U.S. Immigration and Customs Enforcement has quietly assembled a growing arsenal of the sorts of tools the National Security Agency might use to surveil terrorists or drug traffickers: chain analysis software, location tracking devices and communications intercepts. From facial recognition to smartphone hacking and large-scale data broker feeds, the agency’s capabilities are designed to identify, verify, and act on targets at scale — often with scarce public awareness of how these systems interact.
The numbers reflect the results: CNN has documented about 350,000 removals thus far this term; around 200,000 carried out by ICE, and more than 130,000 by Customs and Border Protection — plus almost 18,000 self-removals. The technology described here is a key enabler for this speed.
Facial recognition, ‘hoverboards’ shift from lab to street
(ICE has relied on Clearview AI, the company behind a tool used to scrape billions of images from social media and public websites, to help identify people in photos and videos. New procurement records reveal a $3.75 million contract to expand access to technology supporting Homeland Security Investigations (HSI), that’s in addition to its buying burst for “forensic software” and enterprise facial recognition licenses. The appeal is clear: Investigators pivot from a blurry still image to a name, then to addresses and associates, in minutes.)
But the risks are just as apparent. The National Institute of Standards and Technology has also reported major demographic disparities in error rates among facial recognition systems. In practice, there can be a cascading effect of false matches, whether by disseminating the errant information through database look-ups, field interviews or raids — unless agencies layer in enough safeguards: human review, threshold tuning and strong auditing.
Phone unlocking and spyware broaden digital reach
On the device exploitation side, ICE signed a $3 million deal with Magnet Forensics, which developed GrayKey, an iPhone-and-Android-cracking tool used by law enforcement to unlock and extract data from Apple devices. The process is simple: plug in a confiscated phone, crack its locks where you can, extract eblasts, pin drops and tattletales from downloaded apps and deleted files, then file everything with the rest of your findings.
ICE also awarded a $2 million contract to Paragon Solutions, an Israeli spyware seller, for the “fully configured proprietary solution,” which comes with hardware and licenses as well as maintenance and training. The sale was put on hold amid a government review of commercial spyware use and then resumed. Details are scarce, like whether the system will be fielded by ICE more generally or specifically within units of the HSI that target crimes such as child exploitation, trafficking, and financial crimes.
Operationally, these are not only tools for unlocking phones — they’re tools for unlocking time. Automated extractions can turn what used to take weeks of manual analysis into hours, and that capacity can be multiplied across caseloads. The trade-off: Greater internal controls are required to prevent overcollection — and to guarantee evidence handling can live up to forensic standards if cases make it into court.
Data brokers supercharge person-finding operations
LexisNexis is now a data backbone for ICE. Copies of the records, from Freedom of Information Act requests by activist organizations, revealed more than 1.2 million searches conducted over seven months by ICE’s Accurint Virtual Crime Center, largely to create background profiles. Public records show ICE subscribes to a law enforcement investigative database that pools utility records, credit header data, property records and other commercial sources among mined identity graphs.
Critics warn this is mass surveillance-by-contract that enables the government to circumvent traditional warrant processes by buying commercially available data in bulk. LexisNexis says it endorses responsible and lawful use and points out that it works with thousands of public safety agencies around the country. The scale of the program is significant: Since January, ICE has paid about $4.7 million for these subscriptions this year.
Palantir is the analytics backbone for ICE operations
The company’s Investigative Case Management (ICM) platform serves as the central nervous system linking much of these feeds. Contract records include an $18.5 million award associated with ICM in addition to a much larger multi-year arrangement. As reported by 404 Media, it includes a database made of “tables upon tables” that allow agents to filter people based on immigration status, visa type, physical descriptors like height, color, or weight, and criminal affiliations, travel history, and location data — then export reports for field work.
Alternatively, Business Insider and Wired have reported on an ImmigrationOS program from Palantir that aims to facilitate “selection and apprehension,” offer near-real-time visualization of self-deportations, and highlight the likelihood of visa overstay. ICM and ImmigrationOS together seem to describe a full-stack targeting pipeline: find, score and act on, operational dashboards toward making arrests.
What makes Palantir essential isn’t software as much as integration. The benefit is aggregating identity data in brokers, facial biometric hits from face searches, leads from license plate readers (ALPR) and utility records, as well as field intel into a single pane of glass that investigative teams can query and share.
Scale, oversight and enduring open questions
The farther along this pile gets, the more argument there is. Civil liberties groups, including the ACLU and the Georgetown Center on Privacy & Technology, say a fusion of commercial data with biometrics and predictive analytics could lead to dragnet surveillance or disparate treatment. Government watchdogs have called for clearer regulations around data retention, algorithmic responsibility, and third-party auditing in DHS programs.
The throughline is clear: ICE’s deportation drive is becoming more of a software-powered enterprise. Procurement filings, FOIA disclosures and investigative reporting lead to a fast-improving feedback loop — from picture to identity, phone to dossier, dashboard to doorstep. Whether policy and oversight can match that acceleration will determine not just the efficiency of enforcement, but its legitimacy.