Because there are no federal restrictions on facial recognition technology, the vacuum is being filled with forms of enforcement that could have far-reaching consequences outside immigration. In Chicago, Immigration and Customs Enforcement and Customs and Border Protection agents have been documented using the Mobile Fortify app, a smartphone-based system that detects people by scanning their faces or fingerprints in real time. This tool, confirmed by CBP as “one of many” in its arsenal, can match biometric data against multiple federal databases by merely pointing a phone’s camera at a person.
The technical capabilities of Mobile Fortify are formidable and deeply controversial. It works by capturing an image of a face, converting it into a biometric template, and comparing it to repositories that include records of immigration, criminal databases, and perhaps even driver’s license photos. As with all other face surveillance systems, the accuracy of Mobile Fortify depends on environmental conditions such as light, angle, and image quality-and studies have shown error rates are significantly higher for people of darker skin tones. Although under ideal conditions, some algorithms may reach over 90% accuracy, usage in the real world can severely degrade performance, especially without rigorous human verification.
Internal communications and congressional testimony indicate that ICE’s current practice treats a Mobile Fortify match as dispositive, even permitting officers to ignore documentary proof of citizenship when the app identifies someone as an “alien.” That approach contradicts the basic premise voiced by former NYPD Commissioner Dermot Shea: “A facial recognition match is merely a lead; it is not probable cause.” And DHS agents have reportedly scanned U.S. citizens who asserted their verbal status, detaining at least 170 citizens in 2025 despite Justice Brett Kavanaugh’s assurance that lawful residents would be “promptly let go.”
That legal landscape dramatically shifted with a ruling on the Supreme Court’s shadow docket in Noem v. Vasquez Perdomo. In a 6–3 vote, the Court stayed a ban from the lower court on immigration stops based on race, language, occupation, or location and, in essence, gave ICE the green light to target people at places like bus stops or agricultural sites and to consider “apparent race or ethnicity.” Justice Sonia Sotomayor, in dissent, warned that this risks rendering Latinos as “second-class citizens.” Although temporary, this decision is a harbinger of a lenient approach to profiling, which, when combined with real-time facial recognition, is likely to accelerate discriminatory enforcement.
The prospect of such tools being extended to election contexts raises even more explosive constitutional questions. Earlier this year, the Justice Department asked multiple states for voter roll data in the name of “immigration enforcement” and said it would dispatch federal election monitors to New Jersey and California. Forcing voters to undergo facial recognition verification before casting a ballot is statutorily unsupported, but the ease with which this technology can be moved and the willingness of the administration to push the legal edges make this scenario imaginable. Experts caution that deploying FRT at or near polling places could trigger disputes over voter eligibility in closely contested races, potentially delaying or disrupting certification processes.
The risk is compounded by documented flaws in facial recognition. The case of Robert Williams, wrongfully arrested in Michigan after being misidentified by an algorithm, is a stark illustration of automation bias in leading officers to trust faulty matches. In Louisiana, a man was jailed for a week despite glaring physical discrepancies with the suspect; in Detroit, a pregnant woman was held for hours after a false match in a carjacking case. These incidents happened under policies that at least nominally barred arrests based solely on facial recognition policies; ICE appears to be sidestepping entirely.
The regulatory landscape is patchwork. Fifteen states have enacted limits on police use of face recognition, with Montana and Utah requiring warrants, and Maryland restricting its use to enumerated serious crimes. But no comparable federal statute exists, leaving DHS and ICE free to deploy Mobile Fortify without congressional authorisation or judicial oversight. Civil rights advocates, including the U.S. Commission on Civil Rights, have sounded the alarm that “unregulated use of facial recognition technology poses significant risks to civil rights, especially for marginalised groups.”
The stakes are clear: without enforceable guardrails, the very same smartphone scan that today determines who ICE detains could tomorrow be repurposed to decide who votes. In each of these arenas, the convergence of broad profiling authority, mobile biometric matching, and the absence of federal constraints creates a volatile mix capable of eroding constitutional protections in real-time.
