As Congress plans to reauthorize a major government surveillance authority, privacy advocates say that failing to enact reforms would allow the federal government to spy on Americans at an unprecedented scale with artificial intelligence.
FISA Section 702, which lawmakers must reauthorize by March 2027, allows federal intelligence agencies to conduct warrantless electronic surveillance on foreign nationals of suspicion, whether they are abroad or traveling in the U.S.
But the electronic data of ordinary American citizens – including emails, text messages, and phone calls – are routinely swept up as well. Intelligence agencies can store that data for up to five years, while agents will routinely search through that data without obtaining a warrant.
With the rapid advancement of AI technology, however, Americans’ privacy is even more vulnerable than when Section 702 was enacted in 2008. Federal agents can simply feed the data into an AI model and effectively create dossiers on innocent U.S. citizens.
“The concern is that artificial intelligence AI could be used in these databases to effectively target and collect information in ways that we hadn’t seen before,” Don Bell, policy counsel at The Constitution Project at the Project On Government Oversight (POGO), told The Center Square.
“What we’re seeing now is a truly unprecedented merger of increasing technological capabilities and a federal government that is willing to use these tools to target not just individuals but broad sections of American society.”
Though the U.S. Intelligence Community and the Department of Justice have previously claimed they aren’t using AI in FISA databases, they have failed to answer lawmakers’ requests for proof.
FISA Section 702 currently has no prohibition on the federal government effectively outsourcing its surveillance to private companies via the “data broker loophole.”
“Data brokers collect performance management information, and they basically package that up and sell thousands of data points that they claim to have on people. And the federal government, rather than getting a warrant for a lot of that information, just circumvents the constitution by buying it from these massive data sets,” Bell said.
“But you would need a lot of money and people and time to sort through that and build the profiles necessary for effective surveillance. But with artificial intelligence, that makes it more efficient, cheaper and more doable.”
Bell, along with a number of organizations, proved just how doable the process is. With a basic AI model that they built in two weeks and a couple of data broker subscriptions, they were able to create highly detailed profiles of ordinary citizens in seconds.
“So it looks basically like typing into a search engine bar, ‘find X person at Planned Parenthood’ – but it could have been X person at a gun range, pick your place – and within seconds, this AI agent was able to pull all kinds of information on this individual,” Bell said.
“Not just their name and date of birth, but where they lived, where they worked, who their children were, financial information, things that obviously could kill constitutionally protected speech and really create a dossier on someone,” he continued.
“They were also able to basically draw a circle around the Planned Parenthood clinic – but again, it could be anyplace – and they were able to create a map showing all of the people in the community who were traveling to this place, how long they were there, all of that information.”
The issue, Bell said, is obvious: AI’s ability to collect, organize, and streamline information almost immediately and at such a monumental scale could permanently kill privacy rights if Congress fails to intervene.
“If the government is purchasing our data and our location data and they can use artificial intelligence to create these kinds of profiles, then there is no privacy,” Bell said. “They can reach into our lives in the most invasive way possible, and they can do it at scale. Entire communities can be surveilled, entire groups of people.”
In short, Congress has failed to keep up with changes in technology that are imposing new threats to privacy rights, Bell argued, enabling a “dangerously escalating pattern of surveillance.”
“That’s why it’s so important for Congress to take a stand now to close the data loophole, and also really think about how the changes in surveillance technology really are an existential threat to the Constitution and our privacy,” Bell said.
So far, Congress is still locked in a debate over whether to reform FISA at all. Nearly all Democrats and 19 Republicans in the U.S. House tanked a clean extension of Section 702, over many privacy concerns.
The program can operate without Congressional authorization until March 2027. That’s because FISA Court approves year-long certifications for Section 702, which remain in effect even if the underlying authority has expired.