The unintended consequences of banning face recognition

Face recognition is making the news these days — and not in a good way.

Many articles cite concerns about face recognition misidentifying people of color, the risks of government and industry surveillance, and invasion of privacy. Several cities have banned city agencies’ use of facial surveillance, including the police. And a bill is before Congress to prohibit federal funding from purchasing or using face recognition.

Having worked with face recognition technology for much of my career, I often read fear and confusion rather than facts within many of these articles.

The fear and confusion are understandable — face recognition is a remarkably complex technology. And the idea of government or industry using an image of your face without permission feels like a violation.

While humans recognize known faces instantly, teaching a computer system to do it is very difficult. But with advances in computers, cameras, and algorithms over the past ten years, the technology has accelerated rapidly. And like many of the important technologies of the 21st century (GPS, internet, etc.) we’re only now learning how many ways they can be used — and the implications.

To start to understand how face recognition works, you must understand two important principles:

• One, the true power in face recognition is that it provides attribution. It allows us to link a person to a place, time and actions that are happening at that place and time. That makes it valuable for an extremely wide range of applications.
• Two, it’s not the technology itself, but the data it generates about us and how that data is used that is either helpful or dangerous.

The sad paradox about the newer technologies is that while many people may use them, relatively few genuinely understand how they work. And thus, with face recognition, instead of looking to solve problems — some of which aren’t real — many are leaping to “banning it” as the only solution.

Before we take that step, it’s important to understand some of the potential unintended consequences. If we ban this technology:

• America will lose one of its most powerful tools against domestic and international terrorism. Not using face recognition and other biometrics technologies is a leap backwards in tracking and identifying the bad guys.
• The United States would lose our leadership position in this critical technology. U.S. industry is committed to face recognition technology even if our government bans it for many purposes. Not maintaining our lead will only ensure that U.S. industry will receive the technology from potentially untrustworthy sources. 
• We will lose a major tool essential to many critical applications, including fighting child pornography, finding missing persons, and identifying natural disaster victims. Banning face recognition bans all applications of the technology — including those that aren’t controversial.   
• Our ability to maintain privacy and security will be weakened, rather than strengthened. Many of us already use face recognition to keep our smartphones private. We can strengthen our privacy in many other applications by using these technologies to control access to our data. If we want both privacy and security, the proper use of technology with good governance is our best route to it.

These are just a few consequences of banning face recognition.

While it’s beyond this brief piece to go into more detail, I’ll say most of the issues around face recognition aren’t really about the technology itself, which is constantly improving in accuracy. Rather, it’s about how it’s used, and who controls the data it develops.

In short, “who owns your face?”

Right now, in most cases, copyright belongs to the person or institution who created the image, not the subject; thus, ownership of our data is where we should be focusing our concerns about privacy, because face recognition is just a piece of it. Similar issues are emerging around other technologies, including GPS use, always-listening voice assistants, public and private data aggregation, and artificial intelligence.

America’s privacy and civil liberties controls are many decades old. We need to analyze and update our laws and policies to accommodate this new class of capabilities so that the government and private sectors can use them to make our world safer, while preserving the fundamental rights of individuals and society.

Meanwhile, instead of banning face recognition, I recommend bringing together a body of recognized technology, privacy, and civil liberties experts. This group can help legislators, policymakers, and the press understand face recognition technology — and develop governance based on a sound foundation. 

As for which organization convenes this group, I suggest considering the American Association for the Advancement of Science; Institute of Electrical and Electronics Engineers; National Academies of Sciences, Engineering, and Medicine; National Science Foundation; or National Society of Professional Engineers.

But whichever organization leads the effort, it’s imperative we start now. Because truly understanding the facts is essential to seeing past fear — and avoiding unintended consequences.

Craig Arndt is the Technical Director for Homeland Security Programs at MITRE, which operates the Homeland Security Systems Engineering and Development Institute™, a federally funded research and development center. He’s a recognized expert in biometric systems, human computer interface and human-centered systems, image and signal processing, artificial intelligence, and intellectual property management.