The Dictator's Algorithm: How Authoritarian Governments Are Quietly Shaping the AI That Shapes You

The surveillance state is not a demonstration of strength. It is the most elaborate monument to fear ever constructed.

Let me ask you a question that nobody in the mainstream tech conversation seems to want to answer.

When you type a question into an AI assistant, when you watch a recommended video, when you read a news story that an algorithm decided you should see; who decided what you'd find?

Most people assume the answer is engineers, data scientists, neutral code. The invisible hand of technology working in your interest.

Here's what the research actually shows; some of the most powerful AI systems and the global standards being built around them are being quietly, deliberately shaped by authoritarian governments. Governments that have one goal. Not your convenience. Not your access to truth. Your compliance.

The Global Authoritarian Network

China and Russia are not just building surveillance states for domestic use. They are running a global operation; exporting the architecture of control to any regime willing to buy it.

A 2026 report from Action for Democracy, which used AI itself to track the activity, found that China and Russia are jointly involved in roughly half of all recorded authoritarian collaboration happening worldwide right now. The report documented something chilling; "surveillance infrastructure exported to one regime becomes a template for the next."  The technology of oppression is now a franchise model. China builds it, deploys it, refines it on its own population, and sells the perfected version to the next government that wants to silence its people.

And the information war is accelerating at a rate that should stop you cold. AI-driven disinformation campaigns have increased 350 to 500 percent since 2023. Synthetic videos. Algorithmically generated propaganda. Fake news that is indistinguishable from real news because it was written by a machine trained to sound credible. The projections show another 400 to 600 percent increase by end of 2026.

The Invisible Hand in Your Search Results

Here's where this enters your home directly.

You may have heard of DeepSeek, the Chinese open-source AI model that stunned the tech world earlier this year with its capabilities. Open-source means anyone can use it, including democracy advocates, journalists, and regular people. That sounds good. It might even be good, in some applications.

But here is the question nobody is asking loudly enough; when you don't know what data trained that model, when the fine-tuning process is opaque, when the political guardrails are invisible, how do you know what the model has been quietly instructed not to say?

You don't. And that's not an accident.

Meanwhile, at the United Nations' International Telecommunication Union, the body that writes the technical standards governing global communications, Chinese companies have been the sole proposers of facial recognition standards being fast-tracked for adoption across the developing world. Not one of several competing proposals. The only proposal. One country, writing the global rulebook for how AI identifies human faces.

If you don't find that alarming, I'd ask you to think about it for one more moment. The facial recognition system used to track and disappear Uyghur Muslims in China is the same architecture being exported, standardized, and embedded into infrastructure worldwide. By the time most countries realize what they've agreed to, the foundation has already been poured.

The Lie the Dictators Tell Themselves

Now, I want to share something that gives me a complicated kind of hope. Because the research shows that AI surveillance, for all its power, doesn't actually work the way authoritarian governments claim it does.

Harvard researchers published a landmark analysis this year on what they call the autocrat's calibration dilemma. The problem is fundamental and mathematical. Every AI prediction system must set a threshold, how suspicious does a person have to appear before the system flags them? Lower the threshold and you catch more threats but also sweep thousands of innocent people into the net, creating grievances and backlash. Raise the threshold and genuine dissidents slip through undetected.

There is no setting that solves this. It's structurally impossible. The researchers documented what happens in practice; cycles of tightening, backlash, loosening, and political damage-control that they call "threshold whiplash".

China's COVID Health Code system is the perfect case study. A color-coded app controlling whether you could travel or work, presented as neutral public health technology. Then officials in Henan Province turned citizens' codes red to prevent people from traveling to protest a banking scandal. A public health tool became a political weapon overnight. The public noticed. The White Paper protests erupted. The all-seeing eye blinked.

In Iran, AI cameras were installed to detect women violating the hijab law. In January 2026, millions took to the streets across all 31 Iranian provinces, the largest uprising since 1979. The AI didn't stop them. The regime survived through human violence; live ammunition and a death toll potentially exceeding thirty thousand. Not through technology. Through brutality. Those are two very different things, and the distinction matters.

The surveillance state can produce the appearance of control while the actual grievance, the real human fury at injustice, grows unseen in the dark.

Epictetus, Marcus Aurelius, and the Paranoid King

I bring the Stoics into everything on this platform, and I make no apology for it. Because 2,000 years ago, these philosophers understood power in ways that our most sophisticated data scientists apparently do not.

Marcus Aurelius was the most powerful man in the world. He could have built the Roman version of a surveillance state. He could have used his power to silence every critic, monitor every citizen, and crush every dissenting voice. What he actually did was write private notes to himself at night about how to be less of a tyrant. How to hold power with restraint. How to govern in a way that would leave people better rather than more afraid.

"You have power over your mind, not outside events."

The dictator who needs an AI system watching every citizen is not powerful. He is afraid. The surveillance state is not a demonstration of strength, it is the most elaborate monument to fear ever constructed.

Epictetus, who was an actual slave, who was owned by an actual human being, whose leg was broken as a demonstration of power over him, said; "No man is free who is not master of himself." The autocrat surveilling millions from his palace is the least free person in the entire country. He cannot sleep without the system running. He cannot trust any information that reaches him. He cannot believe the loyalty of anyone around him. He has built a prison so sophisticated that he is the most inescapable resident.

And Seneca, who watched Nero up close and knew exactly what unchecked power does to a human soul, wrote; "A kingdom founded on injustice never lasts."

The data from China, Russia, and Iran says the same thing.

What You Need to Do

I am not going to tell you to just "be aware." Awareness without action is just anxiety with better vocabulary. Here are three concrete things:

1. Question the source of your AI tools. Where was the model trained? Who built it? What is the political context of the company or government behind it? This is not paranoia. This is basic information hygiene in 2026.

   
   

2. Support international AI standards that include democratic accountability. Organizations pushing back against China's unilateral standard-setting at the ITU and similar bodies need public support and public attention. These are unglamorous, bureaucratic fights that determine the rules of the entire game.

   
   

3. Teach everyone around you the single most important question in the information age; "Who benefits if I believe this?" Every piece of content you encounter was placed in front of you by a system that was optimized for something. Know what that something is. Or at minimum, refuse to assume it's your benefit.

Marcus Aurelius wrote; "Waste no more time arguing about what a good man should be. Be one."

The dictators building their AI empires are counting on your confusion. Your passivity. Your assumption that someone else is handling this.

Nobody else is handling this.

This is The Virtuous Machine. And this is exactly what it was built to say.

➡️ Want to understand how AI is being used to manipulate what you see online? [Read: The Great Dumbing Down — How Tech Is Making Us Weaker and Calling It Progress].

➡️ Missed the Yoshua Bengio AI warning post? [Read: He Built AI. Now He Regrets It. Here's Why].

➡️ Want to talk about this with people who are paying attention? [Come find us in the free Tech 4 Grown-Ups community].

— Michael Routhier, Founder of Tech 4 Grown-Ups. I run free digital safety seminars for adults 55+ and write about tech threats as they happen. Learn more about me →