The State Department’s plan to collect the social-media accounts of nearly every visitor to the United States is the clearest sign yet that the Trump administration plans to deploy computer algorithms to fulfill its promise of “extreme vetting” for foreign visitors and temporary residents.
That’s the conclusion reached by several immigration and data-privacy experts who examined the proposal released late last week. If finalized, the rule would ask almost all of the 14.7 million people who annually apply for work or tourist visas to turn over every social-media username they’ve used in the past five years on up to 20 popular online platforms.
A pilot program begun by the Obama administration after the 2015 San Bernardino terrorist attack had previously asked a small subset of high-risk applicants to turn over their account names. But the dramatic planned expansion of that collection suggests machines will soon be examining the online activity of nearly all visa applicants, in a bid to both screen out certain individuals and continuously monitor visa holders already in the country.
“It’s inevitable that if all this information is being collected, there’s some sort of algorithmic program that’s likely to be built to crunch it,” said Rachel Levinson-Waldman, senior counsel at the Brennan Center for Justice’s Liberty and National Security Program.
The potential reach of these algorithms into the lives of the scrutinized applicants is not entirely clear, and there remains widespread confusion over which and how many government agencies plan to review the accounts once they’re collected.
“There’s no oversight of those processes, really,” said Anil Kalhan, a law professor at Drexel University. “They’re really opaque.”
But experts say recent moves by Immigration and Customs Enforcement could shed light on the federal government’s broader plans to vet visa applicants via algorithm. Last summer, ICE floated a proposal for an “Extreme Vetting Initiative” that would automatically scan the social-media accounts of visa applicants to determine not only their national-security and fraud risk, but also nebulous concepts such as their “probability of becoming a positively contributing member of society.”
ICE’s plan has since been rebranded as the “Visa Lifecycle Vetting Initiative” and may undergo more changes before its final iteration. Still, some experts say the agency’s underlying assumptions reveal a mind-set that is not only unworkable, but potentially damaging to civil liberties.
“I’m not sure how to define what a ‘positively contributing member of society’ is,” said Joshua Kroll, a computer scientist at the University of California at Berkeley. “I’m not sure how to sit down and code for that.”
Last fall, Kroll joined 53 other technical experts in writing an open letter to the Homeland Security Department warning that ICE’s algorithmic-vetting initiative would likely be inaccurate and biased. But Kroll says the ultimate effectiveness of any automated social-media scan will depend on whether ICE continues to pursue an overly ambitious predictive model or stick to more limited aims.
“There’s a big difference between something that says [someone] has a 57.3 percent chance of committing a terror attack when he comes to the United States, versus something that looks at known ISIS accounts and then follows who has engaged these accounts and flags—even automatically—this person who is applying for a visa,” he said.
ICE appears to be moving forward expeditiously on its visa-vetting program. Last month, the agency published an acquisition forecast for “over $100 million” to a government contractor for personnel needed to develop the vetting initiative. The contract is slated to take effect later this year.
In an email to National Journal, ICE spokeswoman Carissa Cutrell said that “continuous vetting” will cease once visa-holders become lawful permanent residents. She said the agency’s ultimate proposal will look “very different” from its initial idea last summer, and that any new technology would have to be approved by a slew of Homeland Security sub-agencies.
Cutrell also stressed that the focus of the vetting contract is “on labor, not technology.” The contract includes requirements for personnel to perform “analytical support” and “technical services,” among other stipulations. Cutrell did not respond to a follow-up request seeking an explanation of the distinction.
It’s not clear why ICE—traditionally a criminal-enforcement agency focusing on immigrants and visitors who are inside U.S. borders—appears to be taking an active role in developing and deploying an algorithm to screen applicants ahead of their entry or while applying for benefits.
But there are signs that agencies that work alongside the State Department to screen visa applicants, like U.S. Citizenship and Immigration Services, are also involved. A report from DHS’s inspector general in February 2017—which warned that federal pilot programs to screen applicants’ social-media accounts lacked adequate benchmarks for success—included a critique of USCIS efforts.
“Applicants are continuously screened, both at the time of their visa application and afterwards, to ensure they remain eligible to travel to the United States,” a State Department spokesperson told National Journal in an email. “As the security threat has evolved, this process has been enhanced repeatedly to improve security and more effectively identify individuals who might pose a threat to the United States.”
Don Crocetti, the former head of USCIS’s Fraud Detection and National Security Directorate, said he’s not sure whether USCIS or U.S. Customs and Border Protection plan to develop or deploy their own algorithmic scans of visa applicants’ online activity. But, he added, the State Department’s proposal to collect all social-media accounts from applicants seems to point in that direction. “I would certainly recommend it,” said Crocetti, who now heads the Immigration Integrity Group, an organization that advocates additional controls on immigration. “Because there are known terrorists—I can tell you for a fact that we only identified terrorists, or supporters of terrorism, from email accounts or social-media accounts that were in the computers of known terrorists.
“The world we’re in today is very technological, so the government has to adapt,” he said. “They have to better position themselves to collect the information that’s available.”
But concerns persist that any algorithmic examination of social-media activity—especially if it continues once visitors arrive and begin working in the United States—could infringe upon freedom of speech.
“It seems like it may result in a chilling effect,” said Janice Flynn, a U.S. visa and nationality lawyer practicing in London, who said some of her clients have already expressed trepidation over the new social-media policy. “Some people are scared that if they say something about Trump, or his administration, that they could possibly not get their visa.”
And there’s increasing worry that an expansion of social-media monitoring for visitors to the United States could be a prelude to similar activities ultimately aimed at citizens.
“You’ll see the technologies first applied to visa applicants, and then maybe in the future just applied sort of indiscriminately to citizens, so that if you’re being investigated for a crime you’ll be asked to turn in your social-media handles,” Kroll said. “And those data will be used to evaluate your risk for recidivism or future crime in that context.
“And I think that is also a concern, right?” Kroll added. “It’s sort of a rehearsal.”