The Biden administration has begun using artificial intelligence to carry out a review of more than 55 million people who hold U.S. visas, the State Department and people familiar with the effort told the Daily Mail, a step the department described as part of a new "continuous vetting" process to identify potential immigration violations.
The department said the vetting is intended to identify reasons visas could be revoked or admissions barred, including overstays, criminal activity or links to terrorism. Officials told the Daily Mail that social media accounts, immigration records from applicants' countries of origin and other available information would be incorporated into automated checks.
Administration and former State Department officials described the effort as unprecedented in scale. A former State Department employee told the Daily Mail the public framing of a 55 million-person sweep may be intended in part to prompt voluntary departures, saying, "They don't need to scrub 55 million. They just need to say they are casting the net as extensively as possible, to encourage those who know they are ineligible, probably overstaying their visas, to self-deport before they are caught by the federal government and punished." Another current employee told the outlet, "That sounds insane. I am just happy I am not in consular affairs."
Officials described the technology as a response to staffing reductions and the logistical challenges of manually reviewing millions of records. The Daily Mail reported a roughly 20% reduction in State Department staff in recent months; a former employee said the issue is now one of capabilities rather than manpower. The department has also recently narrowed access to student visas and officials say automated tools were used in that effort.
Immigration policy experts and technology analysts cautioned that automated vetting at scale raises risks of false positives and other harms if the systems are not rigorously tested and paired with human review. Julia Gelatt, associate director of the U.S. Immigration Policy Program at the Migration Policy Institute, told the Daily Mail there is insufficient public information about how the vetting will operate and warned that government databases do not always contain complete or accurate information. "Different government databases are speaking to each other looking for matches, but there are concerns some have incomplete information — like FBI data — so if somebody has an arrest but is ultimately found innocent, that might not be recorded," Gelatt said.
Rob Enderle, president and principal analyst at the Enderle Group, said accelerated, productivity-focused AI systems can sacrifice accuracy. "The odds of this ending very poorly for many people is exceptionally high," he said, adding that platforms currently prioritize speed over quality and that error rates must fall before reliance on automation is increased.
The concerns are grounded in recent cases that authorities and advocates say illustrate problems when automated flags are not carefully reviewed. In April, Japanese BYU student Suguru Onda had his visa terminated after what his lawyer said was an error by automated systems related to minor traffic citations. In March, Turkish Tufts University student Rümeysa Öztürk was arrested after an F-1 visa was revoked and she was transferred to an Immigration and Customs Enforcement facility, a case that drew criticism from lawmakers and civil-rights groups.
A State Department official quoted by Fox News said last spring that every student visa revoked under the current administration had been tied to lawbreaking or expressions of support for terrorism. The department has told media outlets that about 6,000 student visas have been revoked since January, roughly 4,000 of which it attributed to violations of law.
Advocates and analysts said those figures underscore why transparent procedures and strong human oversight are needed. Gelatt said the publicly stated 55 million figure is misleading and likely covers many people who do not currently live in the United States, calling a full sweep "wasteful" and urging the department to explain selection criteria. Sources familiar with the process told the Daily Mail that officials plan to prioritize certain countries for scrutiny but have not publicly disclosed how priorities will be set.
The Department of Homeland Security reported that last year there were nearly 13 million lawful permanent residents and almost 4 million people in the U.S. on temporary visas, numbers that officials said are relevant to the scale of any vetting effort.
Legal and civil-rights groups have long warned that reliance on automated tools in immigration enforcement can produce unequal outcomes and irreversible consequences for people incorrectly flagged. They said automated decisions can be difficult to appeal and that incomplete records or unvetted matches could result in detained or deported migrants who pose no public-safety threat.
State Department spokespeople did not provide additional public detail about the technical safeguards, error rates or the extent of planned human review when reached through channels that disseminated the initial report. Migration experts called for clearer disclosure about data sources, accuracy testing, redress procedures and how the department will avoid political or discriminatory targeting.
The administration's move comes amid other high-profile immigration-policy shifts and heightened scrutiny of student and visitor visas. Officials said the technology is intended to strengthen national security and immigration enforcement, while critics warned that the pace of deployment and scope of data-sharing raise civil-liberties risks if automated results are acted on without thorough human oversight.
As the review proceeds, lawmakers, advocacy groups and immigration experts said they will seek additional information about how individuals will be identified, how errors will be corrected and what recourse people will have if they are wrongly flagged. The practical implementation of large-scale automated vetting, including limits on data-sharing and safeguards to prevent mistaken revocations, will likely shape subsequent debates over the use of artificial intelligence in immigration enforcement.