid the investigators in their process of identifying someone. It is not saying, this is the person. There is no identification made, so it’s not possible that there could be a mis-identification,” Parker said.
A Path Forward
Yes, some municipalities and states use facial recognition technologies in the DMV to thwart fraud. In fact, the most common use for these technologies is to combat the rise of synthetic fraud. Butting heads and turning to the court are not the best path forward for any stakeholders when it comes to facial recognition or any other technologies. “There is room for additional safeguards and transparency measures. Part of the issue is that in some cases, communities have deployed the technology without a lot of education of stakeholders and the community,” Parker said.
It’s important that there are reasonable safeguards that can be found from discussions. There is a lot of negative information out there that is being driven by misconceptions, which is why Parker said that especially in local jurisdictions that are looking to ban the technology “there needs to be a more thoughtful debate about how to make sure this works for everybody.”