China has 4.1 people for every camera, while in America the figure is 4.6. In China, many of these cameras are controlled by the government, while in America the private sector is more active in monitoring security in retail and commercial spaces. 

In response to this trend, it’s reasonable to view all cameras in public places as potential threats to privacy. Ergo, the logic goes, all cameras should be banned, regardless of their use. 

A more nuanced approach that doesn’t compromise the core value of privacy is possible. But it does require considering how to secure people’s privacy while still enjoying the many benefits public cameras can enable.

But before we go there, we need to back up and look at what we’re actually talking about when we use the word “camera.” The kind of cameras most of us think of when we see a camera on a light pole are those whose purpose is to record and preserve images—to surveil. 

Yet not all cameras are installed for those purposes, and in fact some cameras improve conditions in our cities. For example, traffic cameras can provide timely reports to commuters in the event of an accident or road closure, and speed cameras in cities have been shown to reduce traffic collisions, on average, by 15%.

My company deploys cameras to process real-time parking and traffic data. Their video sensors and computer vision software capture data rather than preserving images. The data from these cameras tells municipalities which curbs are available in real time, which helps to decrease traffic and pollution. 

None of this would be possible if we didn’t have a way to analyze traffic patterns. To do that, we need cameras.

Of course, it doesn’t take too much imagination to see what could go wrong. There is the well-documented practice of businesses selling data to third parties, which use that data for their endless marketing mills. And then there is the potential for police departments to use camera footage to racially profile.

Given these threats, we shouldn’t just reflexively trust the private sector to do the right thing. There have been too many infamous examples from the past decade of tech companies promising to be responsible caretakers of data, only for a whistleblower to reveal that such promises were broken. (Our company works with a third-party auditor to independently verify our removal of personally identifiable information from all devices.)

This misbehavior by some of the most powerful tech companies then generates a broader suspicion of all tech initiatives—some of which do benefit the public. One way to solve this mistrust is for companies to commission an independent audit of responsible data management.

Another strategy is to implement best practices around the collection and storage of data. Companies and municipalities should collect only the minimum amount of personally identifiable information necessary to achieve a particular goal, and store it for the minimum amount of time necessary to achieve that goal. For instance, when enforcing parking rules, it is necessary to record images of a vehicle’s license plate when it enters and departs a parking spot along the curb in order to invoice the driver. But those images should be permanently wiped as soon as fees are paid or the violation is resolved. All faces, as well as license plates belonging to surrounding vehicles, should be blurred in these images. 

In short, businesses should explicitly circumscribe the type of data they collect and how long they store it. 

We need a more nuanced approach to the use of cameras in public spaces. We know they exist; now the public needs a better understanding of who owns them and what they’re being used for. Complete honesty in answering these questions will be necessary if tech companies want the public to trust them again.

Prajwal Kotamraju is cofounder and head of computer vision at Automotus.