Citizens are turning face recognition on unidentified police

Moves have been made to limit using facial recognition throughout the globe. In half certainly one of this collection on Face ID, Jennifer Strong and the workforce at MIT Technology Review discover the sudden methods the expertise is getting used, together with how expertise is being turned on police. 

We meet: 

  • Christopher Howell, information scientist and protester. 


This episode was reported and produced by Jennifer Strong, Tate Ryan-Mosley and Emma Cillekens, and Karen Hao. We’re edited by Michael Reilly and Gideon Lichfield.




Strong: A couple of issues have occurred since we final spoke about facial recognition. We’ve seen extra locations transfer to limit its use whereas on the identical time, colleges and different public buildings have began utilizing face I-D as a part of their covid-prevention plans. We’re even utilizing it on animals and never simply on faces with similarities to our personal, like chimps and gorillas, Chinese tech companies use it on pigs, and Canadian scientists are working to determine whales, even grizzly bears.

In different phrases, the variety of methods we’d use this expertise is exploding as are considerations about whether or not that’s possible, not to mention a good suggestion. And so, bans on how face I-D can be utilized are increasing, from Portland, Maine to Portland, Oregon, the place the hardest restrictions on facial recognition within the nation are set to take impact subsequent yr banning not simply its use by police, however business purposes as effectively. 

Wheeler: Colleagues, we are right here this afternoon to think about two ordinances that search to ban using facial recognition applied sciences by our personal Portland metropolis authorities and by non-public entities within the public areas. 

Strong: That’s Portland Mayor Ted Wheeler opening town council assembly that handed these payments in September. As with most issues as of late, the general public feedback and vote happened on Zoom. 

Wheeler: Portland is way from an anti-technology metropolis, and I wish to make that very clear. We are not anti-technology with a big selection of native and nationwide tech corporations. We are one of many quickest rising tech hubs anyplace on the West coast. 

Strong: Over the subsequent few hours attorneys, software program engineers, involved residents, and enterprise leaders all had their say. Then, simply earlier than the vote happened. One final particular person, a neighborhood information scientist, raised his digital hand to ask a query.

Wheeler: Christopher Howell… final however not least, welcome.

Howell: I want to categorical a conditional help for this ordinance, however I’ve considerations.

Strong: That’s as a result of he’s constructing one thing that makes use of facial recognition in a lower than typical approach.

Howell: I’m concerned with creating facial recognition to the truth is use on Portland police officers since they are not figuring out themselves to the general public and so they are committing crimes. Would this grow to be unlawful in case you move this ordinance? 


Strong: I’m Jennifer Strong and over the subsequent a number of weeks we’re going to do one other deep dive on the rise of facial recognition. In this newest miniseries, we’ll discover the way it’s being utilized in sports activities each on athletes and followers, its use in retail shops even the way it’s used to serve the homeless. But first, we kick issues off with a have a look at the way it’s being turned again on police.


Strong: The anonymity of police and different authority figures is steeped in a very difficult historical past. There’s a fragile stability of privateness and accountability on each sides for protesters and for police.

[Sound of Chicago riots.]

Strong: The Chicago Riots of 1968 interrupted the TVs of Americans all around the nation as they tuned in to observe the Democratic National Convention, which was held in Chicago that yr. 

[Sound of Chicago riots.]

Strong: Protesters had been demonstrating towards the Vietnam battle and towards the Democratic Party.  After clashing with the police at Grant Park, protesters marched down Michigan Avenue to the lodge the place lots of the conference delegates had been staying. For 17 minutes, America watched stay because the Illinois National Guard fired tear gasoline… because the police beat the demonstrators. 

[Sound of Chicago riots. “The whole world’s watching”]

Strong: The police weren’t carrying title tags and that is truly fairly frequent when police cope with protesters. Sometimes it is authorized, generally it is unlawful, and generally it’s even mandated within the curiosity of security like this summer season in Buffalo, New York.

Reporter: The Buffalo Police commissioner says it was his determination to have officers put on their badge numbers and never their names on their uniforms.  

Commissioner: There was a rising harassment concern with officers and their household. In order to permit officers to do their job with out worry, I made the choice at the moment.

Strong: But when it occurs, residents and activists push again. Many argue this provides police much more energy than they have already got…as a result of it turns into almost unattainable to carry them accountable.

Interviewee: We want our council to face up and converse for the 255,000 folks you characterize and encourage and demand that this police drive change the coverage again as a substitute of considering solely of the dozen or so officers who had been harassed, who may very well be protected underneath the present present legislation.

Strong: But, effectively earlier than the occasions of this summer season, folks have tried to determine unidentified police officers. They’ve used images from protests, sleuthing on social media, crowdsourcing by way of the web even resorted to stealing private data. As within the case of a hacker collective – known as Anonymous. It’s identified for aligning itself with the Occupy Wall Street protests and briefly taking down the New York Stock Exchange’s web site. That identical group later leaked non-public data it stole from police and authorities web sites together with the house addresses of police officers. 

Prank Call: You have reached the Baldwin County Sheriff’s Office… [ring]

Strong: In this prank name, they declare credit score for the hack.

Prank Call: Sheriff’s workplace….. On the eu continent.. Bring it on.. Ok goodbye…

Strong: Unmasking these wielding drive over others isn’t distinctive to the U.S. Back in 2014 throughout Russia’s annexation of Crimea you would possibly recall that troopers with out badges on their inexperienced uniforms seized management. At the time prime officers together with President Vladimir Putin repeatedly denied that these troops had been Russian.

Putin: There are no troops in any respect. No Russian Troops no less than. 

Strong: This is Russia’s ambassador to the EU chatting with reporters.

Ambassador: The United States being within the custom of interfering in different international locations and sending troops abroad, could also be appearing in response to their very own mentality… I’d say. But this isn’t a case of Russian interference. 

Strong: But it was. By matching images, doubtlessly with the assistance of facial recognition, the Ukrainian Government decided the troops had been the truth is tied to Russian army. And they launched images as proof. Later, Putin additionally admitted the troops had been certainly Russian.

Putin Translation: Of course, the Russian servicemen did again the Crimean self-defence forces. They acted in a civil, as I’ve already stated, however a decisive {and professional} method. 


Strong: Now protesters are more and more making an attempt to show Face I-D again on police – to determine officers who use extreme drive. Including in Hong Kong, the place final yr dramatic photographs of clashes between protesters and police dominated information feeds exhibiting police firing stay bullets and protesters attacking officers. This is New York Times Reporter Paul Mozur talking on public tv.

Mozur: As the protests have gone on and police and Hong Kongers proceed to sq. off week after week the face has grow to be weaponized and identification itself in a approach is weaponized. You know, protesters will exit and police will attempt to seize their photographs on video after which return and determine them by way of all of the social media and on-line supplies that are on the market after which vice versa. We noticed the police truly take their badges off and so now protesters are doing the identical to the police the place they are making an attempt to return and use social media and determine which police are doing what act. //One protester specifically this man Colin Cheung that we discovered created a facial recognition instrument to attempt to determine police… and he didn’t truly launch the product however he says due to that police focused him.

Strong: And Cheung was arrested shortly after he posted in regards to the instrument on Facebook…  however comparable efforts are ongoing elsewhere, together with Portland, Oregon. 

Howell: I’m taking the very same expertise they use on us. And it isn’t like I’m making an attempt to dox folks. I imply, that is for officers that are primarily breaking the legislation by way of their use of drive. 

Strong: Christopher Howell is a knowledge scientist, and protester. 

Howell: And I simply wished to get on the document. Hey, there’s one other approach we may use this. // And actually it is extra about, you realize, preserving the stress on them to determine themselves versus us having to do it. 

Strong: His testimony earlier than the Portland metropolis council about this instrument he’s working on led to his mission getting lined by The New York Times. Since then, he’s gotten a good bit of consideration as did the response of Portland’s mayor, who known as the mission creepy.

Howell: It’s creepy to allow them to exit and tear gasoline folks night time after night time // and I, that was actually, to me was only a, sort of a thoughts blowing second of like, it is creepy that these guys haven’t got their names. You know, that somebody can simply costume up like a policeman you realize [laugh]

Strong: He says he began the mission partly simply to see how it could work.

Howell: I imply, there is a particular technological curiosity there… and that kind of, one thing I can do when it is late at night time and it is irritating to see these things, the information otherwise you go on Twitter and see folks’s, everybody’s movies of the protests and okay, effectively, what can I do? …and in the end I would really like it to not be crucial. I feel they need to put on their names in tall letters in order that we cannot have to make use of facial recognition on cops to strive to determine who they are.  

Strong: Being capable of determine an officer isn’t nearly figuring out if somebody truly is one… It’s extraordinarily troublesome to file a criticism towards somebody with out figuring out who that particular person is.

Howell: One of the explanations I began this was as a result of in lawsuits, you may’t title just like the cop who shoved me. You need to know who it’s. They’ll simply say, effectively, we do not know who it was. So there’s no person so that you can sue and making an attempt to get information on what specific officers did they won’t allow you to go fishing for it. // You must know so, in case you gave me an image of, you realize, this is this cop hitting someone with a baton I wish to, and we will determine, you may see his face. So we will determine who it’s, how will we do this? And initially I used to be extra considering on a reverse picture search on a database. Like I’ll simply acquire in all the images I can, and we’ll have the ability to hyperlink them collectively so we will say these are all the identical particular person. And then I sort of realized I may do higher than that. 

Strong: He realized he may construct a face ID instrument, utilizing photographs of the officers from the web.

Howell: So I began taking a look at their photos the place they are recognized or information, you realize, going again additional and getting like information articles.

Strong: The mission is technically simpler than constructing a system meant to determine anybody.

Howell: We know that who the police are, you realize, there aren’t, there should not be folks in riot gear who aren’t police officers. So if you may get all of their photos and that is the true problem, however then you would have, a more easy mathematical drawback in that sense to determine, okay, we all know this particular person is out of this set. Let’s discover out which certainly one of them they are…

Strong: He says he’s collected hundreds of images to date…principally manually from Twitter and information articles…on common 15 to twenty or so for every officer he’s recognized.

Howell: Quite a lot of the preliminary photographs had been from Twitter. Quite a lot of it was me going into information articles. // I imply, the site visitors accident, one is, you realize, information story is such an ideal instance. Cause I discovered a bunch of these. Or like group barbecue and there are uniformed officers there and it names them.

Strong: It’s not precisely the identical scale because the face ID methods utilized by police…

Howell: I’m simply all doing this proper now on a Python script – you realize, it is all native on my laptop computer. 

Strong: …these are often skilled on picture units of tens of millions or billions.

Howell: Because I do suppose the, the accuracy is essential and, uh, and the variety of photographs is sufficiently small that if I used to be making an attempt to place my face in there, I do not wish to get one other Chris Howell.

Strong: But it’s simply sufficient photographs that it appears to do the trick. He says his instrument has already been used to assist affirm the identification of an officer in a case that’s headed to courtroom.

Howell: Somebody introduced me an image I hadn’t discovered and stated, assist me, you realize, you utilize your system and inform me what it says. And she already knew who she thought it was, however she did not inform me. And then the highest outcome was the one which was anticipated.

Strong: As for the place this goes subsequent, that’s anybody’s guess, although he can think about a day when it is perhaps utilized in partnership with town. 

Howell: I may see a future, perhaps years from now, when issues are a bit totally different the place town makes one thing like this out there and hosts it themselves and makes them take a bunch of images so we will have a well-trained factor. So as a like citizen suggestions factor. And, and it would not essentially need to all be detrimental issues, however in a approach that it is also used for complaints. I imply, a part of me thinks there’s there could be a use they’re partnering with the federal government businesses themselves to say, “Hey, we want to be accountable.” I do not suppose that is very reasonable in 2020, however I feel it, it may very well be not distant! [laugh]

Strong: Something he doesn’t need is to make it open supply. 

Howell: I do not suppose it is an ideal thought for anyone actually to have it simply be like public dealing with on a web site. Think that is, it is asking for issues to go fallacious past just like the actually apparent issues you would have, folks, you realize, faking issues or deliberately placing in deceptive photos…


Strong: Portland’s bans on face I-D will take impact in January however Howell’s mission? It received’t be impacted. Public colleges, spiritual establishments, even how facial recognition is used on the airport, corresponding to the best way airways like Delta board its passengers. None of that will probably be impacted both. Local authorities businesses received’t have the ability to use it. And it received’t be allowed in most public areas or non-public areas that are open to the general public, like a shopping center. Local police won’t be allowed to make use of the expertise both although folks of their non-public properties, like Christopher Howell, will probably be. The bans additionally don’t apply to legislation enforcement on the state or federal degree.

In a approach, using this surveillance expertise is changing into a kind of international arms race between the general public and authority figures, each hoping to peel again the quilt of anonymity to encourage good conduct. It’s laborious to see who the winner is perhaps, however actually the loser is privateness.

In the subsequent episode… 

It’s not trauma knowledgeable to have someone stroll right into a facility and say, sure, you may completely are available in, however let me simply take your fingerprints. 

Strong: We have a look at the transfer to make use of facial recognition in public housing and homeless shelters. 

Facial Recognition is kinder emotionally in that it is passive, it, does not contact them and you may seize it extra shortly and there isn’t any threat of transmission.  

Strong: This episode was reported and produced by me, Tate Ryan-Mosley, Karen Hao and Emma Cillekens. We’re edited by Michael Reilly and Gideon Lichfield. 

Thanks for listening, I’m Jennifer Strong. 


We will be happy to hear your thoughts

Leave a Reply

Reset Password