In June 2018, more than 400,000 people protested the Trump administration’s policy of separating families at the border. The following month saw a host of demonstrations in New York City on issues including racism and xenophobia, the abolition of Immigration and Customs Enforcement (ICE), and the National Rifle Association.
Given the ease of connecting online, it is unsurprising that many of these events got an organizing boost on social media platforms like Facebook or Twitter. A recent spate of articles did bring a surprise, however: the Department of Homeland Security (DHS) has been watching online too. Congress should demand that DHS detail the full extent of social media use and commit to ensuring that the programs are effective, non-discriminatory, and protective of privacy.
Last month, for instance, it was revealed that a Virginia-based intelligence firm used Facebook data to compile details about more than 600 protests against family separation. The firm sent its spreadsheet to the Department of Homeland Security, where the data was disseminated internally and evidently shared with the FBI and national fusion centers; these centers, which facilitate data sharing among federal, state, local, and tribal law enforcement, as well as the private sector, have been heavily criticized for violating Americans’ privacy and civil liberties while providing little of value.
In the meantime, Homeland Security Investigations — an arm of ICE created to combat criminal organizations, not collect information about lawful protests — assembled and shared a spreadsheet of the New York City demonstrations, labeling them with the tag “Anti-Trump Protests.” And as Central American caravans slowly traveled north, DHS’s Customs and Border Protection (CBP) drew on Facebook data to create dossiers on lawyers, journalists, and advocates — many of them U.S. citizens — providing services and documenting the situation on the southern border.
As shocking as these revelations are, DHS’s social media ambitions are both broader and opaque. A recent report I co-wrote for the Brennan Center for Justice, based on a review of more than 150 government documents, examines how social media is used by four DHS agencies — ICE, CBP, TSA, and the U.S. Customs and Immigration Service (USCIS) — and describes the deficiencies and risks of these programs.
First, DHS now uses social media in nearly every aspect of its immigration operations. Participants in the Visa Waiver Program, for instance — largely travelers from Western Europe — have been asked since late 2016 to voluntarily provide their social media handles. The Department of State recently won approval to demand the same of all visa applicants, nearly 15 million people per year; this data will be vetted against DHS holdings. While information from social media may not be the sole basis for denial, it could easily be combined with other factors to justify exclusion, a process that is likely to have a disproportionate impact on Muslim travelers and those coming from Latin America.
Travelers may have their social media data examined at the U.S. border as well, via warrantless searches of electronic devices undertaken by CBP and ICE. Between 2015 and 2017, the number of device searches carried out by CBP jumped more than threefold; one report suggests that about 20 percent are conducted on American travelers. (ICE does not reveal its figures.) CBP recently issued more stringent rules, though it remains to be seen how closely it will follow them; a December 2018 inspector general report concluded that the agency had failed to follow its prior procedures.
ICE operates under a decade-old policy allowing its agents to “search, detain, seize, retain, and share” electronic devices and any information on them — including social media — without individualized suspicion. Remarkably, ICE justifies this authority by pointing to centuries-old statutes, equating electronic devices with “merchandise” that customs inspectors were authorized to review under a 1790 Act passed by the First Congress. This approach puts the agency out of step with the Supreme Court, which recently recognized that treating a search of a cell phone as identical to a search of a wallet or purse “is like saying a ride on horseback is materially indistinguishable from a flight to the moon.”
The breadth of DHS’s social media monitoring begs the question: Is it effective? It is notable that a 2016 DHS brief reported that in three of four refugee vetting programs, the social media accounts “did not yield clear, articulable links to national security concerns,” even where a national security concern did exist. And a February 2017 Inspector General audit of seven social media pilot programs concluded that DHS had failed to establish any mechanisms to measure their effectiveness.
Indeed, content on social media can be difficult to decode under the best of circumstances. Natural language processing tools, used for some automated analysis, fail to accurately interpret 20-30 percent of the text they analyze, a gap that is compounded when it comes to unfamiliar languages or cultural contexts. Even human reviewers can fail to understand their own language if it’s filled with slang.
We now know far more about the scope of DHS’s efforts to collect and use social media, but there is much that remains obscured. Without robust, ongoing oversight, neither the public nor lawmakers can be confident that these programs are serving our national interest.