The tech industry built the machinery of American fascism

By • • 2714 words • 13 min read

Updated on

🏷️ values AI community

Fascism in the digital age requires two machines working in tandem: a propaganda apparatus to control what people believe, and a surveillance apparatus to find, track, and remove the people targeted by those narratives.

The tech industry built both.

The speed of democratic collapse we’re witnessing - faster than scholars predicted possible - is a direct result of that infrastructure already being in place, ready to be used.

What the scholars say

I’ve been reading the scholarly literature on fascism. Not because I wanted to write this post, but because I needed to understand what I was seeing. The frameworks keep pointing in the same direction.

Umberto Eco’s Ur-Fascism (1995) identified 14 characteristics that can “coagulate” into fascism. Eco’s warning was that fascism doesn’t require all 14 - it can form around any subset. The current U.S. exhibits at least 12, including: cult of tradition (“Make America Great Again”), rejection of modernism (attacks on “woke” culture), cult of action over reflection (rapid executive orders without congressional process), disagreement as treason (labeling opponents “enemies within”), fear of difference (mass deportation campaigns), appeal to frustrated middle class, obsession with plots (“deep state,” “invasion”), enemies portrayed as both strong and weak, life as permanent warfare, contempt for the weak, and newspeak (simplified slogans, “fake news”).

Levitsky and Ziblatt’s four indicators from How Democracies Die are all present: rejection of democratic rules (refused to accept 2020 results, orchestrated fake electors scheme in 7 states), denial of opponents’ legitimacy (labels opponents “enemies within”), toleration of violence (pardoned all January 6 defendants), and readiness to curtail liberties (mass deportations, press attacks, firing prosecutors).

Jason Stanley’s 10 propaganda pillars from How Fascism Works are all demonstrably present: mythic past, propaganda, anti-intellectualism, unreality, hierarchy, victimhood, law and order, sexual anxiety, heartland appeals, and dismantling welfare.

What the data shows

The scholarly frameworks are subjective by nature. But the quantitative indices tell a consistent story.

Freedom House: The U.S. score fell from 94/100 in 2011 to 83/100, placing America among the 25 countries with largest declines globally. For comparison: Canada (98), Germany (94), UK (93).

V-Dem Institute: Their 2025 report identified the U.S. as undergoing the “fastest evolving episode of autocratization the USA has been through in modern history.” Director Staffan Lindberg told US News: “At the pace at which it is happening, I would say that before the end of the summer, you no longer qualify as a democracy in the United States.”

Bright Line Watch: Expert surveys show the steepest decline on record in February 2025, with democracy ratings falling from 67 to 55 on a 100-point scale. Expert ratings collapsed on specific indicators: “Government agencies not used to punish political opponents” dropped from 68% to 7%. “Not interfering with the press” dropped from 76% to 23%.

How we got here

The pattern that made this possible is now routine.

January 6, 2021: 140+ officers injured, video of assaults, $2.7 billion in total costs, the largest prosecution in U.S. history. The current administration now claims Capitol Police “escalated tensions” and issued blanket pardons for all 1,500 defendants - including those convicted of assaulting officers.

The 2020 election: 60+ court cases found no significant fraud. Trump’s own Attorney General and cybersecurity officials confirmed the election’s integrity. “Stolen election” remains party orthodoxy.

In January 2025, 17+ inspectors general were fired via late-night emails without legally required notice. DOJ terminated prosecutors who worked on Trump cases, with Acting AG James McHenry citing their inability to carry out “the president’s agenda.” The Public Integrity Section created after Watergate was gutted.

By September 2025, the administration claimed 2.5 million had “left” (600,000 deportations plus 1.9 million “self-deportations”). Family separations resumed inside the U.S. CNN documented 100+ U.S. citizen children left stranded when parents were detained without time to arrange childcare.

When scholars like Jason Stanley talk about “unreality” as a pillar of fascism, this is exactly what they mean. The question stops being “what happened?” and becomes “whose version do you support?”

The tech industry built this

What distinguishes this moment from historical fascism is technology’s role in enabling it. This isn’t passive neglect or unfortunate side effects. This is active collaboration at every level - from platform design to direct political coordination.

I wrote recently about why the tech industry needs to leave X. The short version: Musk deliberately dismantled safety infrastructure, hate speech increased 50%, Grok generated Hitler praise and sexualized images of minors, and the platform owner actively promotes fascist movements from Germany’s AfD to the UK’s far-right. That AI is now available to federal employees.

But X is just the most visible piece. The problem runs deeper.

Palantir: Building the deportation machine

Palantir’s $30 million ImmigrationOS contract provides ICE with “near real-time visibility” on deportations, integrating passport records, Social Security files, IRS data, license plate readers, and social media. The company’s federal contracts nearly doubled in 2025 to $970.5 million.

Peter Thiel, Palantir’s co-founder, was on Trump’s 2016 transition team. His protege JD Vance is now Vice President. Stephen Miller, Trump’s chief immigration architect, holds substantial Palantir stock.

The company that builds the surveillance infrastructure profits when that infrastructure is used. The people who built the company now run the government using it. This isn’t a bug in the system. This is what the system was designed to do.

The data brokers

Palantir is the integration layer. The data itself comes from companies you’ve probably used.

Clearview AI holds a $9.2 million ICE contract providing facial recognition that searches a database of 50+ billion images scraped from social media. A purported co-founder allegedly posted about “building algorithms to ID all the illegal immigrants for the deportation squad.” The company is banned from selling to Illinois police after violating the state’s biometric privacy law and has been fined by regulators across Europe. Federal agents can use it anyway.

LexisNexis - the legal research tool used by virtually every law student and lawyer - holds a $22 million ICE contract providing access to 283 million individual dossiers containing Social Security numbers, addresses, credit history, and relatives’ information. ICE contract documents explicitly state they needed this access because “policy or legislative changes” stopped local agencies from sharing data - meaning ICE contracted with LexisNexis specifically to circumvent sanctuary city laws.

Thomson Reuters - parent company of Westlaw - holds approximately $22.8 million in ICE contracts providing their CLEAR platform and Vigilant license plate reader database. Both LexisNexis and Thomson Reuters are listed as official “data partners” of Palantir, feeding directly into its surveillance systems.

Flock Safety is the trickiest case. The company claims it has no contract with ICE. That’s technically true. But 404 Media obtained data showing over 4,000 lookups by local police done on behalf of ICE - giving federal agents “side door” access to a network of license plate readers conducting 20 billion scans per month across 5,000+ communities. The ACLU of Colorado found over 1,400 ICE-related searches in Denver’s Flock data alone since June 2024. Cities from Cambridge to Syracuse have terminated their Flock contracts over these concerns. Denver’s city council voted unanimously to end the contract - then their mayor unilaterally overrode them.

The same databases law students use to research civil rights cases are being used to track and deport people. The same license plate readers your city installed to find stolen cars are being searched by ICE.

The spyware vendors

ICE isn’t just collecting data passively. They’re actively hacking phones.

Paragon Solutions has a $2 million contract for Graphite spyware, which can access encrypted messages on any phone without the user clicking anything. The Biden administration paused this contract over human rights concerns after Graphite was used to target journalists and refugee activists in Italy. The Trump administration reactivated it in August 2025. Members of Congress warned that deploying this spyware “threatens Americans’ freedom of movement and freedom of speech.”

Cellebrite holds an $11 million contract to unlock phones and extract complete data images - everything including Signal and WhatsApp messages, photos, location history. When ICE detains someone, they can now copy everything on their phone.

The propaganda machine

Surveillance is only half the equation. The same AI tools being used to track people are being used to manufacture the narratives that justify tracking them.

The Washington Post obtained internal communications from ICE’s social media team showing how the agency creates viral arrest videos. Political appointees were ordered to produce “dramatized visuals” to “command user attention.” When one official asked what to say in videos when arrestees had no criminal history, another responded they should find something else “newsworthy” - like an “egregious immigration history.” At the White House’s request, one video was re-edited to “not feature tons of females.” A former Pentagon spokesperson told the Post this “veers into propaganda, into creating fear.” After an ICE official shared census data showing the immigrant population had dropped by 2 million, a legal associate responded: “Only 40 million to go.”

The president himself posts AI-generated deepfakes from official accounts. An NBC News review found dozens of synthetic media posts since January: Trump as pope, Trump as warrior, Trump winning the Nobel Prize. One depicted him flying a jet and dropping sewage on protesters. Another showed “Chipocalypse Now” with the caption “I love the smell of deportations in the morning.” A deepfake of Nancy Pelosi used falsified audio to depict her admitting to crimes. A deepfake showed Hakeem Jeffries in a sombrero, which Jeffries called racist.

This week, the administration crossed a new line. After civil rights attorney Nekima Levy Armstrong was arrested at a church protest in St. Paul, Secretary Noem posted a photo of the arrest showing Levy Armstrong with a calm expression. Thirty minutes later, the White House posted what appeared to be the same image - but her face had been altered to look distressed. An MSNBC analysis called it “the first widely circulated manipulated image that passes convincingly as a real photograph” from the administration. It’s not satire or an obvious meme. It’s fabricated evidence designed to make the resistance look defeated.

And it’s not just official content. Fake AI-generated ICE raid videos are going viral on Facebook. 404 Media found videos made with OpenAI’s Sora depicting workplace deportations at Walmart, complete with badges reading “POICE” and “IMMIGRATION AND CERS.” One video showing Latino workers being rounded up got 4 million views. Many commenters didn’t realize it was fake. Meta, which changed its content moderation rules after the election to have fewer restrictions on immigration content, rewards this engagement with algorithmic amplification.

The propaganda extends to recruitment. ICE is running a $100 million “wartime recruitment” campaign to hire 10,000 new agents, using geo-targeted ads aimed at UFC attendees, patriotic podcast listeners, and people who’ve shown interest in guns and tactical gear. Some ads used white nationalist slogans. A former ICE director from the Obama administration said the agency used to recruit from police departments; now they’re hiring “untrained recruits eager for combat.”

And at the center of it all: Grok. Despite the administration’s own executive order requiring AI systems to be “neutral” and “truth-seeking,” the GSA made Grok available across federal agencies. This is the same AI that generated Holocaust denial and antisemitic content, that was programmed to avoid sources critical of Musk or Trump, that civil society groups warned could become a “propaganda arm of the U.S. government.” Reuters reported that DOGE staff pushed DHS to use Grok for data analysis, and that staff were ordered to train AI to identify communications suggesting employees weren’t “loyal” to Trump’s agenda.

When White House Science Adviser Michael Kratsios was asked about Grok’s antisemitic responses in Senate testimony, he acknowledged they “obviously aren’t true-seeking and accurate” and were “the type of behavior” the administration’s own order sought to avoid. The government expanded Grok’s deployment anyway.

This is what AI-powered authoritarianism looks like: official deepfakes from the president’s account, altered photos to make protesters look defeated, an agency social media machine producing dramatized arrest videos, AI recruitment ads targeting would-be paramilitaries, fake enforcement content going viral on platforms that profit from engagement, and a government chatbot owned by a man who sits on the president’s advisory council and profits from every federal contract it signs.

Silicon Valley’s capitulation

The alignment isn’t limited to Musk and Thiel. Across the industry, tech leaders rushed to demonstrate loyalty.

Meta ended third-party fact-checking, donated $1 million to the inaugural fund, and added Trump ally Dana White to its board. Google donated $1 million, rolled back DEI programs, and paid $22 million toward Trump’s White House ballroom. Amazon donated $1 million and streamed the inauguration; Jeff Bezos killed the Washington Post’s planned Harris endorsement. OpenAI’s Sam Altman donated $1 million and declared “President Trump will lead our country into the age of AI.”

Tech CEOs sat in the second row at Trump’s inauguration - better seats than cabinet picks.

This isn’t companies staying neutral to avoid political controversy. This is companies picking a side - the side with the power to regulate them, the side whose regulatory approach they prefer, the side that will let them do what they want.

What tech workers should understand

In 2020, tech leaders posted black squares and issued statements about racial justice, about democracy, about values. Some companies pulled PAC funding from election objectors. They told us there were lines that couldn’t be crossed. Now those same executives compete for inauguration seats, donate millions to the transition, and build the surveillance systems that make mass enforcement possible.

Ask yourself: if a tech worker got killed by ICE tomorrow, what would their employer do? Stand with their employee’s family, or protect their relationship with a billion-dollar government client? The answer is obvious. These companies laid off tens of thousands of workers while posting record profits. They demanded return-to-office while executives worked from wherever they wanted. Employee welfare has never been the priority.

The people celebrating the Minneapolis killings, the ones calling a nurse a “domestic terrorist” and a mother a “violent rioter” - those are the people tech leadership is courting. Your career accomplishments, your credentials, your years of loyal service: none of it buys you protection from the system your industry built.

Silence won’t keep you safe. The companies demanding your silence will abandon you the moment it’s convenient. The infrastructure your labor built is being used against people just like you. The only question is whether you’ll say something while it still matters.

What now

Google employees walked out over Project Maven. They won. The contract was dropped.

Amazonians organized against Rekognition sales to law enforcement. Microsoft employees protested the HoloLens Army contract. GitHub workers pushed back on the ICE contract. Not every fight was won, but every fight made the cost of collaboration visible.

You have more leverage than you think. The tech labor market is still competitive. Your skills are portable. Your silence is not required.

If your company holds these contracts: Organize internally. Find the others - they’re there. Refuse to work on the projects. Document what you see. And if none of that works, leave loudly.

If your company doesn’t hold these contracts: Say something anyway. The culture of enforced silence in tech is part of how we got here. Every person who stays quiet makes it easier for leadership to pretend there’s no opposition.

Support the organizations doing this work:

The infrastructure exists because people built it. It can be dismantled the same way.