AI is making it way too easy for the government to spy on you, and some lawmakers are worried
Government surveillance has always been a divisive topic, but the conversation just entered a troubling new chapter with the rise of Artificial Intelligence. What used to require entire teams of analysts, hours of work, and bureaucratic processes can now be done in seconds by algorithms capable of cross-referencing millions of data points simultaneously. And that is exactly the problem stirring things up in the American Congress. 😬
The longstanding fight to put limits on the government’s power to sift through phone calls, emails, and text messages of American citizens without a warrant has taken on unprecedented urgency on Capitol Hill. The core concern is that AI will supercharge the state’s surveillance capabilities, enabling faster, more invasive analysis at a scale no human could ever keep up with.
At the center of this discussion is Section 702 of the Foreign Intelligence Surveillance Act, known as FISA. It is a provision that allows the U.S. government to monitor communications of foreign nationals, but in practice, it also sweeps up messages, emails, and phone calls from American citizens — all without the need for a warrant.
The law needs to be renewed periodically, and this time around, the renewal has become a minefield. Lawmakers from both parties are increasingly worried about what happens when you unleash AI on these massive databases, which also include commercial data purchased from private brokers.
The question nobody wants to answer out loud, but everyone is asking behind closed doors in Washington, is simple and terrifying: how far can the government go to know everything about you? 🔍
What is Section 702 and why does it matter so much right now
Section 702 was created in 2008 as an update to FISA, and the central argument was to protect U.S. national security by monitoring communications of foreign targets located outside the country. In theory, that sounds reasonable. In practice, what actually happens is quite different. The internet does not respect borders, and any communication between an American and someone abroad automatically falls within the scope of these mass collection programs. That means billions of data points from ordinary citizens end up in the databases of American intelligence agencies, even though these people have done absolutely nothing wrong or even know they are being monitored.
Republican Representative Thomas Massie of Kentucky summed up the concern well during a press conference where he announced a new bill to close loopholes in data collection. According to him, imagine what happens when, instead of running a search with a person’s name, you turn AI loose on these databases. There is virtually nothing the government cannot know about you.
The problem gained an extra layer of complexity when it became clear that the FBI, the NSA, and other agencies had been conducting so-called backdoor searches — searches through data collected under Section 702 using names and identifiers of Americans. Even though the law says the target is a foreign national, the back door has always been open to dig through information on people inside the country itself.
Democratic Senator Ron Wyden of Oregon, one of the co-authors of the Government Surveillance Reform Act, was pretty blunt about this in a speech on the Senate floor. He reminded everyone that for years there have been shocking abuses involving Section 702. According to Wyden, government officials used the data to find Black Lives Matter protesters, political campaign donors, elected officials, and even a state judge who had complained about police abuse.
To understand the scale of what is at stake, consider that according to reports from the Office of the Director of National Intelligence, Section 702 is considered one of the most valuable tools in the American intelligence arsenal. The CIA even produced materials highlighting how the law helped prevent a terrorist attack at a Taylor Swift concert in Austria. But value for the agencies does not necessarily mean safety for ordinary citizens. And it is precisely that tension — between operational efficiency and individual privacy — that sits at the heart of the entire current debate in Congress. 🏛️
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Artificial Intelligence and commercial data: the combination that has everyone worried
If Section 702 already raised serious questions before, the arrival of Artificial Intelligence in this equation changed the game dramatically. AI tools can process absurd volumes of information in real time, identify behavioral patterns, cross-reference data from completely different sources, and generate detailed profiles of individuals with a precision that would be impossible for any human analyst. When you apply that capability to an already massive database like the one collected under Section 702, the result is a surveillance system with virtually unlimited observational power.
Brendan Steinhauser, CEO of the nonprofit Alliance for Secure AI and a prominent conservative voice in the technology debate, described the scenario pretty clearly. According to him, the technology basically enables a panopticon. You can simply let AI find patterns, aggregate data, and allow the government to build an enormous surveillance state that threatens everyone’s civil liberties.
But the story does not stop there. What really set off alarm bells in Congress was the confirmation that intelligence agencies are also buying commercial data from private brokers, known as data brokers. This data includes:
- Location history collected by mobile apps
- Internet browsing habits
- Purchase history and transaction patterns
- Social media interactions
- Public records and known associations
- Travel history and movement patterns
The directors of both the NSA and the FBI have publicly acknowledged that they buy data on Americans from these brokers to use in investigations. The combination of data collected under Section 702 and this commercial data creates an extremely detailed portrait of anyone’s life, and AI serves as the engine that ties it all together seamlessly and instantaneously.
Jason Pye, vice president of the Due Process Institute, a bipartisan organization that advocates for fairness in the legal system, explained that Section 702 is so broad that it ends up collecting information on Americans incidentally. From there, the FBI can search for data on a person, on an American, without a warrant. That, according to him, is the problem they are trying to solve.
The point that bothers lawmakers the most is that this purchase of commercial data happens in a kind of legal gray zone. When the government buys information that a private company collected from you, it technically is not conducting a direct search, which means many of the constitutional protections of the Fourth Amendment simply do not apply. AI enters as the great enabler of this process, making possible what was previously logistically unfeasible. 🤖
The rebellion in Congress and the fight over renewal
The renewal of Section 702 has turned into one of the most heated and unusual debates in recent American politics, creating unexpected alliances. Libertarian members of the Republican Party and progressives from the Democratic Party find themselves on the same side, pushing for broader reforms and clearer restrictions on the use of AI and commercial data in intelligence operations.
In March, Republican Representative Warren Davidson of Ohio, along with co-sponsors in both the House and Senate, introduced a comprehensive FISA reform bill. The proposal reflects a bipartisan coalition that has been forming in recent weeks specifically to address concerns about AI’s ability to sift through mountains of data obtained through Section 702.
Section 702 was scheduled to expire on a recent Monday, but lawmakers agreed to a 10-day extension to buy more time for debate on new protections and safeguards. The White House pressured congressional Republicans to approve an extension without any changes. In a statement, a White House spokesperson told NBC News that the administration continues to have positive conversations and remains open to proposals that Congress can reach consensus on to reauthorize FISA.
What happened next was pretty dramatic. On the Friday before, around midnight, a group of 20 House Republicans, many of them members of the conservative House Freedom Caucus, rebelled against their own party’s leadership. House Speaker Mike Johnson had called for a vote on a five-year extension of Section 702, which was quickly defeated. A final vote at 2:07 a.m. to reauthorize the legislation for 18 months also failed, forcing Johnson to accept the 10-day extension while members tried to draft a new version.
On the following Thursday, Johnson introduced a new version of the surveillance law that would extend Section 702 for three years. Although the new text added some safeguards, it did not include a warrant requirement for searches involving data on Americans — something several Republicans had been demanding.
Senator Wyden did not hide his frustration with the outcome. In his view, the latest FISA bill from the House is essentially a blank check for President Donald Trump and FBI Director Kash Patel to spy on Americans without a warrant. He urged people not to be fooled by cosmetic reforms.
Even Democrats who previously supported the law now want changes
A notable detail in this story is that even some Democrats who voted in favor of Section 702 in 2024 are now refusing to reauthorize the law without significant amendments.
Democratic Representative Jamie Raskin of Maryland was emphatic at a recent hearing. According to him, FISA must be reformed to protect privacy and civil liberties and to ensure that Section 702 is not used to illegally spy on Americans. Raskin pointed out that the Trump administration has dismantled existing oversight mechanisms, such as the Privacy and Civil Liberties Oversight Board, which makes the need for stronger safeguards even more urgent.
In a statement that neatly captured the sense of distrust, Raskin said that the times have changed since 2024, that the watchdogs have disappeared, and that reforms now depend on Trump administration officials obeying the law — which, according to him, is oxymoronic if not simply moronic.
Raskin also made a point of noting that many of the surveillance activities permitted under Section 702 will continue until March 2027, thanks to a legislative clause that extends the authority for months if Congress cannot agree on a long-term reauthorization.
What AI companies are saying about all of this
At the end of March, Senator Wyden sent letters to the leading AI companies in the United States asking whether they would allow the government to use their technologies to surveil American citizens, including through mass commercial data collection or intelligence data that might inadvertently include information on Americans.
Wyden’s office reported that only Anthropic and Google responded. OpenAI and xAI did not send a reply.
The responses, shared exclusively with NBC News, acknowledge the lawmaker’s concerns but largely avoid giving details about how the companies allow government users to analyze foreign intelligence data.
Anne Wall, head of federal government affairs for Google in the U.S., wrote that the company recognizes that complex challenges can arise from the intersection of rapidly advancing AI and government operations, and that its teams maintain deep respect for the privacy and civil liberties of individuals.
Anthropic’s response, signed by Brian Peters, head of government affairs in North America, was more detailed and revealing. Peters said the company is committed to protecting civil liberties and that its usage policy prohibits unauthorized surveillance or tracking of individuals, as well as the analysis of domestic mass collection.
However, referring to Wyden’s concerns about Section 702, Peters revealed that Anthropic has made an exception for a small number of national security clients, allowing the use of its models for foreign intelligence analysis in compliance with applicable law. Peters said Anthropic’s AI systems can be used to analyze this foreign intelligence information, even if it includes information on American citizens collected incidentally.
It is worth noting that Anthropic, the developer of the popular Claude family of AI models, publicly positioned itself earlier this year expressing concerns about how the Pentagon would use its systems, particularly regarding the use of AI for domestic mass surveillance. The company’s CEO, Dario Amodei, wrote in a statement at the end of February that the company supports the use of AI for lawful foreign intelligence and counterintelligence missions, but that using these systems for domestic mass surveillance is incompatible with democratic values and that AI-driven mass surveillance poses serious and unprecedented risks to fundamental freedoms.
The congressional debate and the future of digital privacy
Among the proposals circulating in Congress, some stand out for the depth of the changes they suggest. One of them would require that any search of data collected under Section 702 involving identifiers of Americans needs prior judicial authorization — which would be a significant reversal in how agencies currently operate. Another proposal seeks to prohibit or heavily regulate the purchase of commercial data by government agencies, closing that back door that has drawn so much criticism. There are also initiatives aimed specifically at the use of Artificial Intelligence in these processes, calling for independent audits and greater transparency about how algorithms are being used to analyze citizen information.
Privacy advocates, such as the ACLU, have long sought to require warrants for searches of American data that ended up in databases fed by Section 702 and organized by data brokers. At the same time, many national security advocates and intelligence community experts argue that such restrictions would harm law enforcement efforts and pose severe risks to national security.
Pye, from the Due Process Institute, wrapped up his analysis with a warning that applies to anyone, regardless of where they fall on the political spectrum. According to him, some of these AI systems, with the data available to them, can essentially track where you come from, where you go, where you work, how much you make, who you know, your political affiliations, your Facebook pages, and your Twitter accounts.
And he added that he finds this genuinely concerning, particularly in this very heated, very polarized, and hyperpartisan political atmosphere.
What becomes clear in this scenario is that the discussion about digital privacy has entered an entirely new phase. It is no longer just about whether the government can listen to your phone calls, but about understanding how AI systems can build a complete profile of your life — your habits, your preferences, and even your political views — from fragments of data you never imagined were being collected.
The debate over Section 702 is, in reality, a mirror of the much larger debate the entire world will need to face in the coming years: how to balance security, technology, and individual freedom in an era when the capacity for digital surveillance has never been so powerful. And the answer Americans find to that question will echo far beyond the borders of the United States. 🌐
