Facial recognition technology is now facing its first official backlash, with San Francisco’s municipal ordinance banning its use by city and county agencies. It’s the first such ban in the US and arose in one of the nation’s most tech friendly cities. And San Francisco is not alone in aiming to put the brakes on facial recognition.
In the neighboring Bay Area, both Berkeley and Oakland are considering bans, as is Somerville, Massachusetts. California’s state senate is considering a ban on facial recognition in police body cameras. Legislators in Washington state have proposed a broad ban on facial recognition, and on the federal level, two senators have proposed a measure to stop companies from using facial recognition data from customers without their consent.
Proponents of facial recognition cite public safety concerns, and a potentially enhanced ability to deter crime and terrorism. While many acknowledge that there are accuracy and bias issues with the technology in its current form, they argue that these issues will be solved by improvements in machine learning—and that they once this happens, facial recognition will be an important tool for government and law enforcement.
Studies have found facial recognition to be far less accurate in identifying non-white individuals and women. Researchers have also found that without proper oversight, police departments manipulate photos to generate matches. Even the FBI’s facial recognition system has shown an unacceptable error rate of nearly 15 percent. In Britain, police have repeatedly failed to test how the technology’s accuracy is affected by the ethnicity of its subjects. Due to these problems, many of those who generally approve of the technology are still calling for increased oversight and regulations, if not a complete moratorium.
“I’m not sure if a total moratorium on this is going to be an answer, because we still have a homeland we have to protect, and because there is still some value in facial recognition,” former president of the National Organization of Black Law Enforcement Executives, Cedric Alexander, said in testimony to the House of Representatives last week.
And indeed, bias and inaccuracy is a great reason to limit its use, and it’s a good sign that these factors have led to fairly broad calls to slow implementation of the technology. But there are also larger issues at hand, and it’s not enough to simply postpone the problem. There are serious questions of civil liberties raised by the use of even accurate and unbiased facial recognition.
In China, facial recognition has become widespread, and is used for a wide range of applications, from monitoring ethnic minorities to publicly shaming citizens who jaywalk. By tracking its 1.4 billion citizens, the Chinese government could eventually aim to predict who will commit crimes ahead of time, according to Fortune.
“There’s more willingness in China to adopt it,” according to Oxford University AI researcher Jeffrey Ding.
And Maya Wang, senior China researcher for Human Rights Watch warns:
“The intention of these systems is to weave a tighter net of social control that makes it harder for people to plan action or push the government to reform.”
The government has already used the technology to track members of the western Uighur ethnic minority, monitoring and reviewing their behavior, according to a report from The New York Times.
Facial recognition is also one component of a wider plan to implement a “social credit score” system, which will track the behavior of Chinese citizens. If their score falls below a certain point, citizens could lose rights and be blacklisted from services.
These clear human rights abuses rely on the accuracy of facial recognition, so simply improving the technology would obviously not prevent similar civil rights violations. And US history suggests such surveillance could not be unique to China.
Similar issues motivated the campaign for the new San Francisco ban, but have been largely forgotten in the wider conversation. The US has its own history of unjustified surveillance without reasonable suspicion of criminal activity. Such politically-motivated surveillance was used by the FBI to monitor and undermine the civil rights movement in the 50s, 60s, and 70s. San Francisco was a focus for this surveillance activity, and that history helped drive the push for the new ban.
Despite measures enacted to prevent such abuses after revelations of that surveillance shocked the public, we’ve seen similar violations post-9/11. In times of political and social crises, the constitutionality of such surveillance tends to become an afterthought.
“You could be at a rally supporting gun rights or protesting gun violence. In all of these cases, the government can monitor you without your knowledge and enter your face in a database that can be used in unrestricted ways,” said Elijah Cummings, chairman of the House Oversight Committee, during a hearing on the issue last week.
We need to weigh the potential benefits against the costs to civil liberties in boosting surveillance capabilities despite such a terrible track record of misuse. Even once the march of technology solves the problems with accuracy, these questions will still be there. San Francisco is right to ban facial recognition entirely, maintaining the right of US citizens to protection “against unreasonable searches and seizures,” enshrined in the constitution.