On January 20, 2025, the first day of his second presidential term, Donald Trump signed an executive order: “Restoring Freedom of Speech and Ending Federal Censorship.” The bad old days of the “censorship-industrial complex,” allegedly responsible for suppressing online speech under President Joe Biden, were over.
Except they weren’t. The driving force behind online censorship had never been the U.S. government, which meant that freedom of speech could not be restored by the stroke of a president’s pen. Rather, the European Union has wielded its Digital Services Act (DSA) to restrict the speech not just of Europeans but especially of Americans and other English-speakers. The E.U. has not violated the free-speech rights of Americans, since it has no obligations under the U.S. Constitution. But it has vitiated those rights, essentially nullifying the First Amendment in cyberspace.
The DSA is not a “threat” to free speech, as some American commentators put it, implying that possible danger lies in the future. Because the DSA is in force now, all major online platforms and search engines must comply with it to remain on the E.U. market. There is effectively no free speech on the internet nowadays, at least not on the major platforms falling under the DSA’s strictest provisions, but only more or less heavily curated, algorithmically managed speech.
Some supporters of President Trump might find this hard to believe. After all, the president’s most prominent ally and advisor is Elon Musk, whose purchase of Twitter in 2022 was said to be motivated by a desire to restore free speech to the platform. But Musk has always insisted that “freedom of speech is not freedom of reach,” and there’s the rub. Using platform algorithms to restrict reach artificially is a form of censorship, one that is not only compatible with the DSA but even encouraged by the E.U.
The Trump Administration can truly restore free speech to the internet only by confronting the European Union. The administration needs to challenge the DSA, to get it repealed or at least neutered. If the E.U. refuses to back down, then the administration will need to work with Congress to pass a law ensuring that American tech companies cannot comply with the DSA by restricting Americans’ First Amendment rights.
Germany Censors the World
H.L. Mencken once said that “freedom of the press is limited to those who own one.” But the advent of the internet and the rise of blogs around the turn of the 21st century made Mencken’s observation obsolete. Now, virtually everyone could be a publisher, with the only barrier to entry being the price of an internet connection. The democratic potential of this development is obvious. The rise of social media extended it further, allowing us, in effect, to publish our every passing thought.
The boundary between private conversation and public debate had been blurred and the distinction between freedom of speech and freedom of the press had thus been effaced. Freedom of speech had precisely become freedom of reach.
This development was not universally welcome. Beginning about a decade ago, Germany and then the European Union as a whole launched a series of initiatives that treated the burgeoning of online speech as a threat rather than an opportunity. These first took the form of task forces or “codes of conduct” into which online platforms like Facebook, YouTube, and Twitter were enlisted on an ostensibly voluntary basis. The earliest efforts were devoted to “hate speech,” some forms of which are even illegal in Germany and other European countries, in which freedom of speech does not enjoy protection equivalent to America’s First Amendment. In 2016 the European Commission unrolled a Code of Conduct on Countering Illegal Hate Speech Online, one year after Germany had organized a task force on the same issue. These “voluntary” arrangements soon gave way to binding laws requiring platforms to restrict speech or face penalties.
Already in the early days of the internet, American lawmakers had recognized that the new communications technology’s great potential would be severely constrained if internet service providers or intermediate users, such as hosted forums, were held liable for everything that other users said and/or did in using their services or platforms. In 1996 Congress thus stipulated in Section 230 of the Communications Decency Act: “No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.” This meant that all internet users would be responsible for their own words. The intermediate services that merely enabled them to be published would neither bear nor share in the responsibility.
Twenty years later, German and European legislators took the opposite approach, insisting that online platforms would bear not only liability but indeed special liabilities for what users said or posted on them. First the so-called Network Enforcement Act or “NetzDG” in Germany in 2017 and then the E.U.’s 2022 Digital Services Act touted this approach under the motto “What is illegal offline, should be illegal online.” But what was illegal offline still was illegal online anyway. This was never the issue. The issue was who is to be held accountable for it.
As Congress realized in 1996, if the platforms were held accountable, then they would be forced to censor. Indeed, they would be forced to censor much more than just illegal content, since they could not know in advance what exactly the competent judicial authorities would judge to be illegal. As Facebook’s German subsidiary put it in 2017 in a statement objecting to the NetzDG legislation, they would have to apply the principle “in dubio pro deleo” [sic]: when in doubt, delete!
There was another problem. “What is illegal offline, should be illegal online” is all well and good—but illegal in what jurisdiction? Many utterances illegal in Germany would clearly be protected speech in America. But the internet is global. Germany’s NetzDG, in effect, extended German laws to the entire world. The real practical import of the law was “What is illegal offline in Germany, should be illegal online everywhere.”
This was made abundantly clear by the NetzDG notices that many American and other English-speaking Twitter users received. Even if the notices typically concluded reassuringly that users’ posts were not in contravention of German law, they left them understandably wondering why their posts had to comply with German law in the first place. What’s more, despite the reassurances, Seth Dillon of the satirical website The Babylon Bee determined that his account would be locked and remain so until he deleted content that had been “reported by people from Germany.”
The E.U. as Arbiter of Truth Itself
Such notices would disappear once the Digital Services Act went into effect. The DSA also superseded NetzDG in Germany, but the same principle applies. The world’s online discourse is now subject to E.U. law, which means, in turn, subject to all the speech laws of each of the 27 E.U. member states. Individuals or “entities” can still report offending content to the platforms. National authorities can name “trusted flaggers” whose reports are given priority treatment. And, under the DSA’s Article 9, both member states and the European Commission can simply issue orders requiring action against specified content. Noncompliant platforms face ruinous fines of up to 6% of their global turnover.
Under the DSA regime, however, platforms are expected to censor much more than just illegal speech. They must also act against alleged mis- or disinformation—not on the grounds that it is illegal, but on the grounds that it is “harmful.” The DSA thus extends online censorship from so-called hate speech to purely factual discourse. In so doing, it makes the European Commission and, in a subsidiary role, E.U. national authorities not just arbiters of compliance with their own laws but arbiters of truth itself.
Already in 2018, the European Commission had rolled out a second “code,” the Code of Practice on Disinformation, to supplement the 2016 Hate Speech Code. Under its aegis, the Commission then launched a dedicated program for specifically combatting COVID-19 “disinformation” in June 2020. In the meantime, in December 2019, incoming Commission President Ursula von der Leyen had first unveiled the DSA as her hallmark piece of legislation, the very purpose of which was to turn the “voluntary” commitments undertaken under the Codes into legal obligations.
Online censorship of alleged COVID-19 mis- or disinformation represented a watershed in the history of free speech and censorship in the West. In light of internal communications revealed in Elon Musk’s “Twitter Files” publications, and several highly publicized though unsuccessful lawsuits inspired by them, many Americans would come to believe that online platforms were censoring lockdown skeptics and COVID-19 vaccine critics at the behest of the Biden Administration. Never mind that the censorship was already well underway by mid-2020, while Donald Trump was still president. Indeed, the Twitter account of the Zero Hedge website was already suspended at the end of January 2020 for posting an article on a potential lab origin of the new coronavirus—a hypothesis President Trump endorsed.
In any case, thanks to the First Amendment, the Biden Administration never had any stick with which to threaten the platforms. It could ask them to remove posts or suspend accounts, but the platforms could always decline.
The European Commission, however, was not just making occasional, informal requests. Under its “Fighting COVID-19 Disinformation Monitoring Programme,” all the major online platforms and search engines were required to report, first monthly and later every other month, on their efforts to suppress such “disinformation.” Twitter, for example, complied by submitting detailed statistics on “global” content removals and account suspensions. This meant that content was being removed and accounts suspended all around the world, to satisfy the European Commission.
There was no need for anyone to reveal private communications with E.U. officials to find out about this. The program was public, and the companies’ reports on their censorship activities were made available in a public archive, even if almost no one paid attention. (For more, see “The EU Files: What Elon Musk Is Not Telling You About Twitter Censorship,” by Robert Kogon, published by the Brownstone Institute, February 22, 2023.)
Moreover, the Commission did have something with which to threaten the platforms: namely, the impending DSA fines. By June 2020, the tech companies had had ample time to read the legislation and knew what was coming their way.
Some of them, including Google, Facebook, and Microsoft, having already been hit with fines in the hundreds of millions or even billions of euros by the Commission, were not going to take the Commission’s expectations lightly in the context of COVID-19. It is true, of course, that they were even less likely to do so after January 2021, with a new Democratic administration that, far from defending free speech, was asking them to do precisely what the E.U. was demanding.
The censorship reached a peak in summer of 2022 with a massive purge of thousands of COVID-dissident Twitter accounts, which left the owners of surviving accounts, like the American COVID-19 vaccine critic Alex Berenson, wondering “what is going on.”
But there ought not to have been any mystery. On July 5, the European Parliament passed the Digital Services Act—to the almost total indifference of the news media and public, in Europe as much as the United States. In combination with the Commission’s rollout three weeks earlier of a “strengthened” Code of Practice on Disinformation—consisting of a long list of far more stringent commitments than the original Code—it was inevitable that passage of the legislation would prompt swift and aggressive action on the part of the platforms. The DSA had arrived. Censorship was becoming law.
Censorship Goes Underground
Today, with Donald Trump having returned to the White House and the owner of X serving as his special advisor, Americans might well imagine that online censorship has already been defeated and free speech restored. But executive orders prohibiting federal censorship do nothing to accomplish this, since the American government was never able to require censorship from the tech companies in the first place. Complaints about a mythical “censorship-industrial complex” distract us from the reality of the E.U. regulatory regime and its transnational success in getting American companies to curtail Americans’ free-speech rights. Furthermore, Elon Musk is also complying with the E.U. censorship regime by curtailing Americans’ free-speech rights. Censorship has not disappeared from Twitter since his purchase and rebranding of the company as “X,” it has merely gone underground.
Account suspensions and content removals—the blunt instruments of censorship employed against COVID-19 “disinformation”—have indeed largely disappeared and seem to be used only in exceptional cases, presumably involving, as a rule, actual or alleged illegality. But the DSA also allows platforms to meet their obligations by way of “demotion” or “restriction of visibility.” In the same spirit, the 2022 “strengthened” Code of Practice commits platforms to mitigating the risk of “viral propagation” of disinformation by taking measures to reduce the latter’s “prevalence, views, or impressions.”
This, of course, is the essence of Elon Musk’s “freedom of speech is not freedom of reach,” a formulation that has allowed Musk to promote X as a “free speech platform” while still satisfying his obligations to censor “misinformation” and other allegedly harmful speech under the DSA. X does not need to remove such content to satisfy the E.U. It is enough, as X CEO Linda Yaccarino put it in a 2023 interview with CNBC, to make it “extraordinarily difficult to see.”
Misinformation “safety labels” are indeed built right into the published part of the X algorithm. These are not the famous “Misleading!” labels that would be prominently displayed on alleged COVID-19 misinformation under the old Twitter regime and that would lead to account suspension after “three strikes.”
Like the old “Misleading!” labels, the safety labels limit the shareability of posts to which they are applied, thereby satisfying E.U. demands. But unlike the “Misleading!” labels, the “safety labels” are not visible to the public. They are back-office labels that are only visible to platform administrators.
This means that while alleged “misinformation” is still being censored on X, unlike under the old regime we do not even know what the platform is treating as misinformation. The content in question simply disappears from public view without a trace. Since the algorithm can ensure that no one or hardly anyone is seeing them in the first place, this also eliminates the need to suspend the offending accounts.
As was true under the Fighting COVID-19 Disinformation program, platforms and search engines are required to submit publicly available reports quantifying their “content moderation” efforts under the DSA. Per X’s April 2024 “DSA Transparency Report,” during roughly just the prior five months the platform took enforcement action on 226,350 of 238,108 items reported to it by E.U. member states or the European Commission—or fully 95% of the reported items. Of the items, 40,331 were deleted and access to 62,802 items was blocked in the E.U. This means, however, that 123,217 items, over half, “merely” had their visibility restricted.
To appreciate the vast extent of the stealth censorship occurring on X, one must keep in mind that the above statistics are limited to content directly flagged by E.U. governmental authorities. They do not include content singled out by individuals, entities, or government-appointed “trusted flaggers.” Above all, they do not include the undoubtedly far larger swath of discourse proactively suppressed by X’s own automated systems or human content moderators to remain in the E.U.’s good graces.
X’s DSA Transparency Report specifies, incidentally, that 1,535 of the then-1,726 members of the platform’s content moderation team—or nearly 90% of them—spoke English as their main language. Nothing could make the overwhelmingly extra-European impact of the DSA more obvious. Post-Brexit, barely 1% of the E.U.’s own population speak English as their native language.
Lest readers have trouble reconciling the foregoing with the highly publicized proceedings against X that were already opened by the European Commission in December 2023, it should be noted that these proceedings, as they now stand, have nothing to do with the platform’s “content moderation” but only with other, more arcane aspects of the DSA. The original proceedings did indeed involve content moderation and could even have had a positive impact on freedom of speech, since X was supposed to be investigated not for failing to suppress user content, but rather for failing to inform users about it. But this aspect has been dropped.
An American Anti-Digital Services Act
So, what does the Trump Administration need to do to defeat E.U. censorship and restore genuine free speech to the internet? The First Amendment’s prohibition on Congress making any law “abridging the freedom of speech” was sufficient to secure free-speech rights for two centuries. But in 2025, when so much speech is online and foreign governments can thus make laws abridging Americans’ speech rights, it is not. This is what Germany did in 2017 when enacting NetzDG and what the E.U. did, to far greater effect, in 2022 when enacting the DSA.
But by what right should Germany or the E.U. be allowed to dictate what Americans can or cannot say? If the German government wants to censor Germans or the European Commission Europeans, that is one thing. But the American government has to make clear that if a foreign government wants to censor Americans or require American companies to do so on their behalf, then the U.S. will take action to defend its citizens’ rights.
Given that American companies are affected, the World Trade Organization could be one venue for challenging the DSA. The aim should be the outright abrogation of the DSA fines: no American company should be subject to fines or any other penalties from a foreign government for respecting Americans’ constitutionally guaranteed rights.
If, however, the E.U. refuses to back down, then the U.S. will need to make it illegal for American companies to cooperate with the E.U. and E.U. member-state governments—or any foreign government—in restricting Americans’ speech rights. The First Amendment prohibits Congress from making any law abridging the freedom of speech. But now, due to the DSA, it needs to enact a specific law protecting freedom of speech against foreign interference.
Such a law should give federal authorities the same draconian powers that the DSA gives the European Commission, but now in the cause of protecting speech rather than suppressing it. If the E.U. is going to fine American companies for respecting Americans’ free-speech rights, then the U.S. government is going to have to fine them for not doing so.
But the power to apply ruinous fines is not the only extraordinary enforcement power that the DSA gives the Commission. Under Article 69, it also gives it the power to conduct what are known as “dawn raids” in the case of suspected non-compliance: i.e., to have investigators break into and seal off company premises, inspect books or records in whatever form, and take away copies of or extracts from whatever books or records they deem relevant to their investigation. The U.S. government needs to have analogous search-and-seizure powers: to prevent the companies from cooperating with the Commission and/or E.U. member state governments. Federal investigators could thus find out exactly what communications the companies are having with the latter. They could find out, for instance, exactly what posts were targeted by the Commission and E.U. member states in the nearly 100,000 reports of “illegal or harmful speech” recorded in the above-cited data from X. If those posts constitute protected speech under American law, then X’s removal of them or restriction of their visibility would constitute a crime.
They could also find out exactly whose user information X has turned over to E.U. member state authorities in connection with such speech. Remarkably, the X data shows that there were nearly 3,000 such requests—or, more precisely, “orders” per Article 10 of the DSA. Nearly 90% of them came from Germany. If the U.S. government may not demand the information of users suspected of speech crimes or other wrong-speak, then surely American companies should not be permitted to provide Americans’ user information to foreign governments on such grounds. This too should be a crime.
Finally, the DSA gives the Commission the all-important power to demand access to social media and search engines’ algorithms (Articles 40, 69, and 72). The Commission has even hired its own programmers and established its own “European Centre for Algorithmic Transparency” (ECAT) to “support the enforcement of the Digital Services Act.”
U.S. investigators should have the same access to platform algorithms—to prevent DSA compliance. Any evidence of tweaking algorithms to satisfy the E.U.’s conceptions of “right-speak” and “wrong-speak” or “correct information” and “misinformation” should lead to sanctions. American online platforms should be neither vehicles of foreign propaganda (via algorithmic amplification) nor enforcers of foreign censorship (via algorithmic suppression).
If the E.U. still insists on its DSA powers when faced with an American law specifically prohibiting American companies from cooperating with the E.U. to censor Americans—or indeed censor any speech available to them—then the companies will simply have to choose. They can either be on the American market or the E.U. market. Or if they want to remain on both, then it will be up to them to find a technical modus vivendi that allows them to comply with both American and E.U. law: for example, by geo-blocking content in the E.U. If the implementation of this solution is financially onerous, as it undoubtedly would be, that is their problem. They can leave the E.U. market and avoid the costs. But censoring Americans’ speech to meet E.U. requirements would no longer be an option.
One thing Republican lawmakers should certainly not do is try to withdraw Section 230 protection from the tech companies as some kind of “punishment” for online censorship. This idea is based upon a fundamental misconception of the origins of online censorship and would only strengthen the hand of the enemies of free speech in the E.U.
The aim of America’s anti-Digital Services Act should be to restore the primacy of American law for Americans. But if freedom of speech is the good that the founders held it to be, then all users of the internet stand to benefit. If the European Union wants to build a new informational Iron Curtain, then that is its business. There is no reason for Americans to be held captive behind it.
