April 16 · Elections
How Hungarian Voters Overcame One of the World's Most Sophisticated Disinformation Machines
An important case study
On April 12, 2026, Viktor Orbán conceded. Péter Magyar's Tisza party had won 138 of 199 seats in the Hungarian parliament, more than the two-thirds supermajority needed to rewrite the constitution. Sixteen years of Orbán ended on a Sunday night in Budapest, with young Hungarians dancing on the banks of the Danube and a 24-year-old law student telling CNN he had to wake up and check his phone to believe it.¹
When Orbán conceded, I went looking for what had changed. I had heard something about EU rule changes on social media ads and information, and I wondered if that was the mechanism. If those rules were what broke the regime, that would be bad news for America, because we do not have those rules and are not getting them. It would also mean states have room to act on disinformation and no real excuse for sitting on their hands.
Worth asking. The answer is complicated.
The EU rules mattered. Meta banned political advertising across the EU in October 2025, citing legal uncertainty under the new framework. YouTube imposed similar restrictions in September 2025. In March 2026, the European Commission activated its Rapid Response System, giving fact-checkers and designated NGOs a fast-track channel to flag election content for demotion or removal across major platforms.²
And Fidesz routed around all of it.
Hungarian journalist Szilárd Teczár, who works with the fact-checking outlet Lakmusz, estimates that roughly 90 percent of the disinformation circulating before the vote was domestic rather than foreign, and the ruling party dominated that domestic share. Fidesz and its ecosystem shifted to proxy Facebook groups with tens of thousands of members, state-adjacent NGOs like the National Resistance Movement, an influencer network called Megafon, AI-generated material designed to duck political classification, and lifestyle pages promoting candidates without ever tripping the platforms' political ad flags. Hungary's own Ministry of Justice confirmed in January 2026 that the country had not designated a national authority to enforce the new political advertising rules and had not adopted the required sanctioning framework. Violations went unsanctioned domestically.³
Meta and YouTube closed one channel when they pulled paid political ads. Fidesz moved its spending elsewhere and the disinformation kept flowing.
What broke Orbán was bad economy plus a candidate who was right for anti-corruption and really stood against it. Analysts at Chatham House landed on the same reading. Economic stagnation, inflation, and falling living standards exhausted voters. Orbán kept campaigning on Brussels and Kyiv and war while people worried about groceries. Magyar gave them somewhere to go that looked like competence rather than continuity, and the disinformation machine kept running at full capacity while voters stopped buying what it was selling.⁴
Hungary was an electoral autocracy, not a fully captured state. Orbán still had to win elections, even on a tilted field, which meant the regime depended on propaganda saturation, media consolidation, and legal harassment of opposition to manufacture its majorities. When economic conditions turned badly enough and Magyar offered a credible alternative, the machinery failed because the machinery was built to shape majority sentiment, not override it.
Fully captured authoritarian regimes do not break the same way. The Soviet Union outlasted decades of material collapse. Nazi Germany intensified as it lost the war. Putin's Russia absorbs sanctions and military disaster because there is no functioning electoral mechanism left to turn against him. The material-conditions argument applies to the electoral version of authoritarianism, not the fully captured version. Once capture completes, force replaces elections as the mechanism that sustains the regime, and force works in the regime's favor.
If you're allowed to make dangerous lies to control people, that's like being allowed to get paid to knowingly shout fire in a crowded theater to cause harm from a rival theater. That intuition is correct, and the current American doctrine has not actually made the case for why we should not go further.
Start with what we already regulate without controversy. Securities fraud criminalizes knowingly false statements made to move markets. Wire fraud criminalizes knowingly false statements made to extract money. False advertising law lets the FTC go after companies that lie about what their products do. Perjury criminalizes knowingly false statements under oath. Defamation lets private parties recover damages for knowingly false statements that harm reputation. Every one of these regulates speech based on content, requires proof of falsity, requires proof of knowledge or recklessness, and none of them violate the First Amendment.
The principle running through all of them is simple. The First Amendment does not protect knowingly false statements made to cause specific harm. The government does not have to let a hedge fund defraud investors because fraud uses words. It does not have to let a witness commit perjury because perjury uses words. The First Amendment protects the marketplace of ideas, not the marketplace of knowing lies deployed for profit or power.
Apply that to political disinformation. A candidate's voting record is verifiable. A candidate's criminal history is verifiable. Whether a bill does what its opponents claim it does is verifiable. Whether an official took a specific action is verifiable. An operation that knowingly spreads false versions of those claims, paid to amplify them at scale for political return, is doing exactly what a securities analyst does when knowingly misrepresenting a company's earnings. Same structure. Same knowledge requirement. Same measurable harm to a market the government can protect. Different market. Why would we treat that differently?
The Supreme Court doctrine that governs this is narrower than the conventional wisdom suggests. United States v. Alvarez struck down the Stolen Valor Act in 2012 because it criminalized lying about military honors with no requirement of harm. The plurality opinion written by Justice Kennedy explicitly preserved the government's power to regulate false speech when it causes legally cognizable harm, listing fraud, defamation, perjury, and speech integral to criminal conduct as traditional categories where content-based regulation remains constitutional. Justice Breyer's concurrence, which controlled the outcome, applied intermediate scrutiny and signaled that narrower statutes tied to specific harms would survive. That ruling reaffirmed fraud and defamation regulation; it did not overturn them.⁵
Brandenburg v. Ohio, the 1969 decision that governs when the government can criminalize speech for its consequences, protects expression unless the speaker directs it at inciting imminent lawless action and the speech is likely to produce that action. An industry built to manufacture false factual claims about candidates and official acts, paid and coordinated at scale, meets that standard more cleanly than most of the speech courts have tested against it.⁶
The statute you would draft looks something like this. It targets paid or coordinated dissemination, not organic speech. It requires actual knowledge of falsity or reckless disregard, the same standard from New York Times v. Sullivan that governs defamation of public figures.⁷ It targets specific categories of factual claim capable of objective verification, such as candidate voting records, criminal histories, official acts, and the content and effects of legislation. It provides civil penalties with enhanced damages for repeat offenders and criminal penalties for coordinated operations. It includes a safe harbor for good-faith correction. It excludes opinion, satire, and contested interpretive claims. It sits outside direct political control the way the SEC sits outside direct political control, through independent commissions, judicial review, and evidentiary standards that force prosecutors to prove their case.
Some states have moved on deepfake disclosure, requiring labels on synthetic political media. That is better than nothing. But apparently we're allowed to commit fraud as long as it's political, and there's nothing in the First Amendment that tracks with that. Putting a disclosure label on the lie is not the same as treating the lie as fraud.
The objection is that political speech is special, that we cannot trust government to decide truth, that any regulation opens the door to partisan enforcement. Those concerns are serious and the statute has to address them. They also prove too much. Courts define truth every day. Juries decide whether a defendant committed murder. Judges decide whether a contract was breached. The SEC decides whether an analyst misrepresented earnings. The FDA decides whether a drug company lied about side effects. The entire legal system rests on the premise that properly structured institutions, with rules of evidence and burdens of proof and adversarial process and appellate review, can determine what happened and act on it. The claim that this becomes unworkable the moment the subject matter turns political is not a constitutional principle. It is a selective exemption, written by the people who benefit from it.
The same concerns apply to securities fraud enforcement and we solved them. The SEC is not a partisan weapon because the definition of securities fraud stays narrow, the enforcement machinery sits outside direct political control, and the burden of proof sits where the Constitution requires. The same architecture works here.
A constitutional order that protects investors from fraud but does nothing about fraud against voters has its priorities inverted. That is a convenient choice for corrupt courts rather than a constitutional mandate. A company that lies about what its toaster does gets sued into the ground. A political operation that lies at industrial scale about a candidate's record, about what a bill does, about anything verifiable, faces nothing. The legal system treats buying a defective appliance as a more serious injury than voting under false pretenses for offices that control armies and courts.
Part of what we need to talk about here is that there were incredibly sophisticated things at play related to the media in Hungary and they still lost. That is a big deal. A note of hope. You can have captured media. It cannot get more consolidated than it was there. A government aligned with Putin while Russian disinformation operations ran through the campaign. And it can still be won.⁸
Three days after winning, Magyar walked into Hungarian state television and called it a factory of lies. He compared what the network had been doing since 2010 to what Goebbels and the North Korean leadership would admire. He told the host that not a single true word had been spoken there. He announced that his government will suspend the broadcaster until its public service character is restored.⁹ That is the energy. Any American candidate who wants to win in 2026 or 2028 needs to be able to walk onto Fox News and say the same thing. You are deeply corrupt. You are done when I am in charge. The fraud argument and that posture are the same argument. One says the legal system should name what they do. The other says the candidates should name what they do. Both are necessary.
Things are bad and they are getting worse. Trump's approval sits around 40 percent, which is typical for him. Democrats have overperformed in special elections. When Trump is not on the ballot, his supporters seem to blame the Republican Party and stay home. The work available to us runs race by race, seat by seat, statute by statute. Treating paid political disinformation as the fraud it is would be one piece of that work. Not the whole answer. One lever we are refusing to pull while the regime consolidates around us.
Apparently we're allowed to commit fraud as long as it's political? No. There's nothing in the First Amendment that tracks with that.
We need 10 subscribers per article, we need 10 people per article to keep this independent research, media, and activist journalism alive. Be one of the 10 so our millions of readers can keep growing and making change.
Click on this sentence to find everything as a free download, listed in the store for $0.00.
Works cited
Bell, M., et al. (2026, April 12). Hungary election 2026 results: Petér Magyar wins, Trump ally Viktor Orbán concedes landmark defeat. CNN. https://www.cnn.com/2026/04/12/world/live-news/hungary-election-orban-magyar
Brandenburg v. Ohio, 395 U.S. 444 (1969).
Chatham House. (2026, April). Hungary election: Orbán has been defeated – but will Orbánism survive? https://www.chathamhouse.org/2026/04/hungary-election-orban-has-been-defeated-will-orbanism-survive
EDMO. (2026, March). A dystopian electoral campaign will be over soon in Hungary, but disinformation's damages are here to stay. https://edmo.eu/publications/a-dystopian-electoral-campaign-will-be-over-soon-in-hungary-but-disinformations-damages-are-here-to-stay/
Euronews. (2026, April 15). Magyar vows to shut down Hungarian state TV accusing it of 'North Korean' propaganda. https://www.euronews.com/my-europe/2026/04/15/magyar-vows-to-shut-down-hungarian-state-tv-accusing-it-of-north-korean-propaganda
Euronews Next. (2026, April 13). Fabricating proof: The disinformation tactics that shaped Hungary's election. https://www.euronews.com/next/2026/04/13/fabricating-proof-the-disinformation-tactics-that-shaped-hungarys-election
Liberties.eu. (2026, March). How political advertising is adapting ahead of Hungary's 2026 elections. https://www.liberties.eu/en/stories/political-advertising-hungary-elections/45607
New York Times Co. v. Sullivan, 376 U.S. 254 (1964).
United States v. Alvarez, 567 U.S. 709 (2012).
Subscribe by email →