Are Police Body-cams the Next Step to a Panopticon Society?
Face recognition is gaining prominence in the news, which is a good thing. People need to understand what this technology can and can't do. A recent article predicted that police body cameras are going to make life a "perpetual line-up."
The article doesn't go into detail about this at all, but I can tell you how it would work in theory. Police body-cams should be on all the time while they are on duty because you never know for sure when something important might happen. If you wait for the officer to turn on his body-cam it might be too late. So that means you have around 8 hours of video per day, per officer. This video can be stored on a vendor site or at the police department.
When the office sends his video to HQ, a back-end process scans the videos and finds all of the faces in them and converts them to biometric templates. This includes video of criminals, of course. But it also includes video of people at the mall, at the gas station, everywhere the officer went. Lots and lots of people who are not even accused of a crime. But all of these templates get stored in a biometric database. For most of these, the database doesn't know who they are, they are just "person143124". But, if they have some other database to link to, they might be able to know that person 143124 is Alex Kilpatrick.
Now some crime happens - maybe a bank robbery. They recover security camera footage and run it against their database and find a match to Alex Kilpatrick. Now they come to my house and question me, arrest me, whatever, depending on how much evidence they have. It is worth noting that they wouldn't be able to arrest me solely based upon a face match. But it still might cause me a lot of problems. If they don't know who person 143124 is, they might be able to use their database to know I am always at Starbucks at 4:30 on Fridays, so they can look for me there.
The linked article has a great quote that brings this home by analogy with our current situation:
There is a knock on your door. It’s the police. There was a robbery in your neighborhood. They have a suspect in custody and an eyewitness. But they need your help: Will you come down to the station to stand in the line-up? Most people would probably answer “no.”
Few people would want this. But at the same time, if you were falsely arrested because you happened to look like a criminal based upon an eye-witness report you would understand the mistake. What's the difference? In a word: scale. The eye-witness false arrest is going to affect at most 1 or 2 people. For that to happen you would have to have the bad luck to look like a criminal AND be in the wrong place at the wrong time. It is awful if it happens to you, but it won't happen often.
With the automated solution they can look for "potential matches" at a much bigger scale. They might decide for a large-scale crime they are going to go look at the top 20 face matches from their database and interrogate those people. Even worse, maybe they are going to start surveilling the top 10 people all day for a week. That doesn't sound like a free society to me.
What's the answer? I think we can mirror some of our existing laws that protect innocent citizens from unreasonable search. I want the police to be able to archive video and use facial recognition technology, and I want them to be able to look at a video from 5 years go if it might help solve a crime. But at the same time I don't want that video to become a privacy destroying database just because it exists. Doing a face search against a video database needs to be limited, tightly controlled and regulated, and only done under a warrant. We will still have innocent people occasionally arrested (just like we do with regular search warrants now), but with a regulated, warrant-based system we will be able to monitor and hold police accountable for following the rule of law in how they access it.
It is always hard to know how a new technology will affect society and what the best policy is for preventing its abuse by government. However, the very first step is to make sure government use of surveillance/biometric technology is not being hidden.