A small US company is selling an app that might end our ability to walk down the street anonymously. Among its clients: authoritarian states and US immigration enforcement. Jacinta Gonzalez, an organizer with the NGO Mijente, talked with us about why this puts the 11 million undocumented people in the United States at an even higher risk for deportation.
In January, a New York Times article revealed that 600 US police agencies use a tool by a previously unknown company, Clearview AI, to instantly identify almost anybody based on their face. The small New York-based company admits having built a database of more than two billion images scraped from the internet, mostly from Social Media sites, without permission or notice. After taking a biometric scan (a “faceprint”) of a person, it combs through these databases to match it with all faces with identical features.
This is a technique that even Google, the “pioneer of surveillance capitalism” according to Shoshana Zuboff’s seminal work “The Age of Surveillance Capitalism” (p. 16), displayed reservations against. Google also offers image search and facial recognition, but keeps both has separate. While it does have the capacity to combine both techniques, ethical safeguards kept it from developing a Clearview-like tool. Way back in 2011, then Google CEO Eric Schmidt said it was the one technology the company had held back because it could be used “in a very bad way”. He predicted that sooner or later, a company would, however, cross “the creepy line”. Nearly a decade later, the NYT article shows that he was right.
For the first time in history, an app has been introduced that “could end your ability to walk down the street anonymously” as the NYT author writes. So far, the surveillance apparatus built by the Chinese government has been understood to be unmatched; it has become infamous for violently targeting the Uighur minority and everybody with a low “social credit”. But in all of this, China still relies on DNA samples that can only narrow down a manhunt but not provide definite matches. The Clearview app is more precise than anything used in China so far.
As if this was not enough, the full extent of the scandal only became clear when Clearview’s entire client list was stolen and published a month later: Contrary to the company’s claims, the app has not at all been exclusively used by US law enforcement. Instead, more than 2,900 organizations are using the software according to BuzzFeed News who reviewed the list, including businesses, schools, foreign governments in countries such as Saudi Arabia – as well as immigration enforcement.
All these clients warrant a critical look. However, the leaked information suggests that the 11 million undocumented people in the US are at risk of being systematically targeted with the help of this app – especially given US president Donald Trump’s promise to his voters to deport as many of them as possible, no matter their background. According to BuzzFeed, agents working for the US Immigration and Customs Enforcement (ICE) have made 8,000 searches via 60 different accounts registered in New Jersey, New York City and the border town El Paso; Customs and Border Protection (CBP) made 7,500 searches via 280 accounts.
The Latinx advocacy group Mijente has worked at the intersection of big data and immigration rights for years. The leader of its “No tech for ICE” campaign, Jacinta Gonzalez, spoke with us on the phone about how the Clearview app might be used to target undocumented immigrants.
The interview was edited for length and clarity.
NUDGED: As we learned last month, the immigration enforcement agencies ICE and CBP have both secretly made thousands of faceprint searches with a previously unknown app. Jacinta, were you surprised to learn about this?
Jacinta Gonzalez: I find it terrifying that a small company with few employees can build such a powerful, invasive tool that dramatically shifts the way that immigration enforcement looks like in this country without any sort of oversight or accountability. However, I was not very surprised. In recent years, small tech and data companies have been building very powerful tools and handing them over to police departments and federal agencies, without any sort of protection for people’s basic rights.
A major example of this is the Silicon Valley tech firm Palantir, which was co-founded by “surveillance enthusiast” Peter Thiel who invested in Clearview AI and happens to be an outspoken Trump supporter and high-level donor for his 2016 presidential campaign.
Definitely. Since 2014, Palantir provides ICE with the Falcon software tool which has been used to process large amounts of data on undocumented people and to plan workplace raids as we have shown – contrary to the company’s claims.
In what way does the Clearview app go beyond the Palantir software?
The Palantir app requires ICE agents to have a name before they can pull up immigration databases to see whether that person has already been tracked as an undocumented person. This means the agents rely on tips – as in the Mississippi workplace raid where anonymous sources blew the whistle on people working in the factory without documentation – or they have to stop people and ask for their passports.
By contrast, Clearview allows for an immediate, single-handed, secret background search. An ICE agent could just walk into a supermarket, select a worker based on racial profiling, take a picture without her noticing and use the Clearview app to pull up all pictures of her face that are visible to a web scraper or public API.
Even if she does not have a Social Media profile?
Yes. No matter if she appears in a photo published by her former college friend on Facebook or in a photo of a demonstration published on an NGO website, all of those will come up. If her face has ever been linked to her name online, the Clearview app will immediately let the ICE agents know who she is.
But what would they do with her name and photo? Can they run it against a database with all US citizens, residents and tourists, and if she does not come up, they assume that she must be undocumented?
No, usually it’s not a process of elimination but a process of evidence. Most undocumented immigrants in the US have had some sort of contact with either immigration enforcement or law enforcement: Many of them overstayed their visa, many have registered under the Deferred Action for Childhood Arrivals (DACA) program [and will lose that protection if the Supreme Court decides to end it later in 2020], some have been deported and have come back to their families, others lost their Temporary Protected Status (TPS) after the Trump administration declared their country of origin safe again, and yet others have applied for asylum or for a U visa for victims of crime and lost. All that information is already in the ICE database and it will come up.
If they take a photo and see that she is undocumented, what will they do?
They can either arrest her on the spot, or they also look at her colleagues, make a target list of undocumented people working in that supermarket, and then use that list to justify and facilitate a workplace raid. In both cases, her arrest will have been based on racial profiling which is prohibited in most US states.
Did this ever happen?
Technically, this can already have happened. But we would not know it, because there’s no accountability and transparency around how the Clearview app is being used. Without the leak in February, we would not even know that ICE agents are using it.
Will the Clearview app even be used in sanctuary cities like San Francisco or New York or on sanctuary campuses, e.g. of the UC Berkeley, who promise to protect their people from deportations?
Yes, ICE agents can technically use the app wherever they want. Clearview AI undercuts sanctuary cities and makes their policies irrelevant. What’s more, it can turn local and state-wide measures geared toward protecting undocumented people like municipal ID-cards, driver’s licenses or health benefits against them. To benefit from these measures, immigrants had to have their photos taken – which can now probably be accessed by Clearview as well. Paradoxically, what has previously been a protection, can now backfire.
Does it mean that photographers and journalists should be cautious when uploading photos of demonstrations or immigrants’ rights events because Clearview will scrape these photos?
Not only the press but we organizers ourselves have to be more cautious and weigh our right to free speech against safety concerns. How many times do we publish pictures of our protests? But by doing this, we’re handing over information to ICE. Technically, the Clearview app can reveal everybody’s name and digital history. And given that the Trump administration has already persecuted immigrants’ rights activists in the past, we are worried about ourselves as well.
Barack Obama focused on deporting only the few undocumented immigrants whom he deemed security threats. By contrast, Trump promised his voters to deport millions of undocumented people. Do you know how ICE prioritizes arrests?
ICE agents are currently arresting whomever they identify as undocumented. They have quadrupled their workplace raids during the Trump administration. They are even using the lockdown during the current Corona pandemic for home and neighborhood raids, going from door to door to make arrests, with no regard as to whether immigrants will get infected. Any raid is usually justified with a search for a so-called “criminal”, most often somebody working with a fake Social Security Card, but if they find other undocumented people in the same place – which is usually the case – they take them in as so-called “collateral arrests”.
ICE’s actions are inexcusable. As people have gone to hospitals & rushed out to get food, ICE has arrested them, in the middle of a public health crisis, & shuffled them into overcrowded, unsanitary detention centers.
Their actions will get people killed. They must stop. pic.twitter.com/8KNiKK6Qh4
— Mijente 🐜🐜🐜 (@ConMijente) March 19, 2020
It is unclear since when the Clearview app has been used. But did the number of deportations increase because of the Palantir software?
It is very difficult to get accurate numbers because deportations from the border and from the interior were counted differently in different years and under different administrations. What we do know is that the Obama administration started using analytics tools for immigration enforcement, and this correlates with the time in which interior enforcement spiked. The same infrastructure that Obama built to go after traffickers is now used against workers and their families. If you like it or not: Obama built a Cadillac and then handed the keys over to a maniac, Donald Trump.
It is difficult however to prove any causal effects between the use of a particular software and deportations. As part of their contracts with ICE, tech companies ask not to be named in any charging documents because they don’t want the public to know how the technology is being used. We are worried that this creates a parallel legal track in which ICE is relieved of its duty to prove whether it gained evidence in a legal way.
According to the information leaked to BuzzFeed News, agents have downloaded a trial version of the app on their personal phones. Why would they do this – are there any incentives for individual ICE agents to deport as many immigrants as possible?
We do know that the 22 ICE field offices in the US receive quotas of how many immigrants they have to arrest during specific raids, as we have shown in regard to Operation Mega. However, I don’t know if these translate into individual quotas. No matter what, it is incredibly worrying that agents have this faceprint tool on their private phones, which are not being monitored for abuse. There are numerous ways for this to be abused: whether it is officers stalking their (ex) partners or using it for racial profiling, given that white supremacists have tried to recruit police officers for decades.
Does the undocumented community know that they are being targeted with the help of faceprints?
They are slowly starting to understand what is happening – just like everybody else. We are just barely seeing workers inside tech companies starting to resist and regulators starting to catch up. Undocumented immigrants have a deeply ingrained fear of immigration authorities, and at the end of the day, they don’t care if they get detained because their neighbors snitched on them or because of Palantir or Clearview.
Did you ever talk with Clearview AI about your concerns?
I did not contact Clearview yet, but Palantir. They do not want to talk to us. We have asked. They won’t.
How come that data scientists from liberal, immigrant-friendly communities like California’s Silicon Valley and New York City agree to build such tools?
Workers in these companies tend to be very removed from the finished product and its intended use. New hires are kept in the dark, but for the unlikely case that they do realize what they’re working on, the companies make them sign NDA’s [non-disclosure agreements]. Apart from that, we know that the level of coercion and intimidation within these companies is very big. In this culture, workers feel nervous to take action, especially when they think that this would be inconsequential. We are all the more excited to see that, even under these circumstances, workers have started to organize. But we need more of this to have an impact.
Thank you for your time, Jacinta!
Jacinta Gonzalez is one of 15 staff members at Mijente, an NGO that describes itself as “political home for Latinx and Chicanx changemakers”. She leads their “No tech for ICE” campaign. Gonzalez has 15 years of experience as an activist on immigration and deportation issues. In 2016, she was arrested during an anti-Trump rally together with two fellow protesters. Unlike the latter, white men, she was held detained overnight at the request of US Immigration and Customs Enforcement (ICE) – although she is a US citizen.
Header image: The photomontage does not show an undocumented immigrant but Ricky Martin. (CC) Clearview Logo/Wikimedia + (CC) Eva Rinaldi/Wikimedia, photo montage: Christina Felschen
Portrait of Jacinta Gonzalez: Puente Arizona (screenshot)
Are you worried about* Clearview’s use of faceprints? Which of the app’s potential uses concerns you most? Let us know in the comments or on Twitter!
(*If you indeed worry about Clearview AI collecting and using your data, and if you live in a country or a US state with privacy laws (like the EU or California), you can follow an idea by VICE reporter Anna Merlan and request information. She sent an email to email@example.com, asked what personal data Clearview AI obtained, how they obtained it and how it was used. She also asked that all said data be deleted after it was given to her and prohibited the company from collecting any data of her in the future.
I did this, referring to my rights under the European General Data Protection Regulation (EU GDPR), and I will keep you posted about Clearview’s reaction.)