Brussels' long war against freedom of expression

There's a document you should read. It's 160 pages long, authored by the staff of the U.S. House Judiciary Committee, and it contains something remarkable: evidence. Thousands and thousands of pages of internal corporate communications, produced under parliamentary subpoena by ten of the world's largest technology companies—Meta, Google, TikTok, X, Apple, Amazon, Microsoft, and others—that meticulously and devastatingly document how European Commission officials, for a decade, quietly pressured Silicon Valley to suppress opinions they didn't like. Not illegal opinions. Not dangerous opinions. Political opinions. Conservative opinions. Opinions on migration, COVID-19, gender ideology, and electoral integrity. Here's an inventory in nine acts, for anyone who still believes democracy and thought control are compatible.

The committee has published two interim reports. Part I appeared in July 2025, Part II in February 2026. Together, they constitute the most comprehensive public reckoning with what critics have long suspected and EU defenders have always denied: that the Digital Services Act—the EU's epochal regulatory framework—is not a security law. It is a censorship regime. With global reach. Designed and implemented with partisan intent. And one that has already influenced election results on both sides of the Atlantic. Welcome to the brave new world of democratic information control. It's now called "risk prevention." It used to be called something else.

Prologue: The Philosopher-King of Censorship
Before we get to Brussels, we should pause briefly in Palo Alto. For the intellectual framework for everything the European Commission has built – the justifications, the framing, the moral self-assurance – was articulated with unusual clarity by an American on American soil, four years before this controversy reached its current boiling point.
On April 21, 2022, Barack Obama delivered the keynote address at a symposium on “Challenges to Democracy in the Digital Information Age,” co-hosted by the Stanford Cyber ​​Policy Center and the Obama Foundation. It was a polished, hour-long speech, delivered with the former president’s characteristic eloquence and genuine self-assurance. At its core, it was also an elaborate argument for why governments must control speech to save democracy—and why those who oppose this argument are, at best, naive and, at worst, complicit in authoritarianism.
Obama's central thesis: The free flow of information—precisely what the First Amendment is meant to protect—has become democracy's greatest weakness. Social media platforms, he argued, "accelerate" humanity's worst impulses, amplify disinformation, sow distrust, and create the conditions in which autocrats thrive. He mentioned Putin and Steve Bannon in the same breath. He declared, "People are dying because of misinformation," citing COVID-19 vaccine skepticism as prime evidence.
Obama's solution: Technology companies must "reinvent" themselves under government oversight. Content moderation doesn't go far enough. Platforms have financial incentives to circulate dangerous content and must be compelled to do more through regulation. Decisions about what is true and what is dangerous should not be left solely to private companies. They must be subject to government oversight.
What Obama didn't say—and what he carefully avoided as a lawyer and politician—was that he didn't openly call for government officials to maintain lists of prohibited opinions. He framed his argument cautiously, acknowledging the First Amendment and insisting that he was only talking about mitigating the "worst harm" of disinformation.
But the logic of his argument, followed to its logical conclusion, leads precisely to this point. If disinformation is an existential threat to democracy, and if platforms cannot be trusted to combat it voluntarily, and if state regulation is necessary to enforce it—then someone must be authorized to decide what constitutes disinformation. In a democracy, that someone is ultimately the state. And the state, composed of people with political interests and ideological convictions, will exercise this power in a politically motivated and ideologically driven manner.
Who decides what constitutes disinformation? Whoever holds the regulatory pen. And the pen is never held by no one. Obama didn't answer this question in his Stanford speech. He sidestepped it with characteristic elegance, offering assurances of independent oversight, journalistic standards, and citizens' own responsibility for critical media consumption. Not bad ideas. But no answers to the structural problem his own argument creates. The European Commission answered the question for him. It decided who holds the pen. The Commission. And the Commission has wielded that pen—as we shall see—with a political precision that should alarm anyone who took Obama's feigned concerns about democracy seriously.
The irony is exquisite and deserves to be savored. Obama delivered his Stanford speech, warning of the dangers of foreign authoritarian interference in democratic information ecosystems. He explicitly called Russia's manipulation of social media platforms a threat to Western democracy. He called for regulatory frameworks to protect the integrity of public discourse. And the European regulatory framework, born from precisely this ideological tradition, driven by precisely these justifications, built by bureaucrats who share precisely these values—this framework, as the House Judiciary Committee has extensively documented, has been used to annul election results, suppress political opponents, and shape the information environment of European voters in ways that just happen to favor the established political order. Obama's Stanford speech is the original text of the ideology of "censorship as a safeguard for democracy." It is deliberate, sincere, and wrong in a way that matters enormously.

I. The machine is built in the dark.
The story doesn't begin with the DSA itself, which was adopted in 2022, but with a decade of preparatory work. Starting around 2015 and 2016, according to the House Judiciary Committee's investigation, senior European Commission officials began holding a series of meetings with the major social media platforms. The stated purpose was innocuous: combating "hate speech" and "disinformation." The actual purpose, the documents suggest, was something far more specific.
Europe experienced the same populist revolt that shook establishment politics across the Western world. Voters, angered by mass immigration, economic stagnation, and elitist condescension, flocked to parties the European press labeled "far right"—parties that, in most cases, simply espoused views that had been mainstream a generation earlier. The platforms, with their algorithmic indifference to editorial gatekeepers, gave these movements a megaphone that bypassed state broadcasters and mainstream media.
The Commission worked to censor true information and political opinions on some of the most important political debates in recent history, including the COVID-19 pandemic, mass immigration, and transgender issues. The Commission's solution: to turn the platforms into enforcers. Over the course of more than 100 closed meetings documented in the seized documents, Commission officials pressured company representatives to tighten their content moderation rules globally—not just within Europe. The trick, as the documents reveal, was that they understood a fundamental reality of how platforms operate: you can't simply run separate content moderation regimes for different countries. If Europe demands that a certain type of speech be suppressed, the practical effect is that it will be suppressed everywhere.
The COVID-19 pandemic dramatically accelerated the process. In November 2021, the Commission contacted TikTok, asking how the platform planned to combat “disinformation” about COVID vaccines—specifically in the United States, not Europe. The Commission requested information about TikTok’s plans to “remove” certain claims about vaccine efficacy that were directed at American children. This was a foreign government bureaucracy ordering an American company to censor American speech on American soil. It was done via email.

II. The law that makes it permanent
The informal pressure campaign was codified and amplified when the Digital Services Act 2023 came into force. The DSA is, at first glance, platform regulation. It requires large platforms to assess and mitigate “systemic risks” to civic discourse, electoral processes, and public health. It empowers regulators to appoint “trusted whistleblowers”—approved organizations whose requests to remove content from platforms must be given priority. It threatens non-compliant companies with fines of up to six percent of their global annual revenue—for a company like Meta, that could amount to billions. The DSA isn’t just regulating Europe. It’s exporting European standards of discourse to the entire world.
The extraterritorial scope of the law is key to understanding why it is relevant to everyone. Platforms like Facebook, YouTube, X, and TikTok do not maintain country-specific content moderation systems for practical reasons. Users travel. VPNs are ubiquitous. The cost and complexity of geographically precise moderation are prohibitive. This means that changes to the global community guidelines that the Commission has been explicitly urging platforms to implement—as the seized documents show—apply to users in Zurich just as much as to users in Zagreb.
The Part I report contains a particularly striking example from a Commission workshop. Regulators labeled a hypothetical social media post containing the phrase "We must take back our country"—a phrase used by politicians across the ideological spectrum—as "illegal hate speech" that platforms are obliged to censor under the DSA. This is not a fringe interpretation. This is what Commission officials taught the platforms' compliance teams in an exercise documented in internal company files. It's astonishing. "We must take back our country"—illegal. Guilt admitted, proceedings dropped. But "We are building the largest digital censorship network in history"—that, of course, is protecting democracy.

III. Elections: The clearest proof
While the broad claims about content moderation policies may seem abstract, the election-specific evidence is harder to ignore. The Part II report identifies at least eight European national elections where the Commission activated a “rapid response system”—a mechanism through which approved fact-checkers and state-designated “trusted whistleblowers” ​​can submit priority requests to platforms to remove content in the days before and after voting.
The elections mentioned in the report span the continent:

• Slovakia 2023: Under pressure from the Commission, platforms censored statements, including "There are only two genders," as hate speech – content that had nothing to do with the election itself.
• Netherlands 2023 and 2025: Government agencies were granted the status of "trusted whistleblower," which allowed for faster removal of content before elections won by populist Geert Wilders.
• France 2024: A coordination meeting before the election between Commission officials, national regulators and "left-wing NGOs" discussed which political content should be moderated.
• Ireland 2024 and 2025: The Irish Media Authority held “DSA Election Roundtables” with the Commission and platforms. Meta confirmed that it had updated its “election risk assessment and measures”.
• Romania 2024: The most dramatic case – and the one that puts everything in a different light.

The commission's response to all this was to dismiss the reports as "pure nonsense" and "completely unfounded." But the documents are not the committee's invention. They were produced by the companies themselves under legal duress. The question is not whether these meetings took place. The emails prove they did. The question is whether they represent legitimate work to ensure electoral integrity or partisan influence disguised as security jargon.

IV. Romania: The censorship that annulled an election
In November 2024, a political outsider named Călin Georgescu unexpectedly won the first round of the Romanian presidential election with 23 percent of the vote, having risen from single-digit polling numbers in just a few weeks. Romania's intelligence services, the Constitutional Court, and the European Commission immediately pointed the finger at TikTok: Russian-linked bot networks, they said, had artificially amplified Georgescu's content and manipulated the platform's algorithm. On December 6, 2024, the Constitutional Court made history. It annulled the election results—the first time a European nation had done so on the grounds of social media manipulation.
The narrative was neat, convincing, and politically extremely convenient. Georgescu was portrayed as pro-Russian and anti-NATO. His victory would have been an embarrassment for the EU establishment. The interference allegation gave the authorities the legal basis to annul the result and exclude him from the runoff election in May 2025, which a pro-EU candidate won. Then the evidence came. TikTok's own statement to the European Commission—documents produced under the House Judiciary Committee subpoena—stated that the company "had found no evidence of a coordinated network of 25,000 accounts connected with Georgescu's campaign, nor had any such evidence presented to it." This is the platform that was accused of being a vehicle for interference. It found no evidence of interference.
TikTok informed the European Commission that it had found no evidence of the Russian bot network that the Romanian Constitutional Court had cited to justify the annulment of the election. Furthermore, the Romanian investigative journalism outlet snoop.ro, citing confidential sources within Romania's own tax authority, reported that at least one of the TikTok influencer campaigns had in fact been funded by Romania's National Liberal Party—a member of the established mainstream coalition, not a foreign government.
None of this led to the reinstatement of the election results. The rerun took place. The pro-EU candidate won. And the Commission, which had used the Romanian case to justify its TikTok investigation under the DSA, never publicly addressed TikTok's own denial of the central issue. This is the most serious accusation in the entire debate: that the censorship apparatus built under the DSA was used not to protect an election, but to alter it. Whether one considers Georgescu a genuine threat or a legitimate democratic election, the principle is the same. Governments should not have the power to decide which election results they recognize based on retrospective disinformation findings that the platform in question denies. It cannot be repeated often enough: An election was annulled. Based on evidence that the main platform denies. By a constitutional court acting under pressure from a foreign bureaucracy. And the winner of the annulled election was barred from the rerun. In a different geopolitical context, this would be called a coup. In the EU, it's called combating disinformation.

V. The German Paradox
The German federal election of February 2025 is revealing because the evidence disrupts the simple censorship narrative—albeit in a way that raises its own troubling questions. Before the election, the commission conducted stress tests on major platforms and convened roundtables with the German digital services coordinator to discuss “risks.” Germany’s domestic intelligence agency warned of Russian disinformation campaigns. In Hesse, a task force was established to “analyze opinions on social media platforms and coordinate measures.” State officials warned police officers against membership in regional branches of the AfD—a party polling above 20 percent.
And yet: Independent research by Global Witness and the Institute for Strategic Dialogue revealed that TikTok and X's algorithms disproportionately amplified AfD content compared to other parties. Elon Musk used X to openly support the AfD, hosting its leader in a livestream and telling his 220 million followers to vote for her. The AfD came in second place with 20,8 percent of the vote. Then, months after the election, Germany's domestic intelligence agency designated the entire AfD—the main opposition party in the Bundestag—as an "extremist organization" subject to increased surveillance. The designation was quickly suspended pending a legal challenge.
What the German case demonstrates is that the DSA's electoral integrity framework coexists with other forms of state pressure and does not prevent their use against opposition parties. The AfD was not suppressed on TikTok. It was suppressed in other ways: through intelligence agency labels, other parties' refusal to cooperate legislatively, and an overwhelmingly hostile media environment. The DSA was merely one tool in a broader toolbox.

VI. What this means for the rest of the world
The Trump administration reacted to these findings with unusual resolve. Secretary of State Marco Rubio imposed visa bans on five European officials, including former DSA architect Thierry Breton, and labeled them leaders of the “global censorship industrial complex.” The State Department launched an internal investigation into DSA enforcement in early 2025.
The DSA Researcher Access Provision—enforced in the December 2025 Commission fine against X—claims the right to demand that an American company release data about American users to researchers approved by European regulators. This is an extraordinary extraterritorial claim to the private information of American citizens, imposed by an unelected foreign bureaucracy.
When a European regulator tells a platform to change its “global community guidelines,” it makes content moderation decisions for Iowa just as it does for Ireland. More broadly, the structural reality of global content moderation means that the protective effects of the First Amendment are being quietly eroded by foreign regulatory pressure. This is not a hypothetical future threat. The seized documents show that it has already happened: platforms changed their global moderation rules in response to Commission pressure, and these rule changes applied to everyone else as well. The COVID-19 content labeled as misinformation in Brussels was also labeled misinformation elsewhere. The immigration discussion that crossed the line into “hate speech” by EU standards crossed that line for other users, too.
The $120 million fine imposed on X in December 2025 was officially for transparency regarding blue checkmarks and ad repositories. But the underlying principle is clear: American companies serving American users should not be regulated by unaccountable foreign bureaucrats whose definition of acceptable speech fundamentally contradicts free speech.

VII. The Defense of the Censors
Fair is fair: The Commission's defenders make several arguments that deserve attention. First, they note that the DSA contains no indication of a bias against conservative, populist, or right-wing content. Its language is neutral: It addresses "illegal content," "systemic risks" to civil discourse, and algorithmic transparency. European legal scholars point to the regulation's first article, which commits to upholding freedom of expression.
Secondly, they argue that the German example – where algorithmic analysis showed that the AfD was strengthened rather than suppressed – proves that concerns about anti-conservative bias in DSA enforcement are at least partially unfounded.
Third, they claim that the House Judiciary Committee reports themselves are political documents, produced by the Republican majority under Trump ally Jim Jordan, to attack EU regulations that restrict American technology companies whose executives are close to the Trump orbit.
These are not frivolous objections. But here's what's undisputed: The meetings took place. Pressure was exerted. The platforms changed their rules. And in Romania, a national election was annulled based on allegations of interference, which the platform at the center of the story denies.

VIII. A new kind of war
There is a framework that makes everything described in this essay understandable. It is not the framework favored by the defenders of the European Commission, who portray the DSA as straightforward consumer protection regulation. It is the framework of military doctrine: fifth-generation warfare.
In fifth-generation warfare, the primary battlefield is not territory. It is not infrastructure. It is the human mind. The target is not an enemy army, but the ability of an enemy population to perceive reality clearly, to form coherent political judgments, and to act collectively in its own best interests. The weapon is information itself—or more accurately, the management of information. Victory is not achieved when the enemy surrenders, but when the enemy population can no longer distinguish between truth and falsehood, friend and foe, or their own interests and those of their adversaries. In fifth-generation warfare, the battlefield is the human mind. Victory is achieved when a population can no longer think clearly enough to defend itself.
The concept was primarily developed in the context of non-state actors and foreign adversaries. Russian theorists call their version the Gerasimov Doctrine, after General Valery Gerasimov's 2013 essay, which argues that the lines between war and peace have become indistinguishable and that informational, psychological, and political tools are now as crucial as tanks and missiles. Obama himself alluded to this in his Stanford speech when he quoted Putin's alleged insight that you don't have to make people believe disinformation. You just have to flood a country's public sphere with enough noise so that citizens no longer know what to believe.
What has received far less attention is the question of what happens when these techniques are used not by foreign adversaries, but by governments against their own citizens. It is sometimes called "reflexive control"—a term from Soviet and Russian military psychology that describes the manipulation of an adversary's decision-making process by providing a carefully curated picture of reality. Operating on false or incomplete information, the adversary makes decisions that serve the manipulator's interests, while believing they are acting freely. Reflexive control: The manipulation of a population's decisions through the curation of the information they receive, so that people choose what the controller wants, while believing they are choosing freely.
The architecture of the DSA fits this framework with uneasy precision. Let's look at what the law actually does in practice. For the most part, it doesn't order the removal of specific content by government decree. That would be too obvious, too legally vulnerable, too much of a reminder of the censorship regimes that Europeans should have rejected in 1945. Instead, it creates a system of incentives and pressure that compels platforms to curate their own information environments in directions favored by the Commission. The threat of ruinous fines—up to six percent of global revenue—creates a powerful incentive for platforms to err on the side of over-removal for content in gray areas. The Trusted Flagger system gives approved organizations—selected by government regulators—the power to prioritize reporting content for review in the critical days leading up to elections. The systemic risk provisions require platforms to assess and mitigate broad categories of language – including fully legal language – if regulators determine that such language poses risks to civil discourse or electoral integrity.
The result is an information environment shaped at the margins in a way that favors established political narratives over dissenting ones, institutional authority over popular skepticism, and approved experts over unofficial voices. Citizens navigating this environment believe they are encountering a natural marketplace of opinions. What they don't know is that the marketplace has been quietly reorganized through regulatory pressure applied in closed workshops between Commission officials and platform compliance teams.
The psychological warfare dimension is not incidental to the design of the DSA. It is structural. When the Commission labeled the phrase “We must take back our country” as illegal hate speech in a training exercise for platform compliance teams, it was not simply making a legal decision. It was encoding a political judgment into the informational infrastructure of European society. The word “infrastructure” is important here. Infrastructure is what you don’t notice until it fails. For most of history, governments that wanted to control information had to do so visibly: burning books, shutting down newspapers, arresting editors. These actions were recognizable as censorship and generated corresponding resistance. The genius of the DSA model—whether intended or not—lies in the fact that it operates at the infrastructure level. It doesn’t tell citizens what to think. It shapes the informational environment through which citizens form their own thoughts without their knowledge or consent.
Military psychologists distinguish between first- and second-order effects in information operations. First-order effects are direct: false information is believed. Second-order effects are more powerful and lasting: the epistemic trust of the target population is eroded. They become less certain about what is true, more susceptible to official narratives, and more dependent on authoritative sources to interpret reality.
This is not merely hypothetical. This is a precise description of the cultural moment we find ourselves in – a moment in which questions that were publicly debatable ten years ago are now considered "dangerous disinformation," in which certain political positions are not refuted but rather de-platformed, and in which more and more people sense that the space for free speech is narrowing, without being able to pinpoint exactly where the line has been drawn and by whom.

IX. The bigger picture
Step back from the partisan noise, and the picture that emerges is disturbing, regardless of political sympathies. Democratic self-governance requires that voters be free to receive and discuss information. Social media platforms have become the dominant venue for this discussion. And these platforms are now subject to a regulatory framework—enforced by unelected officials in Brussels—that gives state-approved bodies the power to pre-rank requests to remove content in the weeks leading up to elections, defines broad categories of political speech as “systemic risks” that need mitigation, and threatens companies with ruinous fines for non-compliance.
Whether this framework is used with partisan intent or genuine neutrality, its structure creates a serious vulnerability. Whoever controls the definition of “disinformation” controls what voters can easily find and discuss online. The EU’s response is that independent regulators and judicial review provide sufficient safeguards. The response from the House Judiciary Committee—and increasingly from many European voices—is that these safeguards are inadequate and the framework itself is the problem. The right answer probably lies somewhere between these two extremes. But the conversation cannot be honest as long as the Commission insists its actions are “pure nonsense” instead of addressing tens of thousands of pages of its own internal communications.
What makes the Romanian episode so significant goes beyond its immediate facts. It sent a message to every voter in Europe: the result of an election can be overturned by state authorities on the grounds of informational threats that citizens cannot independently verify, using evidence that the platform in question denies, in processes that are not subject to full public scrutiny. The message—whether received consciously or unconsciously—is that electoral democracy operates within limits set by the state's informational judgment. This is not democracy. This is administered democracy—the form of government that Vladimir Putin, the man Obama identified as a master of information warfare, practiced in Russia for twenty years.
The Founding Fathers of the American Constitution designed the First Amendment precisely because they understood that governments will always be tempted to manage the information environment in a way that serves the ruling class. The DSA is the most sophisticated version of this impulse ever constructed. It is not the work of obvious tyrants. It is the work of sincere bureaucrats who sincerely believe they are protecting democracy while creating the conditions under which democracy degenerates into a caricature of itself. The paradox is complete: A regulatory framework erected in the name of democracy threatens democracy's most fundamental prerequisite—the free exchange of ideas among citizens without government approval, curation, or post-processing.

Conclusion: Read the documents
The two House Judiciary Committee reports are fully public. They are imperfect documents—accusatory in tone, written for political impact, produced by people with strong institutional interests in their conclusions. Read them with appropriate skepticism. But read them. Because whatever you think of the Trump administration, whatever you think of Jim Jordan, whatever you think of Elon Musk—the documents behind these reports are real. The emails between Commission officials and platform compliance teams are real. TikTok's denial of the Romanian bot network is real. The Irish regulator's DSA election roundtables are real. The global changes to content moderation rules are real.
A foreign, unelected bureaucracy has spent a decade trying to shape what can be said and read online. The question isn't whether you like the speech it has suppressed. The question is whether you believe unelected foreign officials should have this power in the first place.
Freedom of expression is the foundation of all other freedoms. If Brussels decides what is disinformation, it decides what is true. That is a power no government should possess.
The censorship machine is running. It's running quietly. And it's running with your consent—a consent you never explicitly gave because you were never explicitly asked. This is called administered democracy. It used to be called something else…

Sources
The following official documents support this analysis:
• House Judiciary Committee: Part I Report (July 25, 2025): https://judiciary.house.gov/sites/evo-subsites/republicans-judiciary.house.gov/files/2025-07/DSA_Report&Appendix(07.25.25).pdf
• House Judiciary Committee: Part II Report (February 3, 2026): https://judiciary.house.gov/sites/evo-subsites/republicans-judiciary.house.gov/files/2026-02/THE-FOREIGN-CENSORSHIP-THREAT-PART-II-2-3-26.pdf
• Part I Press Release: https://judiciary.house.gov/media/press-releases/foreign-censorship-threat-how-european-unions-digital-services-act-compels
• Part II Press Release: https://judiciary.house.gov/media/press-releases/new-report-exposes-european-commission-decade-long-campaign-censor-american