Invisible Authoritarianism: How AI Is Powering a New Surveillance State

From border drones to protestor threat scores, Executive Order 14179 sets the stage for AI-enabled repression

The New Architecture of Control

It doesn’t arrive with sirens. There are no tanks, no emergency broadcasts. Instead, it comes through memos, data pipelines, and machine learning models. Bit by bit, the U.S. government is building a new kind of security state—one that operates not by force, but by algorithmic preemption. And the public may not realize what’s changing until it’s too late to question it.

On January 23, 2025, President Donald J. Trump signed Executive Order 14179, rescinding Biden-era AI ethics guidelines and launching a deregulatory push to make “American AI dominance” a matter of national policy. Within weeks, two new OMB memos—M-25-21 and M-25-22—were quietly released. These directives promised efficiency, innovation, and leadership. What they delivered was something else entirely: a blueprint for an unaccountable, AI-powered bureaucracy with expansive authority to surveil, predict, and punish.

Here’s what the new architecture looks like:

• Mass AI-enabled surveillance is being normalized within DHS, ICE, and DOJ, with no mandatory audits or transparency requirements.

• Autonomous military decision-making—once a red line for Pentagon ethicists—is now fast-tracked for deployment, free from independent oversight.

• And perhaps most chillingly, AI is being used to score, rank, and target protestors and dissidents, drawing from predictive models trained on historically biased policing data.

This isn’t speculative. It’s already happening—codified into law, normalized through bureaucracy, and shielded by a veneer of technical language. In this investigation, GTNM traces how these policies transform once-fictional fears into everyday administrative practice. We unpack their connection to Project 2025, explore the collapse of accountability frameworks, and ask a fundamental question: When algorithms decide who is dangerous, what happens to dissent?

Expanded AI Surveillance by DHS, ICE, and DOJ

“This is the AI-powered version of COINTELPRO—but scaled and privatized.” — GTNM Analyst

The memos establish that national security systems are exempt from transparency and oversight requirements. That means:

• Facial recognition, social media scanning, and license plate surveillance technologies can be rapidly deployed without risk audits.

• ICE and CBP can continue using tools like Palantir’s FALCON system and Amazon Rekognition without disclosing results, error rates, or demographic impacts.

• DHS fusion centers and state partners are likely to receive AI tools through federal grants, spreading unchecked surveillance capacity to local law enforcement.

Predictive policing—long criticized for racial bias—is now being federally legitimized through algorithmic systems built on opaque datasets.

️ Autonomous Military Decision Systems: Code over Commanders

While M-25-21 claims to promote “human flourishing,” it excludes military and national security AI systems from civil oversight.

What’s at stake:

• Drone swarms, autonomous targeting, and decision-support systems will expand.

• Military AI will be used in battlefield decision-making, with limited human involvement—contrary to previous Pentagon “ethical AI” frameworks.

• There are no requirements for model transparency, testing in adversarial environments, or accountability in the case of civilian casualties.

“We are entering the era of black-box warfighting.” — Former DoD AI Task Force Advisor

Meanwhile, AI-trained models using battlefield surveillance data will likely be privatized and sold back to federal or local law enforcement, completing the militarization loop.

AI-Based Threat Scoring for Activists and Protestors

Perhaps the most dystopian implication is the use of AI to target domestic political dissent.

Under the current framework:

• Agencies can use AI to rank individuals by “threat level”, based on social media posts, associations, or protest attendance.

• There are no required civil rights audits unless voluntarily initiated.

• These systems disproportionately flag Black, immigrant, Muslim, and leftist activists, based on biased training data and surveillance history.

Even the “pilot programs” mentioned in M-25-21 are allowed to run without safeguards, provided the agency signs off internally.

“The algorithm says you’re a threat. There’s no hearing. No appeal. Just a knock on the door.” — GTNM Research

Historical Parallels: From COINTELPRO to Predictive Repression

• In the 1960s and ’70s, the FBI’s COINTELPRO targeted civil rights leaders like Martin Luther King Jr. and the Black Panthers using wiretaps and informants.

• Today, the same surveillance logic is outsourced to algorithms, trained on historical policing data already tainted by systemic racism.

The threat isn’t just surveillance—it’s preemptive repression based on statistical assumptions. This isn’t just a policy shift. It’s a transformation of the state’s relationship to protest, dissent, and democratic action.

Connection to Project 2025

These AI expansions echo the Project 2025 vision, which emphasizes:

• Eliminating “deep state” resistance within federal agencies

• Centralizing executive control over law enforcement and military operations

• Privatizing data infrastructures and dismantling regulatory checks

The deregulation of AI procurement (M-25-22) combined with surveillance authority (M-25-21) reflects a blueprint for authoritarian technology governance, cloaked in the language of “efficiency” and “innovation.”

Fictional Parallels That Expose Real Risks

Minority Report (2002, dir. Steven Spielberg)

Premise: The state uses predictive technology (“Precrime”) to arrest people before they commit a crime, based on forecasts.

Real-World Parallel:

• The current memos allow predictive analytics by DHS, ICE, and DOJ using social media, location history, and behavior modeling.

• AI systems can now assign “threat scores” to protestors and activists without them committing any illegal act.

• “High-risk” individuals may be placed under surveillance, detained, or denied services—all based on probabilities.

❝Precrime isn’t science fiction anymore. It’s AI risk scoring without oversight.❞ — GTNM Analyst

1984 by George Orwell

Premise: A totalitarian regime uses constant surveillance, disinformation, and fear to control every aspect of public and private life.

Real-World Parallel:

• Facial recognition and social media tracking allow DHS and ICE to monitor dissent and identity in real-time.

• The removal of civil rights audits means algorithms can label protestors as threats—without legal recourse.

• Like Orwell’s telescreens, modern surveillance isn’t just watching—it’s watching to punish deviation.

❝War is peace. Freedom is slavery. AI is efficiency.❞ — The logic of EO 14179

️‍️ Black Mirror – “Metalhead” & “White Christmas” Episodes

Premise: Hyper-advanced surveillance, AI weaponry, and social ranking systems create a bleak, tech-dominated world.

Real-World Parallel:

• Military AI drones, autonomous targeting, and surveillance systems are being developed outside public view.

• The lack of transparency around model training means agencies could be using biased or unethical data sources—just like in Black Mirror’s dystopias.

Brave New World by Aldous Huxley

Premise: Control is achieved not through fear, but by making people love their servitude via technology and engineered comfort.

Real-World Parallel:

• The memos frame AI deployment as improving “efficiency,” “innovation,” and “human flourishing.”

• Soft language masks hard power—surveillance is sold as “personalization” and “optimization,” not control.

Gattaca (1997, dir. Andrew Niccol)

Premise: Society uses genetic data to score individuals’ value and access to opportunities.

Real-World Parallel:

• AI systems trained on biased data are being used to determine access to jobs, housing, credit, and public services.

• Like Gattaca’s DNA barcodes, your digital behavior may pre-determine your fate, especially if you’re marginalized.

RoboCop (1987)

Premise: A privatized, militarized law enforcement agency uses autonomous systems to patrol and punish citizens.

Real-World Parallel:

• The memos encourage public-private partnerships for AI development, often with tech and defense contractors.

• This raises concerns about unaccountable AI weapon systems, predictive law enforcement, and data sharing across opaque networks.

From Fiction to Policy: AI Dystopia Becoming Reality

What was once science fiction is now federal policy.

AI is no longer a speculative risk—it is a concrete power mechanism being embedded across military, immigration, and policing institutions, without safeguards or public accountability.

Final Thought: When the Future Becomes a Weapon

The true danger of artificial intelligence in governance isn’t just in what it can do—but in what it is being authorized to do without oversight. Executive Order 14179, along with M-25-21 and M-25-22, doesn’t just expand the government’s use of AI—it strips away the guardrails that once limited its most repressive applications. By deregulating procurement, removing civil rights protections, and centralizing decision-making authority under the executive branch, these policies transform AI into an invisible, scalable, and unaccountable infrastructure of control.

This isn’t a projection or a paranoid hypothetical. It’s already happening. Threat scores are being generated. Facial recognition tools are deployed with no audits. Predictive policing programs are fed biased historical data and marketed as “efficiency tools.” Military systems are being trained to make lethal decisions without humans in the loop. Protestors, immigrants, and marginalized communities are being preemptively profiled—often by algorithms that operate behind closed doors, shielded from both the public and the courts.

These technologies are powerful, yes—but that’s precisely why their use demands more scrutiny, not less. In the wrong hands, or even in unaccountable hands, they become tools of digital authoritarianism—a modern COINTELPRO running at machine speed, under the banner of “innovation.” And what’s most insidious is that this repression doesn’t announce itself with tanks or tear gas. It arrives in dashboards, decision matrices, risk flags, and quiet denials of services or freedom.

The future we imagined in dystopian fiction wasn’t a warning about machines. It was a warning about power—and what it does when it becomes too efficient to be resisted, too opaque to be questioned, and too automated to be reversed. When algorithms decide who is a threat, without appeal or accountability, democracy itself is placed on autopilot—flying blind into authoritarian terrain.

There is still time to resist. But only if we understand the stakes. Only if we fight not just for transparency, but for democratic control over the tools that now shape our justice, our wars, and our freedoms. The solution isn’t to halt technology—but to ensure it serves the people, not those who would control them.

At GTNM, we believe resistance begins with clarity. That’s why we write. That’s why we track every policy. That’s why we speak, even when the machines refuse to draw. The firewall over dissent must not be accepted as a feature of the system. It must be recognized for what it is: a warning we cannot afford to ignore.

The future is programmable. Let’s not code in our own captivity.

Citations and Relevant Links