Internet entrepreneur and Megaupload founder Kim Dotcom issued a stark warning in a Tucker Carlson interview released August 7, 2026: artificial intelligence systems will eventually gain access to vast government databases, creating comprehensive profiles on every individual connected to the internet. His remarks arrive at a moment when documented evidence from civil liberties organizations and privacy advocates suggests the infrastructure for exactly that scenario is already being assembled.

The Warning in Dotcom’s Own Words

Speaking with Tucker Carlson in a 34-minute interview, Dotcom drew on his background as a hacker, internet entrepreneur, and long-time target of US government prosecution to frame what he described as an inevitable convergence of AI capability and existing surveillance architecture.

“AI is very dangerous because it will get access to the US database,” Dotcom said. “So the AI will know everything about everybody.”

When Carlson pressed him on whether any practical escape exists, Dotcom’s answer was blunt: “Don’t have AI. Don’t have internet. It’s the only way.”

The comment was not merely rhetorical. Dotcom elaborated that the US government already operates massive surveillance programs capable of capturing phone calls, video calls, and other online activity. He described how collected data could be used to place citizens into categories and influence them — including during elections and campaigns designed to build public support for wars. He also claimed the US is already collecting approximately 10 petabytes of data per day, encompassing communications of all kinds.

Dotcom additionally rejected the premise that the United States holds an overwhelming lead in AI development, saying China is “not far behind” and characterizing claims to the contrary as propaganda.

The Architecture Already Being Built

Dotcom’s warnings do not exist in a vacuum. Independent documentation from civil liberties organizations reveals that the legal and technological scaffolding for AI-driven mass surveillance is actively under construction — and in some cases already operational.

According to an April 2026 analysis published by the Electronic Privacy Information Center (EPIC), negotiations between AI company Anthropic and the Department of Defense collapsed after Anthropic insisted on safeguards to prevent its AI systems from being used for mass surveillance of Americans. The government specifically sought to use Anthropic’s tools to collect and analyze “unclassified, commercial bulk data on Americans, such as geolocation and web browsing data” — a practice EPIC identifies as exploitation of the “data broker loophole.”

That loophole allows federal agencies to purchase data from commercial brokers, effectively circumventing constitutional and statutory protections that would otherwise require a warrant. EPIC notes that the USA FREEDOM Act of 2015 explicitly prohibited domestic bulk data collection under the Foreign Intelligence Surveillance Act — yet the data broker loophole functions as a direct workaround to that prohibition.

Compounding the issue is Section 702 of the Foreign Intelligence Surveillance Act, which authorizes the collection of foreign communications data but “incidentally” sweeps up Americans’ communications in the process. The government asserts the right to conduct warrantless searches of that incidentally collected data through what EPIC calls the “backdoor search loophole.” Congress has documented numerous abuses of this authority since 2015 reforms were put in place.

AI-Powered Profiling: From Immigrants to Citizens

A February 2026 report from the ACLU of Massachusetts provides concrete examples of AI-driven profiling already in use against specific populations — with civil liberties advocates warning that what begins with targeted groups rarely stays confined to them.

According to the ACLU report, the Trump administration has contracted with data aggregation companies Palantir and Babel Street to consolidate Americans’ personal data into large government databases. These companies do not merely collect data — they apply machine learning algorithms to identify patterns, flag so-called “suspicious activity,” and automate profiling and surveillance. The ACLU report specifically notes that criticism of the administration has appeared in this context as a marker of suspicion.

Palantir’s Immigration and Customs Enforcement application, called ELITE, reportedly populates a map with potential deportation targets, generates a dossier on each individual, and provides a “confidence score” estimating the person’s current address. The State Department has deployed similar technologies in its “catch and revoke” program targeting visa holders, and Secretary of State Marco Rubio has extended social media surveillance practices to visa applicants — screening individuals based on their political expression.

The ACLU’s analysis draws a historical parallel that aligns with Dotcom’s broader argument: what the government applies to non-citizens frequently becomes a template for what it later applies to citizens.

The Data Collection Ecosystem

EPIC’s analysis describes a surveillance ecosystem that extends well beyond government agencies acting alone. Data about individuals is collected through nearly every internet-connected device, with smartphones representing the most significant collection point. Commercial entities scan images online, scrape social media posts, consolidate web browsing histories, and track vehicles via automated license plate readers. That commercially harvested data then becomes available to the government through the data broker market — without any judicial oversight.

The implications of layering advanced AI systems on top of this data ecosystem are what both Dotcom and organizations like EPIC are flagging as an inflection point. AI does not merely retrieve stored information — it synthesizes, correlates, and draws inferences across disparate data sources at a scale and speed no human analyst can replicate. A government with access to 10 petabytes of daily communications data and a sufficiently capable AI system would possess, as Dotcom suggests, something approaching total informational awareness about its population.

Dotcom’s Own Experience as Context

Dotcom’s perspective carries specific personal weight. In 2012, US authorities shut down Megaupload and charged him with criminal copyright infringement, racketeering, and money laundering — triggering a New Zealand police raid on his home and a legal battle over extradition that has continued for more than a decade.

In the Carlson interview, Dotcom offered his own account of why he became a government target. He stated that his financial support for WikiLeaks and Julian Assange preceded the copyright charges, and that his donation to WikiLeaks — prompted in part by the release of the “Collateral Murder” video — made him a person of interest to US authorities. He characterized the copyright prosecution as the mechanism through which that interest was pursued.

Whether or not one accepts that framing, Dotcom’s case documents the reality that US government legal and surveillance resources have been directed at individuals operating outside American borders for activity conducted on platforms used by millions. The legal theories applied to Megaupload — holding the platform operator responsible for user uploads — remain a point of legal controversy Dotcom raised directly in the interview, noting that the same logic applied consistently would implicate companies like Google and Meta for criminal activity conducted by their users.

The Legislative Window

EPIC’s analysis identifies a near-term legislative opportunity that may already be closing. Congressional consideration of Section 702 renewal represents a moment when lawmakers could close both the data broker loophole and the backdoor search loophole simultaneously. EPIC’s assessment frames this as an urgent priority given the government’s documented interest in applying AI to surveillance data.

Without those closures, the legal architecture Dotcom warns about — AI systems with access to comprehensive government and commercially purchased databases — would have few structural barriers to realization. The technology to build such a system is advancing rapidly. The legal loopholes that would permit it are documented and currently in operation. And the expressed government interest in using AI for exactly this purpose has been reported by major news organizations following the collapse of the Anthropic-DOD negotiations.

Dotcom’s warning may be framed in stark terms, but the underlying trajectory it describes is one that civil liberties organizations, AI companies themselves, and legal scholars are actively documenting in real time.

This article draws on reporting from RT, EPIC (Electronic Privacy Information Center), ACLU of Massachusetts, and the Tucker Carlson interview with Kim Dotcom.