AI Surveillance on the Rise in US, but Tactics of Repression Not New

Dia Kayyali is a fellow at Tech Policy Press.

On March 8, ICE showed up at the apartment of Mahmoud Khalil, a permanent resident of the United States and a graduate student at Columbia University until last December, and arrested him. His arrest was part of the Trump administration's apparent goal of deporting students who joined protests in support of Gaza. President Donald Trump promised, “This is the first arrest of many to come.”

The fact that this happened days after Axios reported on a new “AI-fueled” plan from the State Department to revoke the visas of students with alleged “terrorist sympathies” has inspired a lot of alarm over how dangerous AI is in the hands of the Trump administration. I was tempted to make the same kind of proclamation myself. But I’ve come to notice that the people most shocked by what is happening are people who have had little personal experience with state repression.

As a Syrian-American, I started working on technology policy because of surveillance. I was young when 9/11 happened, but after horror for the victims, my next feeling was fear for Arabs, Muslims, and anyone who would be mistaken as one of us. My fears played out as hate crimes of all sorts increased, including against groups mistaken as Muslim, most notably Sikhs. Sikhs generally wear turbans and beards, like Balbir Singh Sodhi, who was murdered outside the gas station he ran in Arizona for the crime of “looking Muslim” to an ignorant person.

After my initial response to recent events, I stepped back and asked myself what was really different now. Is the use of AI and machine learning systems for the surveillance of immigrants and other marginalized communities new? No. Is it possible that the Trump administration will use these technologies much more aggressively and ignore the few constraints that were in place to protect people in the United States from government overreach? Yes. But it is dangerous to over-fixate on the role of AI in these events.

Indeed, reports of the AI-driven “catch and revoke” program are coming at a time when AI hype has never been higher. But while technology will make things worse for immigrants and other oppressed people in a number of ways, it is not the fuel for this fire–racism, xenophobia, and other forms of hatred are. No, these tactics of repression are not new. What is new is that they are being deployed at a shocking speed, with an unprecedented level of carelessness for the legal limits of the Constitution of the United States.

It’s important to understand the context of AI as a part of a toolkit of fascism, and how it fits into the history of political and social repression in the United States. As someone who has been working on surveillance for over two decades, I share some of that history here.

In the United States, the rule of law is weak, but racism is strong

Even as each day brings new reports of the myriad ways the Trump administration is flouting the law, it’s important to remember that the country has a long and sordid history of bending the letter of the law to its service. After all, this is the country where the Supreme Court held that “Any person descended from Africans, whether slave or free, is not a citizen of the United States” in the infamous Dred Scott decision. In more recent memory, the US relied on former Assistant US Attorney General John Yoo’s infamous torture memos to justify inhuman torture at the Abu Ghraib prison in Iraq in 2002 and 2003 and found legal underpinnings to surveil and harass the entire Arab, Muslim, and Middle Eastern population long after 9/11, alongside mass, suspicionless surveillance of people’s communications by the NSA. For that, and the history I lay out below, many Black, Indigenous, and otherwise marginalized people are deeply unsurprised by many of Trump’s actions.

Trump is simply using tools created for him by previous presidents—Republican and Democrat alike. Trump invoked the Alien Enemies Act to justify deportations for the first time since President Franklin D. Roosevelt used it as grounds for the forced incarceration of tens of thousands of Japanese-Americans in internment camps. Understanding how new repressive measures may be used and how the government may make legal justifications for anything it wants to do must reference this history.

“Catch and revoke” in historical context

Axios reported that the “catch and revoke” plan to root out students that “appear" to be Pro-Hamas will be “assisted [by] reviews of tens of thousands of student visa holders' social media accounts” and news coverage of protests.

Certainly, new technologies make it possible to do this type of surveillance faster and at a greater scale. But attempts to police thought and ferret out political dissidents are not new. Like the country’s history of racism, unjustified (and often unjustifiable) surveillance and thought policing have a long history, particularly when it comes to immigrants and people of color. Each of these phenomena has similar features: secret lists, undercover agents and informants, harassment of specific communities, and the application of new technologies. During the Red Scare, the Executive Branch, with the help of private companies and the local, state, and federal government, aggressively investigated suspected communists and ensured they lost their jobs or were imprisoned. The most famous example of this was the blacklist of suspected communists in Hollywood, which included Charlie Chaplin. A few decades later, the same FBI director oversaw the notorious COINTELPRO program, which was a systemic attempt to infiltrate, spy on, and disrupt activists in the name of “national security.” This program attempted to incite violence between Black Panthers and gangs and tried to convince Dr. Martin Luther King, Jr. to kill himself.

Many of Trump’s tactics now are particularly reminiscent of the post-9/11 shift that led to the creation of the Department of Homeland Security and the beginning of the “War on Terror.” FBI Joint Terrorism Task Forces emerged in major cities around the country, where, according to a 2010 Justice Department report, they “improperly spied on domestic advocacy organizations based on ’speculative, after-the-fact rationalizations.” The FBI used infiltrators, informants, and agent provocateurs in Arab, Muslim, and South Asian communities to entrap people into “terrorist plots”—often plots initially suggested by the informant themself. The DHS also established “fusion centers”—intelligence information clearinghouses for local, state, and federal law enforcement, including social media information.

The post-9/11 period saw the emergence of the “no-fly list” that relies in part on algorithmic flagging and often ensnares innocent individuals, prohibiting them from flying or making them subject to detention or additional screening at airports. (I received plenty of extra security screening, like many other Arab-Americans.) This kind of list, opaque and riddled with inaccuracies, could be responsible for the recent and inexplicable ICE detentions of Europeans and Canadians, including Jasmine Mooney and Jessica Brösche. These failures provide insight into the low quality of tools the government is willing to deploy- or perhaps simply the high number of false positives the government is willing to force people to endure.

Social media monitoring plays a role in all of these activities. A 2022 report from the Brennan Center details the US government's history of social media surveillance and why it raises “a host of civil rights and civil liberties concerns.” Despite these issues, social media surveillance has been used by law enforcement to track, harass, and prosecute members of every major social movement for more than a decade, from the “Occupy” movement in 2011 to Black Lives Matter protests following the murder of Trayvon Martin to the George Floyd protests in the summer of 2020 and pro-Palestine protests last year. As early as 2012, a CBS report noted that “nine in 10 law enforcement agencies say they monitor social media (and use) what they find to make cases against demonstrators.”

The DHS has been using social media surveillance since at least 2010, but under Democratic Presidents Obama and later Biden—“began ramping up the use of social media for immigration vetting” in 2014, as detailed by the Brennan Center. The program reviews social media posts and uses “algorithmic tools” to find less obvious connections. Unfortunately, efforts to create basic privacy protections and to constrain government surveillance in the US have repeatedly failed. Now, these tools are in the hands of Donald Trump and Stephen Miller.

Tech firms have always been complicit

Tech companies have a sordid history when it comes to upholding human rights. As detailed by historian Edwin Black in the book IBM and the Holocaust, IBM knowingly manufactured and maintained the technology that allowed Nazis to track Jewish people and other populations. A 2002 court case In re South African Apartheid sought to hold IBM responsible for updating and knowingly supplying that technology to enable apartheid in South Africa.

The leaders of many of the biggest US tech firms have already pledged their allegiance to Trump. There’s no reason to think that Meta, Google, Amazon, and other companies will not gladly preserve and supply reams of data for the government to use to illegally target citizens and non-citizens who dissent. What’s more, all of these companies, as well as firms such as Palantir, are racing to provide AI solutions to government agencies. They will each help build systems of oppression, one API at a time.

What to do now

This shouldn’t even need to be said, but it does—this phenomenon will not stop with Mahmoud Khalil, Badar Khan Suri, or any of the future names we will no doubt learn in the days ahead. Anyone who is not the key demographic of President Trump and Elon Musk is at risk as the campaign to target people for their political beliefs expands. With its social media surveillance tools and AI systems, the government will be in a position to find artifacts to justify whatever it wants to do—to immigrants, to political activists, to trans people. In fact, it’s more likely that people who are very secure in their safety may have posted things on social media that will put them at risk if this administration continues to indulge its fascist appetites. And at this point, deleting your posts isn’t going to save you. But that doesn’t mean we should be quiet. Quite the contrary—people in the US should be out in the streets protesting right now, especially everyone with a US Passport.

It’s also a good time to get creative. When the use of facial recognition ramped up in the US, artist and technologist, Adam Harvey created anti-facial recognition makeup. His project, CV Dazzle, may be outdated, but it can still serve as an inspiration. Technologists, in particular, should look for ways to fight back. There are many ways to feed information to the government that are completely legal that may nevertheless reduce the quality of their data, from meaningful tactics like sharing misleading information on your social media posts to creative ones like wearing high-quality 3D printed masks of Elon Musk’s face in public. There are also dozens of tried and true mutual aid tactics that folks aren’t utilizing fully yet—every major city should have a legal hotline for everyone who might be targeted by the Trump administration, in particular immigrants and trans people.

If nothing else, resistance sends a message that intimidation tactics won’t work and to those targeted that they are not alone. Finally, it sends a message to the entire world that the US is not okay with the attack on its democracy. This last point is important—the whole world is watching.