Op-ed | Face Recognition and the ‘Trump Terror’: A Marriage Made in Hell

ICE and CBP are smashing their way not only through car windows but also through any constraints on the use of face recognition

by Jay Stanley, Senior Policy Analyst, ACLU Speech, Privacy, and Technology Project

Face recognition is a dragnet surveillance technology and its expansion within law enforcement over the last 20 years has been marred by systematic invasions of privacy, inaccuracies, unreliable results, and racial disparities. As some parts of law enforcement have rushed to deploy it, this technology has sparked deep opposition, and over 20 jurisdictions across the country have banned their local police from using it.

Now ICE and CBP appear to be aggressively adopting face recognition as part of the destructive path they are bludgeoning through American communities and the Constitution. The ACLU has been calling these departments “rogue agencies” for years, and now their lawlessness — long present at the border— is not only moving into our cities but also merging with the nightmarish capabilities of face recognition technology (which has already been over-used and mis-used by normal American law enforcement agencies). Face recognition and the “Trump Terror” deportation drive is truly a marriage made in hell.

This marriage comes in the form of a new mobile app being used by ICE and CBP called “Mobile Fortify.” The app, which was only made public through leaked emails and documents obtained by 404Media, allows agents to point a phone at anyone in public, compare their faces against a variety of government databases containing 200 million images, and get instant access to their name, date of birth, and a potentially deep well of intimate data contained within those databases. It also allows the contactless collection of fingerprints.

One program document declares ominously, “ICE does not provide the opportunity for individuals to decline or consent to the collection and use of biometric data/photograph collection.” It’s unclear exactly how these agencies are using the tool (see below), but this raises serious constitutional and ethical issues. Very few law enforcement agencies have attempted to harvest people’s biometric information during stops on the street, but in a recent ACLU case one court ruled that police officers cannot take people’s fingerprints during such stops, even if they have reasonable suspicion of criminal activity (the low standard of suspicion that is required to carry out a brief stop to question someone, aka a “Terry Stop”). And while it’s easy to photograph people in public places, extracting a unique biometric identifier by doing a face recognition scan is akin to a fingerprint and should follow the same rules. (And the government should not be intimidating people peacefully exercising their First Amendment rights by photographing them at all, with or without face recognition.)

The use of this technology on anyone, regardless of immigration status, raises constitutional issues, but it’s important to note that the impact here will not be limited to people who are undocumented. The same document asserts that officers

may use Mobile Fortify to collect information in identifiable form about individuals regardless of citizenship or immigration status. It is conceivable that a photo taken by an agent using the Mobile Fortify mobile application could be that of someone other than an alien, including US. citizens or lawful permanent residents [LPRs].”

The fact that this dragnet catches citizens and noncitizens alike is not an operational error — it is by design. The photos collected by Mobile Fortify of U.S. citizens and others with legal status are emphatically not discarded — they are retained as part of DHS’s Automated Targeting System (ATS), the foundation for DHS’s myriad watchlists. As the documents acknowledge: “CPB will retain all photographs, including the non-match photographs (to include US. citizens /LPR photographs), as part of ATS holdings.”

In short, this program unleashes roving agents across the country to take photographs of anybody, citizen or not, and to retain those photographs in government databases, whether they are a match or not.

“Definitive” identification
ICE and CBP (specifically its Border Patrol subagency) are also heedlessly trampling the need for careful treatment of any face recognition results. Recognition of that need, which arises from not only the technology’s power but also its unreliability —is widespread within the American law enforcement profession. Yet Congressman Bennie Thompson, ranking member of the House Homeland Security Committee, reported that “ICE officials have told us that an apparent biometric match by Mobile Fortify is a “definitive” determination of a person’s status and that an ICE officer may ignore evidence of American citizenship—including a birth certificate” when the app says a person is undocumented.

If true, this practice is utterly, jaw-droppingly irresponsible. Face recognition is fussy and unreliable even in the most controlled conditions. But a study earlier this year by the U.S. National Institute of Standards and Technology (NIST) noted specifically that compared to controlled conditions, border crossing photos “present challenges for face recognition” because

subjects often exhibit non-zero yaw and pitch (associated with the rotational degrees of freedom of the camera mount), low contrast (due to varying and intense background lights), and poor spatial resolution (due to inexpensive cameras). There are often subjects standing in the background, usually at very low resolution.

Those kinds of characteristics are even more likely to be found in photographs snapped on the street using handheld phones than those taken by CBP equipment at the border. And as studies have consistently shown, face recognition is also less accurate when used on Black people and others with darker skin, and less accurate with women than men.

Because of such problems, police departments around the country are careful to say that they use face recognition only for such purposes as “lead generation” and that officers should always verify a person’s identity through means other than a face recognition match. It’s certainly true that police officers and departments around the country have failed to live up to those promises and to use the technology with the care it merits. The result has been innocent people that we know about (virtually all Black), and probably others we do not know about, who have been wrongly arrested and jailed because of face recognition errors. All that is bad enough. Still, those cases have resulted in nationwide news coverage, condemnation, and lawsuits. To allow officers participating in the current deportation effort to stop, arrest, detain, or deport someone — perhaps violently, as we’re seeing all too often — based on this technology would be a severe violation of due process, human rights, and the Constitution.

All of this raises the chances that any of us or our children, other family members, friends, or neighbors will find themselves being photographed by one of these federal officers and suffering who-knows-what consequences due to some false match or database error. Given these agencies’ unabashed racial profiling, that is especially true if we are Hispanic or an immigrant of color from anywhere in the world— or just appear to possibly be in the eyes of a none-too-conscientious federal agent of the kind we are seeing roam our streets.

Vast amounts of data
Aside from the hair-raising prospect that life-changing official decisions will be made based on false face recognition matches, this app reportedly also lets agents in the field access a stunning amount of data about the subjects of those matches. The data contained within the databases that it reportedly incorporates include:

• Face prints, iris scans, fingerprints, DNA records, and other biometrics, as contained in the Automated Biometric Identification System (IDENT, aka HART), a giant DHS database that holds records on more than 270 million people.
• Photographs taken when people enter or leave the United States, as collected by the Traveler Verification Service, CBP’s misguided entry/exit face recognition system — or even during the course of domestic air travel, as TSA Precheck facial recognition scans are shared with CBP for verification. In both cases, TSA and CBP currently promise to delete U.S. citizen’s photos, but that protection only exists by whim of TSA and CBP internal policy.
• Local, state, federal, and foreign criminal records and accusations of gang membership, as contained in the giant NCIC (National Crime Information Center) database, the FBI’s primary national criminal database.
• State driver’s license databases and other state data through access to an interstate sharing database called Nlets, which is under scrutiny by members of Congress. U.S. Citizenship and Immigration Services (another DHS component) is actively building a centralized database of U.S. citizens built on Nlets data, passport numbers, Social Security numbers, and more.
• Travel, banking, social media, and license plate reader data, as well as any data collected as part of any immigration application or proceeding, as contained in a CBP database called TECS (not an acronym), the agency’s giant catch-all information storehouse. Part of TECS is the mysterious Automated Targeting System (ATS), which is used to make unaccountable algorithmic security judgments about U.S. citizens and non-citizens crossing the nation’s border. We don’t even know what data is used by ATS.
• Commercial data broker data (possibly). User manuals obtained by 404Media state that “currently, LexisNexis is not supported in the application,” strongly suggesting that such support was in the works when the manual was written. LexisNexis is one company in a giant, ethically sketchy industry whose business is to collect vast dossiers of information on nearly everyone in America, including a lot of sensitive personal information.

And these databases contain errors. In 2019 a federal court ruled that ICE’s own databases are so unreliable that they could not serve as the basis for probable cause warrants against detention targets. The NCIC database is also notoriously inaccurate, and people have been badly harmed as a result. Federal law requires that government agencies maintain records used to make “any determination about any individual” with “such accuracy, relevance, timeliness, and completeness as is reasonably necessary to assure fairness to the individual in the determination.” That doesn’t seem like too much to ask, yet the FBI felt compelled to exempt the NCIC from that law. DHS did the same with respect to the ATS program. Meanwhile, data brokers hold a lot of error-ridden “dirty data” and make a lot of inferences that are “inaccurate trash,” as one reporter summed it up after an investigation.

In short, just by pointing a phone at you, an ICE or CBP officer can tap into a potentially vast amount of data — more personal information in once place than any government official should be able to get, and more than Congress intended for any official to get when it enacted the Privacy Act. And that information is not even necessarily correct.

Other problems with Mobile Fortify
There are other problems with this technology:

• It’s unauthorized. Congress has not authorized this dramatic expansion in federal police power, nor has there been any meaningful public debate over it. Do the American people want to live in a world where the authorities have the power conveyed by this app? They have not spoken.
• A lack of checks and balances. This app is being deployed without any clear statutory restrictions, usage auditing, or accountability mechanisms, so far as we know.
• Questionable security. A 2024 audit by the DHS inspector general found that ICE “did not effectively manage and secure its mobile devices or the infrastructure supporting the devices,” creating a risk of cyberattacks and unauthorized access to sensitive information. The reported power and scope of the Mobile Fortify app would raise the stakes of such sloppiness even higher.
• A lack of transparency. Americans only found out about this dramatic expansion of face recognition because of leaked documents and reporting on a few social media videos by people subjected to face scanning during stops by immigration agents, leaving many unanswered questions about the program.

Unanswered questions
In September, nine senators wrote a letter to the ICE director demanding greater clarity about Mobile Fortify. There are many questions around this technology that the public is owed answers to. They include:

• How are officers using the app? A “Privacy Threshold Analysis” (PTA) of the program obtained by 404Media says vaguely that “to conduct identity verification during an ICE operation,” ICE agents “will use the Mobile Fortify app… to take a photograph” when they encounter “an individual or associates of that individual.” Is it being used only to confirm the identities of people previously selected for targeting? Is it used on random people that agents encounter and decide might look “foreign”? Is it being used on protesters? Does the reference to “associates” mean that if an officer picks somebody out, they will photograph everyone else nearby? Not only are these behaviors not subject to statutes, rules, or guidelines that we know of, but we don’t even know if they’re taking place.
• Is usage confined to ICE and CBP? Social media videos viewed by 404Media suggest that not just ICE but also CBP officers are using face recognition in some of the abusive immigration operations we’re seeing. Some report that CBP is at least as bad and perhaps even worse than ICE, which owns the formal role of locating and detaining those the government wants to deport. Are personnel from any other federal agencies that are participating in the immigration operations using the app?
• Is usage of this app spreading to local police? From the beginning a major question about this program is whether local police who are cooperating with the federal deportation drive are also being provided with this app. That cooperation takes place through a program called 287(g), which the Trump Administration has massively expanded to over 1,000 agencies, up from only around 150 at the end of the Biden Administration. 404Media reports that local police have, in fact, been given what appears to be a limited face recognition app that doesn’t involve all the data that Mobile Fortify can access. We know little about this app, however.
• Are the agencies incorporating any safeguards? Are ICE and CBP implementing any risk management practices or procedures to protect civil liberties? Are they logging and analyzing its usage? Do they allow its use on children? How are images collected by the app shared if at all? Further, official Trump administration guidelines on high-risk AI — including facial recognition technology — requires public use case inventories and for agencies to mitigate harms to civil rights and liberties from the AI. Are ICE and CBP complying with the administration’s requirements?
• What is ICE and CBP’s current policy governing face recognition technology, and does Mobile Fortify violate it? In 2023 the Department of Homeland Security issued a detailed policy governing use of face recognition technology by ICE, CBP, and other agencies. But the policy quietly disappeared from the DHS website early in the Trump Administration, with no replacement yet posted. Is that policy still in effect? Does DHS still enforce its core requirements, including that face recognition technology “may not be used as the sole basis for law or civil enforcement actions”?
• Are the agencies testing or tracking the app’s accuracy? Have ICE and CBP carried out real-world testing of the app and its accuracy rate? Are they tracking its in-the-field error rate now?
• What face recognition algorithm does the app use? Studies of face recognition by NIST reveal wide variation in the accuracy of different algorithms. Who did ICE contract with to build the app and what was contained in that contract?
• Is the use of the app being disclosed in legal proceedings? Is the use of, and information about, the app and the face recognition algorithm it uses being disclosed to immigration judges and magistrate judges considering whether to grant arrest warrants being informed that the app was used? Is it being disclosed to criminal defendants, respondents in immigration court, or others subject to arrest or apprehension on whom the app is used?

“The border is everywhere”
ICE and CBP are smashing their way not only through car windows across the nation but also through any constraints on the use of face recognition. Their use of this technology represents the merging of the abusive ICE/CBP Trump Terror with the most powerful surveillance technology available today. It also represents an importation of abusive Border Patrol practices into the rest of the country (“The border is everywhere now,” as one official declared) and appears to be characterized by what one analyst termed the Trump Administration’s “hostility to rules-based administrative regulation.”

The Mobile Fortify program represents a dangerous expansion in the government use of face recognition in American life and would fundamentally reorient the relationship between the authorities and individuals in this country if it is allowed to continue. It must not be.

Categories:

Tags: