Skip Navigation
  • Home
  • Our Work
  • Research & Reports
  • Social Media Surveillance by Homeland Security Investigations: A Threat to Immigrant Communities and Free Expression

Social Media Surveillance by Homeland Security Investigations: A Threat to Immigrant Communities and Free Expression

This resource looks at ICE Homeland Security Investigations, the backbone of the White House’s immigration enforcement apparatus, and its dangerous exploitation of social media information.

Published: November 15, 2019

Immig­ra­tion and Customs Enforce­ment (ICE) has become the subject of national ire over its role in the Trump admin­is­tra­tion’s harass­ment, intim­id­a­tion, and deport­a­tion of immig­rants and immig­rant famil­ies. While most atten­tion has focused on ICE’s Enforce­ment and Removal Oper­a­tions (ERO), which is respons­ible for deten­tion and deport­a­tion, far less atten­tion has been paid to the agency’s Home­land Secur­ity Invest­ig­a­tions (HSI). HSI, primar­ily known for invest­ig­at­ing cross-border crim­inal activ­ity, has quietly become the back­bone of the White House’s immig­ra­tion enforce­ment appar­atus. Its oper­a­tions increas­ingly focus on invest­ig­at­ing civil immig­ra­tion viol­a­tions, facil­it­at­ing deport­a­tions carried out by ERO, and conduct­ing surveil­lance of First Amend­ment-protec­ted expres­sion, often using inform­a­tion pulled from social media — data that is frequently unre­li­able and diffi­cult to inter­pret. Risks of online surveil­lance include the chilling effect on speech, intru­sions into privacy, and the like­li­hood of discrim­in­a­tion in deploy­ment and impact. And there is little proof that the use of such data by HSI — which will affect Amer­ican citizens and resid­ents as well as immig­rant popu­la­tions — will keep the coun­try safer. 

1. HSI conducts over­broad, open-ended, and unac­count­able social media surveil­lance. HSI’s scope of oper­a­tions has expan­ded far beyond enfor­cing the laws under the agency’s purview, includ­ing using social media to target vocal crit­ics of the admin­is­tra­tion and to collect inform­a­tion on First Amend­ment-protec­ted activ­it­ies.

  • Track­ing “Anti-Trump” Protests: In summer 2018, HSI used Face­book to monitor “Anti-Trump” protests in New York City. Agents main­tained and dissem­in­ated a spread­sheet track­ing the protests, the left-wing organ­iz­ing groups, their polit­ical goals, and the number of people who signed up on Face­book to attend the protests. 
  • Inter­rog­at­ing Lawyers, Journ­al­ists, and Advoc­ates at the Border: Since 2018, lawyers, journ­al­ists, and advoc­ates have been invest­ig­ated, harassed, detained, and depor­ted in retali­ation for their human­it­arian and docu­ment­ary work at the U.S.-Mexico border. As part of an inter­na­tional, inter­agency oper­a­tion involving ICE, Customs and Border Protec­tion (CBP), the Federal Bureau of Invest­ig­a­tion (FBI), and the Mexican govern­ment, HSI collec­ted inform­a­tion and created dossiers on Amer­ican and foreign advoc­ates, journ­al­ists, and lawyers, in part based on their social media activ­ity.
  • Search­ing Elec­tronic DevicesWithin 100 miles of any U.S. border and during ICE’s wide-ranging invest­ig­a­tions, HSI claims the right to access the inform­a­tion on elec­tronic devices (e.g., cell phones, laptops, tablets, thumb drives) and in social media accounts without a warrant. When HSI agents search the device of someone trying to enter into the U.S., near the border, or other­wise encountered during ICE oper­a­tions, they can use hand-held tools to instantly unlock, decrypt, and down­load the full contents of the device, includ­ing all data from the indi­vidu­al’s Face­book and Twit­ter accounts. These searches inev­it­ably pick up inform­a­tion about Amer­ican friends, family members, busi­ness asso­ci­ates, and online contacts. HSI uses the social media data extrac­ted from devices to map out social networks to fuel further invest­ig­a­tions and intel­li­gence oper­a­tions. 

2. HSI exploits social media inform­a­tion to target vulner­able popu­la­tions and those speak­ing out against the admin­is­tra­tion’s immig­ra­tion policies, often based on extremely thin evid­ence.

  • Social Media Surveil­lance for “Extreme Vetting”: HSI has plans to continu­ously monitor the moment-by-moment social media activ­ity of 10,000 foreign visit­ors flagged as “high-risk,” from the time of trav­el­ers’ visa applic­a­tions through­out their time in the U.S. HSI person­nel are expec­ted to conduct this surveil­lance manu­ally, though the program origin­ally aimed to use social media posts to algorith­mic­ally predict whether a trav­eler would be a “posit­ively contrib­ut­ing member of soci­ety” and “make contri­bu­tions to the national interest” – criteria drawn from the Pres­id­ent’s blatantly discrim­in­at­ory first Muslim ban, and likely to impact the same group of people as the ban itself. This monit­or­ing will impinge upon free speech; anyone want­ing to come to the U.S. will be reluct­ant to post on contro­ver­sial topics, as will Amer­ic­ans commu­nic­at­ing with them.
  • Target­ing Immig­rant Activ­ists: HSI tracks the social media activ­ity of immig­rant rights groups and activ­ists to target outspoken immig­rants for deport­a­tion.
    • ICE targeted the Vermont-based farm work­ers’ rights group Migrant Justice, using social media to track and monitor the group’s advocacy campaigns, locate members for arrest, and compile dossiers on members’ activ­it­ies and social circles. HSI and ERO jointly planned the arrest of two prom­in­ent Migrant Justice activ­ists whom the agency labeled “high-profile cases” despite their lack of crim­inal records. ICE and CBP arres­ted over 40 immig­rants involved with Migrant Justice between 2016 and 2018.
    • HSI used Face­book to track events organ­ized by the New York City-based immig­rant advocacy group New Sanc­tu­ary Coali­tion as part of its “Anti-Trump” protests spread­sheet in mid-2018. ICE arres­ted the two co-founders of the group, prom­in­ent activ­ists Jean Montre­vil and Ravi Ragbir, in Janu­ary 2018. Montre­vil was depor­ted that month and Ragbir remained in immig­ra­tion deten­tion for nearly three weeks until a federal court ordered his release, express­ing “grave concern” about ICE’s possible target­ing of Ragbir based on his protec­ted speech. In April 2019, the Second Circuit Court of Appeals ruled that the First Amend­ment bars ICE from target­ing activ­ists for deport­a­tion based on their polit­ical speech. 
  • Migrant Youth ArrestsSocial media is a key part of HSI’s crack­down on alleged “gang members,” a label HSI uses loosely to target and racially profile indi­vidu­als, partic­u­larly young migrants, suspec­ted only of civil immig­ra­tion viol­a­tions. Agents use social media to track and locate these indi­vidu­als for arrest. During gang raids, HSI agents claim full author­ity to arrest any indi­vidual based merely on the indi­vidu­al’s immig­ra­tion status, even if there are no concrete facts support­ing a claim of gang affil­i­ation. In raids target­ing alleged members of the Salvadoran gang MS-13, HSI officers were encour­aged to extract and save social media data from “any and all avail­able mobile devices.” Inform­a­tion from those devices, such as a list of Face­book friends, is exploited to map out social networks, identify new targets, and build cases to prove gang affil­i­ation. HSI submits memor­anda to immig­ra­tion courts to support alleg­a­tions of gang affil­i­ation that often rely on specious social media inform­a­tion, such as online contacts or cloth­ing worn in photos. These memor­anda ensure that those labeled gang affil­i­ates are denied bond in immig­ra­tion court and that all future immig­ra­tion services and applic­a­tions are denied. 

3. The auto­mated tools HSI uses to analyze social media have not been proven effect­ive or object­ive and are likely to result in discrim­in­at­ory assess­ments. Auto­mated systems have a poor track record of making complex judg­ments, rais­ing red flags about the push toward auto­ma­tion. Internal eval­u­ations of earlier DHS pilot programs magnify the concern, as they did not find a suffi­cient basis to justify scal­ing up the use of social media screen­ing to the rest of the Depart­ment due to the diffi­culty of spot­ting “indic­at­ors of fraud, public safety, or national secur­ity concern” on social media. Never­the­less, HSI has multiple contracts with data mining and analyt­ics compan­ies to facil­it­ate its auto­mated analyses.

  • Deport­a­tions based on Auto­mated, Unac­count­able “Threat Level” Assess­mentsHSI has recently ramped up its enforce­ment against trav­el­ers who have allegedly over­stayed their visas, claim­ing that over­stays pose a seri­ous national secur­ity threat, although there is little evid­ence to substan­ti­ate that asser­tion. HSI awar­ded a $37 million contract to the data mining firm Giant Oak to continu­ously monitor, aggreg­ate, and analyze social media data to provide the agency with prior­it­ized rank­ings of poten­tial over­stay leads for deport­a­tion. Giant Oak formu­lates these rank­ings based on auto­mated assess­ments of each indi­vidu­al’s “threat level” from their social media activ­ity, among other inform­a­tion. Because social media posts are notori­ously contextdepend­ent and diffi­cult to inter­pret, complic­ated assess­ments about indi­vidu­als’ threat risks will inev­it­ably rely on prox­ies that tend to reflect stereo­types and assump­tions about the groups being vetted. There is no public inform­a­tion about the valid­ity of Giant Oak’s assess­ments or whether they are discrim­in­at­ory. 
  • Mapping Social Networks for a Digital Drag­net: HSI funnels the inform­a­tion it collects into a large analyt­ical data­base, FALCON, which was developed by the data mining company Palantir, to analyze and assess massive amounts of personal data on indi­vidu­als not suspec­ted of any wrong­do­ing. Palantir’s tech­no­logy enables HSI to conduct “social network analysis” and link analysis, creat­ing a digital drag­net to identify people who may be only very tangen­tially related to crim­inal activ­ity. There is no check on the vera­city of the inform­a­tion in this system or the connec­tions it draws between indi­vidu­als, and as the volume of inform­a­tion grows, so does the risk of error.