Could the US State Department use AI to revoke student visas?

Some observers say use of the technology could have disproportionate effect on protests

There are concerns the US State Department's apparent plan to use AI to revoke visas will disproportionately affect students taking part in demonstrations against the Israel-Gaza war. Getty

Concern is mounting in the US amid a report that suggests the US State Department will be using artificial intelligence to potentially revoke the visas of international students accused of supporting Hamas.

The story, first reported by Axios, describes a “catch and revoke” AI effort that cancels visas of foreign citizens who support designated terror groups.

Some have viewed the effort as a way to target participants in the wave of recent protests related to the Israel-Gaza war on university and college campuses.

Some say the potential use of AI to revoke certain visas in the US could infringe upon due process legal rights. REUTERS

Since the beginning of the Israel-Gaza war in 2023, hundreds of demonstrations have taken place on US campuses, involving thousands of student protesters

In a post on X, Secretary of State Marco Rubio did not directly refer to AI, but did touch on the topic of international students and national security.

“Those who support designated terrorist organisations, including Hamas, threaten our national security,” he wrote. “The United States has zero tolerance for foreign visitors who support terrorists. Violators of US law – including international students – face visa denial or revocation, and deportation.”

The State Department told The National it “is committed to protecting our nation and its citizens by upholding the highest standards of national security and public safety through our visa process”.

“The Department has broad authority to revoke visas under Section 221(i) of the Immigration and Nationality Act (INA). We exercise that authority when information comes to light indicating that a visa holder may be no longer be eligible for a visa,” it said.

Compounding matters, President Donald Trump also recently signed an executive order that would revoke the visas of foreign students and others who took part in pro-Palestinian protests, although the executive order lacked teeth and was somewhat ambiguous.

Speculation about the potential use of AI to revoke visas also comes amid a recent request from the Trump administration to have the public provide opinions about artificial intelligence for a forthcoming AI Action Plan that hopes to clearly define US policy on the topic.

While details are few and far between about how AI might be implemented by the US State Department in this particular scenario, such a move would touch a nerve among those who worry about the unchecked and glitchy aspects of AI.

It also raises many questions about whether or not due process, and more specifically, the right to fair and consistent legal procedures is being infringed on. The topic of AI potentially affecting legal processes has grown in recent years, as the technology has advanced.

“Despite the best efforts of programmers and software engineers, an algorithm may be trained on incomplete, biased, or flawed data, or there may be data black holes,” wrote Christine Chambers Goodman, a professor of law at Pepperdine University in Los Angeles county, in a 2022 article.

“One of the additional issues brought on by algorithmic decision-making relates to scale. One algorithmic error could simply be one independent error in one case, but it could also be one error in a line of code creating hundreds of thousands of erroneous decisions.”

Craig Smith, a partner at the Boston-based Lando & Anastasi law firm, echoed those sentiments, telling The National that the State Department's potential use of AI raises several concerns.

"In addition to the free speech issues, the use of artificial intelligence tools without transparency is concerning. AI tools are only as good as the models they are based on and how they are trained," he said, referring to the potential for bias with AI.

"AI tools are effective at summarizing known information, but interpreting the meaning of that information is more difficult. Social media posts and reactions to them can be very nuanced, which requires a sophisticated understanding of the issues and the context in which the posts were made," said Mr Smith, who specialises in technology law.

He also noted that despite tremendous technological developments in recent years, AI tools and chatbots have been widely known to have hallucinations in response to various user prompts.

"An analysis of this information by an AI tool would need to be subject to quality control and verification," he said.

"These errors may be inevitable in certain AI tools, and therefore verification of any analysis or conclusions is critical."