US Immigration and Customs Enforcement (ICE) has dropped a controversial plan to monitor the social media activity of visa applicants and holders to gauge whether they would be “positively contributing member[s] of society.”
According to reports, ICE has dropped the requirement to use machine learning or other automated predictive decision-making technologies in its proposals to contractors vying for government business. Instead, ICE will emphasize human oversight of the vetting process.
Last November, Human Rights Watch joined a coalition of more than 100 nongovernmental organizations and technology experts to challenge ICE’s plan to use surveillance and predictive analytics to monitor and screen visa applicants seeking entry and visa holders currently in the US.
The Extreme Vetting Initiative was ICE’s effort to monitor the social media and other digital information on all visa applicants and holders. The criteria for screening were culled directly from President Trump’s January 2017 “Muslim ban” Executive Order. They include determining whether a visa applicant would be a “positively contributing member[s] of society” and “make contributions to the national interest” – and if they intended to commit a crime or terrorist act.
Critics of the plan argued these traits are impossible to measure and flag in an automated system, which would be both ineffective and biased. The system would also risk hiding politicized, discriminatory decisions behind a veneer of scientific objectivity. Knowledge that they might be monitored could have a chilling effect on visa holders and applicants who may censor themselves online to avoid attracting official scrutiny.
In place of the extreme vetting system, ICE is funding training of human visa screeners. But, ICE will continue to use the potentially discriminatory criteria culled from the Muslim ban executive order to manually screen applicants. While ICE’s move away from automated flagging is a positive one, the fundamentals of “extreme vetting” remain.