Photo Illustration by Elena Scotti/Splinter/GMG, photos via AP, Shutterstock

The federal government is launching a program to automatically and continuously monitor the social media posts of students, tourists, and other visitors to the United States, hunting for any “derogatory” information and tracking their whereabouts, associations, and activities.

The new program, which The Intercept first reported Monday, would give a high-tech boost to President Trump’s call for “extreme vetting” of immigrants by supposedly helping the government detect visitors who might be undergoing “radicalization. “

Advertisement

But civil liberties and immigration advocates charge that the unprecedented monitoring program could unfairly target visitors from Muslim-majority countries and help accelerate the deportations of undocumented immigrants. The government’s own contracting documents, they point out, show that the program’s social media surveillance capabilities will be used to “locate and detain” people who have merely overstayed their visas, many of whom may not necessarily pose national security risks. Since 2007, the majority of undocumented immigrants in the United States have gained entry through visa overstays, not illegal border crossings.

The program, which an official familiar with ICE’s plans—who asked not to be named as they were not authorized to speak about a contract in the procurement process—said it aims to deploy by October 2018, would fuse the now-fragmented databases containing information on banking activity, travel history, and various watch lists that ICE agents use to track people into one powerful tool. But it would also add a key new component: constant surveillance of social media posts by visitors to the U.S., starting the moment they apply for a visa and extending throughout their stay.

The program would apply to people entering the U.S. on a wide range of visas, including students, travelers, people applying for refugee or asylum status, and individuals seeking longer term green card benefits. It would automate a series of complex tasks—including surveillance of social media for “derogatory” posts— that are currently performed by agents, and it would track people more consistently and for the entirety of their visa cycle.

A spokesperson for ICE declined to comment further on the program.

The legality of the proposed system has yet to be formally established by the courts, though ICE notes in contracting documents that it does not foresee any major complications. Visa-holders consent to a degree of surveillance which is conditional for entry, but they also enjoy some protection provided by the US Constitution while in the United States under the 14th Amendment’s equal protection clause.

Advertisement

Such techniques are supposed “to further investigations or support any prosecution by ICE or U.S. attorneys in immigration or federal courts.” The program would thus bolster U.S. attorneys carrying out Attorney General Jeff Sessions’ directive to crack down on immigration violations by prosecuting a greater number of offenders under more serious charges.

The program is still in its formative stages. ICE has announced that it plans to contract with companies to implement it, describing in detail what it seeks and answering written questions. It has also held two “Industry Day events attended by over 50 companies.

Slide from an Industry Day event held by ICE about the new program.

The program hopes to make ICE’s visa review process more efficient. Currently, the visa security screening process occurs at the Visa Security Coordination Center, where the State Department sends them for initial vetting by ICE. The process generates investigative leads for ICE’s case management system, called LeadTrac. Each lead is based upon allegedly “derogatory” information which must be vetted manually by personnel. This new vetting program would streamline this process “while simultaneously making determinations via automation if the data retrieved is actionable,” according to a background document posted with others regarding the prospective services on the government contracting website FBO.gov.

“It looks like in order to do this ‘extreme vetting’ they’re expanding their mission,” Claude Arnold, a former ICE Homeland Security Investigations Special Agent, said. ICE HSI’s National Security Unit already vets applicants prior to the interview process on a limited basis via its Visa Security Units, “but they’re not everywhere, they’re in high threat locations,” Arnold said.

Supporters of the program say that the continued vetting of visitors once they have entered the U.S. could help ensure potential threats do not fall through the cracks.

“Let’s say you go in and get a tourist visa, and you enter the US, and there’s nothing derogatory on you,” John Sandweg, acting director of ICE from 2013 to 2014, said. “Now two weeks goes by and an informant shares some serious information about you about raising money for a terrorist organization. It’s possible that, though that information gets loaded into a system, now nobody is running a search against you. So the idea would be for a ping to come up when new derogatory information related to you gets entered.”

Advertisement

“Gangbangers love to have photos of themselves holding an AK-47,” Arnold said. “It doesn’t just have to be criminal offenses, I mean, people talk about where they hang out, who their friends are. Someone who maybe has become radicalized, they advertise that on their social media pages, so it’s fair game.”

(Both Sansweg and Arnold work for Frontier, a company which is in the homeland security market but not in a position to participate in the new ICE program.)

Civil liberties advocates, however, contend that the automated data-mining program could unfairly keep people out and mistakenly drag innocent people into constant monitoring.

Advertisement

Cathy O’Neil, a data scientist and author of Weapons of Math Destruction, said that ICE needs to be transparent about the assumptions that will be embedded within their algorithm.

“When a person optimizes any kind of algorithm, you will implicitly weigh false positives against false negatives,” she said. “At the end of the day, someone has to choose a ratio: How many innocent false positives are you going to keep out of the country for each false negative?”

The program also needs to have built-in audit systems to ensure the process is accurate and asks when, how, and for whom it fails, O’Neil argued. “A lot of the times when police use social media, for example, to track gangs, they implicate people in criminality simply because of their social networks,” she said. “I’d be afraid of that kind of thing happening, for example, to a Muslim individual simply because of ties to other Muslims.”

Advertisement

ICE has brushed off such concerns. In response to an industry question about likely pushback from privacy advocates, an unnamed ICE spokesperson said that the targeting of foreign nationals, rather than U.S. citizens, would neutralize any concerns. The spokesperson then pointed to the San Bernardino terrorist attacks, which involved a recent arrival who had come on a J-1 “fiance” visa. “Agents have an ever-changing environment and we cannot look and just do nothing,” the spokesperson said. “We remember the San Bernardino case where the data was not there.”

Screenshot of the ICE response to an industry question about civil liberties concerns.

Immigration advocates also charge that the programs could be used to locate and deport a wide variety of individuals suspected of low-level immigration violations, not just terrorist suspects.

Advertisement

“What we’re seeing under this administration is a rapid buildup of data infrastructure for U.S. Attorneys’ Offices to be able to do more immigration prosecutions,” Jacinta González, field director for the Arizona-based advocacy group Mijente, told Splinter. “Already under Obama more than half of their prosecutions were for immigration violations. This program is just a continuation of the ramp-up under Sessions and Trump.”

ICE’s Counterterrorism and Criminal Exploitation Unit, which supports the State Department in vetting visa applicants, already accesses automated DHS systems like Custom and Border Protection’s Automated Targeting System. To fulfill President Trump’s extreme vetting directive, ICE wants a contractor to “leverage all available datasets to create a Federated Query Vetting System.”

Numerous technology companies with expertise in big data analytics, such as IBM, Unisys, and SAS, have shown interest in the contract.

Advertisement

Kelly Jones, a federal accounts executive with Nuix USG who attended the July 19 industry day, said her company was proposing its Insight and Analytics solution to meet ICE’s requirements. “We already have a deployment within ICE, so the relationship is already well established,” she said, adding that Nuix has a meeting with ICE later this week for deeper discussion of the extreme vetting system.

In response to questions posed by vendors after the initial briefing about the data-mining aspects of this program, ICE noted that the government will provide firms with “required key terms/words/criteria” and expect firms to perform “cross‐lingual searching & translation across multiple languages.” Public social media posts, and forms of open source intelligence, are slated to be a key part of this service, supporting ICE’s year-old Visa Overstay Lifecycle pilot, which maintains persistent online surveillance of visa holders from the time of their application to their eventual departure.

The services sought by ICE specify that social media exploitation techniques should not only support established investigations but provide intelligence leads “derived only from free and publicly available sources through unattributed computers,” according to background contracting documents. Some social media companies, such as Facebook, have opted to make their application programming interfaces private in response to surveillance concerns. But in an industry Q&A document, posted online on July 13, ICE insisted that contractors must still go after this data.

Screenshot from a July 13th industry Q&A

Neema Giuliani Singh, legislative counsel with the American Civil Liberties Union Washington Legislative Office, said that DHS should refrain from expanding social media monitoring for vetting purposes, pointing to a recent DHS Inspector General report which found that automated attempts to identify travelers’ social media accounts performed worse than manual ones.

Advertisement

“Social media can be easily misunderstood because of language differences, and there are also cases where people post things critical of government, so the question becomes what ‘derogatory’ information pings the system?” Singh said. “And if it does, do you get a court hearing, or a chance to rebut determinations based on those alerts? Because it seems like right now the government gets to make decisions behind closed doors and all of a sudden we have a whole class of people powerless to fight them.”

Arnold, however, argued that what people post online is fair game. “If they’re dumb enough to be putting themselves out there....of course law enforcement is going to exploit that,” he said. “The internet and social media is modern-day dumpster-diving.”

While the program may be immigration-related and still in its early stages, its monitoring applications will likely spread in the coming years, González said. “What starts as HSI’s secret toy of the month, in a few years, becomes used on the whole population,” she said. “So this isn’t just an immigration issue, it’s an issue for everyone. In some ways, DHS is the gateway to spying on the American people in a very evasive way.”

Advertisement

For now, ICE seems confident the program will be implemented without significant legal challenge, especially given the lack of congressional rules at the moment.

“The prediction is that in the near future there will be legislation addressing what you can and can’t do,” said an unnamed ICE spokesperson in a July 18th Industry Q&A, noting the system may have agents apply for warrants. “We will continue to do it until someone says that we can’t.”