DHS' Constant Vetting Initiative: a Muslim-Ban by Algorithm
The initiative is an effort to develop an automated system to continuously monitor visitors and immigrants to the U.S. by looking at what they say, including on Twitter, Facebook and LinkedIn.
Cross-posted from Just Security
Late last week, the Congressional Black Caucus (CBC) wrote to the Department of Homeland Security, urging it to stop ICE’s “Extreme Vetting Initiative,” saying that the program would “unfairly target the minority communities [the CBC] represent[s].” The initiative – now relabeled “Visa Lifecycle Vetting” – is an effort to develop an automated system to continuously monitor visitors and immigrants to the United States by looking at what they say, including on Twitter, Facebook and LinkedIn. Among other things, the system is meant to predict whether a traveler will be a “positively contributing member of society” and “contribute to national interests” (broad criteria drawn from Trump’s blatantly discriminatory first Muslim ban order). It is required to generate a minimum of 10,000 annual investigative leads.
The CBC’s letter is timely: after sounding out potential contractors about the technical feasibility of such a program last summer, ICE just published an acquisition forecast that indicates it intends to go forward with spending $100 million on the project. It is likely to be money wasted. As 54 leading technologists pointed out in a letter to DHS, no algorithm can make judgments about such subjective matters as who will make a positive contribution to society (which in any event seems irrelevant for people who, for example, just want to visit Disneyland). Nor can a computer predict in any reliable way if a person intends to commit a terrorist attack, which occur too rarely to allow for an accurate forecast. Instead, as is the case with other predictive algorithms, such a venture will inevitably resort to using proxies that reflect the biases of programmers, with minority communities left most impacted.
The free speech implications of such monitoring are obvious. Knowing the government might be watching, anyone wanting to come to the U.S. to visit family or for business will be reluctant to opine on controversial or otherwise politically relevant topics, as will the Americans communicating with them. An American citizen who wants her brother in Iraq to get a visa so he can visit will certainly think twice before sending out tweets criticizing U.S. policy towards that country.
The initiative appears to be part of a broader paradigm shift in how those coming to the U.S. are vetted – “continuously,” instead of at gatekeeping intervals, like during visa issuance and entry into the country. This January, DHS Secretary Kirstjen Nielsen testified to the Senate that “[w]e are … ensuring that…individuals are continuously vetted…rather than only at the time of their application.” In February, President Trump issued a Presidential Memorandum creating a “National Vetting Enterprise,” specifying that it was U.S. policy to “process biographic, biometric, and contextual information … on a recurrent basis” to identify security threats. Additionally, legislation introduced during the Senate immigration debate last month included a provision that would have authorized “continuous screening” of those “who are present, or expected to arrive [in the U.S.] within 30 days.”
This is a significant change in how the U.S. treats those who come to its shores as visitors or immigrants and requires – at a minimum – a full airing of the need for such an expansion, its feasibility, and its costs. After all, our existing visa vetting system is among the world’s most thorough and effective.