AI Is Supercharging Global Disinformation — And Authoritarian Governments Are the Biggest Users

The Weapon Is Already Deployed

According to Freedom House's Freedom on the Net 2023 report — authored by Allie Funk, Adrian Shahbaz, and Kian Vesteinsson — global internet freedom declined for the 13th consecutive year. At least 47 governments deployed paid online commentators to manipulate public debate during the coverage period. That's double the number from just ten years ago.

Generative AI accelerated the trend. Freedom House documented AI-generated disinformation tools being used in at least 16 countries to smear opponents, sow doubt, and tilt public opinion. The technology is no longer experimental — it's operational.

What Authoritarian Governments Are Actually Doing

China retained its title as the world's worst environment for internet freedom for the ninth straight year, according to Freedom House. Myanmar came close to displacing it. Iran experienced the sharpest single-year decline, shutting down internet service entirely, blocking WhatsApp and Instagram, and ramping up surveillance to suppress anti-government protests.

Beatriz Saab, writing for the National Endowment for Democracy, documented the mechanics in her report Manufacturing Deceit: How Generative AI Supercharges Information Manipulation. Authoritarian actors are incorporating generative AI directly into their strategies — not just for one-off propaganda campaigns but as a systematic infrastructure for controlling what citizens believe.

The NED report offered a stark assessment: even if it's too early to measure decisive electoral wins for authoritarian powers using AI, the damage is already occurring. Citizens are losing trust in all online information. When nobody knows what's real, authoritarian narratives fill the vacuum.

The Scale Is Staggering

In 55 out of 70 countries surveyed by Freedom House, people faced legal consequences for expressing themselves online. In 41 countries, people were physically attacked or killed for what they posted. In Myanmar and Iran, governments carried out death sentences over online speech.

In Belarus and Nicaragua — run by Alyaksandr Lukashenka and Daniel Ortega respectively — people received years in prison for online speech. This is what AI-enhanced authoritarian information control looks like in practice.

What Mainstream Media Is Getting Wrong

Outlets like The Atlantic have focused heavily on Donald Trump's late-night Truth Social posts as the primary disinformation story of our time. Trump resharing conspiratorial content at 10:15 p.m. — including, per The Atlantic's own reporting, material originating from a satirical website — receives treatment as the defining threat to democratic information systems.

Meanwhile, China, Iran, Russia, Myanmar, and dozens of other authoritarian governments are running state-sponsored AI disinformation operations at industrial scale, targeting elections, suppressing protests, and imprisoning people for speech. The Atlantic devoted more coverage to Obama conspiracy theories than to the fact that 47 governments are systematically manipulating the information environment their citizens inhabit.

Both stories warrant attention. The proportions deserve reconsideration.

AI-Powered Censorship Infrastructure

According to Freedom House, at least 21 countries have enacted legal frameworks that mandate or incentivize digital platforms to use machine learning to remove disfavored political, social, or religious content. These aren't rogue hackers — they're governments building AI censorship into law.

The MDPI academic journal Surveillance, Disinformation, and Legislative Measures in the 21st Century identifies this convergence: AI-powered surveillance plus disinformation plus legislative cover creates a system difficult to reverse once operational.

Civil society organizations are using generative AI to push back — fact-checking at scale, identifying coordinated inauthentic behavior, mapping disinformation networks. But resources are unequal. Governments have budgets. NGOs have volunteers.

The 2024 Election Cycle

2024 saw more than 50 national elections worldwide, per the NED report. AI-generated disinformation was present in virtually all of them. Voters in dozens of countries made decisions inside information environments that were deliberately poisoned by state actors.

Authoritarian disinformation doesn't favor one American political tribe over another. China does not care whether Americans are Republican or Democrat. Iran's operations target American trust in institutions — and that strategy is working.

U.S. media outlets continue debating presidential tweets while authoritarian governments run coordinated AI operations designed to destabilize democracies worldwide.