How AI Is Reshaping U.S. Immigration Enforcement | DOCUMENTED NY

By Benjamin Rubin | Mar 25, 2025
Since the detention of Mahmoud Khalil, a green card-holding pro-Palestinian activist and Columbia University graduate student, many have asked which legal residents the Trump administration will target next. Trump promised that the 30-year-old Columbia graduate and Algerian citizen of Palestinian descent would be the first of many such arrests to come. 

In early March, Axios reported that the administration was using AI to scan tens of thousands of social media accounts, targeting student visa holders believed to show “pro-Hamas” support. The endeavor, run by the State Department as well as the Departments of Justice and Homeland Security, will be used to revoke student visas, officials said.

Whether or not the “Catch and Revoke” system reported by Axios was behind ICE’s deportation efforts, DHS has been well positioned to use AI surveillance of social media in targeting “foreign threats” for years.

“The previous administrations built the runway that can be used by the Trump administration and more authoritarian regimes to crack down on the civil rights of communities in the U.S.,” wrote Citlaly Mora Hernandez, a spokesperson for a racial and immigration justice advocacy group fighting against tech surveillance, Just Futures Law.

Also Read: What Has Happened Since Arrest of Legal U.S. Resident of Palestinian Descent Mahmoud Khalil: Trump Says “First of Many”

Early in March, DHS announced that under Trump’s executive order they would be collecting the social media handles of people applying for visas and citizenship. But this was only an update to an existing policy. In 2019, visa applicants were required to submit their social media identifiers as well, a policy maintained through the Biden administration, affecting the millions who apply for visas each year. But that rule has been challenged in an ongoing court case, primarily on the grounds of protecting the First Amendment due to self-censoring by those surveilled.

A 2019 report by the Brennan Center for Justice found that DHS had been broadly monitoring social media of noncitizens in the name of national security, in many cases that of Muslim populations. By then, DHS had been using automated systems to monitor social media, but the report noted concerns — like accuracy and bias — around using such algorithms to assess actual threats. 

Studies show that it’s very difficult to determine if a person is an actual national security threat based on their online presence, said Saira Hussain, a senior staff attorney at the Electronic Frontier Foundation, a digital privacy advocacy group. 

“But when you add AI into the mix, or some sort of automated tool, I think that makes it even more likely that you’re going to end up making mistakes,” Hussain said. 

Similarly, a 2017 inspector general report assessing early pilot programs surveilling social media for ICE’s visa issuance found that the initiatives had a shortfall of “criteria for measuring performance to ensure they meet their objectives.” 

Spokespersons from the State Department and DHS did not respond to a request for comment on the record. The Department of Justice also did not respond to a request for comment.

DHS has capitalized on the growth in federal spending on AI, which has reached billions of dollars, according to the Brookings Institution. A DHS document from 2021 outlined a strategy for “the responsible integration of artificial intelligence into the Department’s activities,” for example.  

Also Read: Haitian Community Protests in Brooklyn to Defend Temporary Protected Status

According to a database maintained by DHS, the “Artificial Intelligence Use Case Inventory Library,” as of this year, the agency lists close to 160 active efforts using AI. ICE is listed for about 20 of the active cases, while CBP is listed for 60. 

Federal agencies associated with the “Catch and Revoke” program did not respond to a request for information on the specific contracting behind the system.  

Nonetheless, at least one program used by DHS offers a window into how the agency plans to use AI to surveil social media.

“Babel,” for instance, is an active case listed in the inventory, described as “a commercially procured tool that helps CBP compile social media and open-source information on travelers who may be subject to further screening for potential violation of laws that CBP is authorized to enforce or administer.”

Babel is further described in the inventory as an AI tool for text and image detection on social media, used to screen travelers and help analysts “identify potential threats to the United States.” 

Babel Street is a government-contracted data analytics company, with a website describing services that fits the descriptions in the inventory under “Babel.”

In 2023, reporting by Vice found that under CBP, Babel Street’s AI surveillance could in some cases be used to link social media posts of noncitizen populations to their location data and Social Security numbers. 

Babel Street was identified in the Brennan Center’s 2019 report, with contracts under CBP as far back as 2015. Across the decade since, DHS has awarded Babel Street over $21 million in contracts, according to USAspending.gov. 

This article was originally published on Documented NY on March 25, 2025

The post How AI Is Reshaping U.S. Immigration Enforcement | DOCUMENTED NY appeared first on The Haitian Times.