The EU Digital Services Act: what it actually requires
Blog · censorship

The EU Digital Services Act: what it actually requires

Statements of reasons, trusted flaggers, systemic risks and the crisis clause, read from the primary text, plus the documented clashes over X, TikTok, Germany's trusted flaggers and the US visa bans

Series: How minds are moved · Censorship, enforced and invisible · 4 of 7

On 5 December 2025 the European Commission fined X €120 million, the first penalty ever issued under the Digital Services Act. President Trump called it “a nasty one”; Vice President JD Vance and Secretary of State Marco Rubio called it censorship. A Commission spokesperson answered: “Our digital legislation has nothing to do with censorship.” Less than three weeks later the US State Department denied visas to five Europeans it accused of censoring Americans, among them Thierry Breton, the former commissioner who had steered the law.

What was the fine actually for? Not for any post, opinion or political view. It was for X’s paid blue checkmarks, its advertising archive and the barriers it put in front of researchers. That gap, between what the law says and what each side says it does, is why this post goes to the primary text.

Contents · 8 sections

TL;DR #

  • The Digital Services Act (Regulation (EU) 2022/2065) applies to every online platform in the EU since 17 February 2024, and to the biggest ones, those with more than 45 million monthly EU users, since August 2023.
  • It does not define what speech is illegal. National and EU laws do that. The DSA sets procedures: notice and action, reasons, appeals, transparency.
  • The largest platforms must also assess and reduce “systemic risks”, which include effects on “civic discourse and electoral processes” and on freedom of expression itself.
  • Fines can reach 6% of global annual turnover.
  • Supporters say it replaces secret phone calls with public rules and appeal rights. Critics say “systemic risk” and “trusted flaggers” give governments a lever over lawful speech. Both sides cite the same articles.

What it requires, article by article #

For every hosting service and platform:

  • Article 16, notice and action. Anyone can report content they think is illegal. The platform must process the notice.
  • Article 17, statement of reasons. If a platform removes your post, demotes it, demonetises it or suspends you, it must tell you why, and on what basis (law or terms of service). These statements go into a public EU database.
  • Article 20, internal complaints. You can appeal for six months after the decision.
  • Article 21, out-of-court dispute settlement. Certified independent bodies can review moderation decisions. One, the Appeals Centre Europe in Dublin, was certified in 2024.
  • Article 22, trusted flaggers. National regulators can certify organisations whose notices get priority. They must be independent of platforms and publish yearly reports.
  • Article 28. No targeted ads to minors based on profiling.

For very large platforms and search engines only:

  • Article 34, systemic risks. Each year they must assess risks from their services: the spread of illegal content; negative effects on fundamental rights, including freedom of expression and media pluralism; negative effects on civic discourse, elections and public security; and harms related to gender-based violence, public health, minors, and people’s physical and mental well-being.
  • Article 35, mitigation. They must take “reasonable, proportionate and effective” measures, which can include changing algorithms, terms and moderation.
  • Article 36, crisis response. In a crisis, on a recommendation from the Board for Digital Services, the Commission can require specific measures for up to three months, which can be extended.
  • Articles 37 to 40. Independent audits, a non-profiled recommender option, public ad repositories, and data access for vetted researchers.

The clashes, as documented #

X: from proceedings to the first fine #

The Commission opened formal proceedings against X on 18 December 2023, covering illegal content, information manipulation, Community Notes, the blue checkmark, ad transparency and researcher access. Preliminary findings came on 12 July 2024. On 12 August 2024, before Musk’s live interview with Donald Trump, Commissioner Breton sent Musk a public letter warning about “amplification of harmful content”. The Commission’s own spokespeople said the letter had not been agreed with the Commission President. Breton left the Commission the next month.

The December 2025 fine covered only the transparency parts: the checkmark as “deceptive design”, the ad archive, and researcher access. The parts about content and Community Notes remained open. X appealed on 16 February 2026. On 16 July 2026 the Commission accepted X’s plan to fix the ad archive and researcher access, with six months to implement it and an external audit after that, while the appeal continued. The Board for Digital Services had found the plan only partially adequate, and the blue checkmark is not part of it. The X fine was the first under the DSA but not the last: the Commission later fined Temu €200 million (May 2026) and AliExpress €550 million (July 2026), both over illegal and unsafe products, not speech.

TikTok and Romania #

After Romania’s Constitutional Court annulled the presidential election in December 2024, citing a coordinated TikTok campaign, the Commission opened proceedings against TikTok on 17 December 2024 over election risks. To supporters, this was the DSA working as intended. To critics, it showed a law that could be used to second-guess an election result. Two years on, the question still splits Romania. On 10 November 2026 President Nicușor Dan said the promised report on the annulment was not ready. On LSN News, my news engine, the same press conference produced two opposite headlines: one outlet, gandul.ro, wrote that he “admits there is no new evidence of Russian influence”; another, capital.ro, wrote that “there was evidence of Russian influence” and undeclared campaign spending (https://lsn-news.shuffle-on.com/s/669661). He had said both things: evidence was presented a year ago, and nothing new has come since. Each outlet chose its half.

Trusted flaggers in Germany #

In autumn 2024 Germany’s regulator certified its first trusted flagger, REspect!, an organisation that reports online hate. The regulator’s head said illegal content, hate and fake news could now be removed quickly. Critics read that as giving “fake news” a removal track it does not have in law. He later said the word “illegal” applied to the whole list. In 2025, according to documents made available to Euronews, REspect! is about 95% government-funded; Euronews also reported that it passes some reports directly to federal police, which the DSA does not require. The government said the DSA only requires independence from platforms, not from the state.

The transatlantic dispute #

In July 2025 the US House Judiciary Committee published a staff report calling the DSA a “foreign censorship threat” to Americans’ speech. In December 2025 came the visa bans. Breton called it a “witch hunt”. The Commission said it “strongly condemns” the bans.

The arguments, from both sides #

Supporters point to the text. The DSA gives users rights they never had: a reason for every decision, an appeal, an independent reviewer. It forces transparency on algorithms and ads. Article 34 explicitly lists negative effects on freedom of expression among the risks platforms must assess. It does not make disinformation illegal, and none of the fines so far has been about speech: X’s was about deceptive design and transparency, Temu’s and AliExpress’s about illegal products.

Critics also point to the text. “Negative effects on civic discourse” is not defined, and platforms facing a 6% fine have every incentive to over-remove. Trusted flaggers can be state-funded. The crisis mechanism lets the Commission, on the Board’s recommendation, require measures during an emergency. Breton’s August 2024 letter, sent before a US election interview, is their evidence that officials will use the law’s weight beyond its letter.

What I can say from the record is that both readings rest on real clauses. How it is used will decide which reading wins.

What it feels like from the inside #

For an EU user it can be good news: the next time a post disappears, you should get a reason and an appeal path. In practice, statements of reasons range from a specific explanation to a template with a checkbox. For a platform, it feels like a regulator in the room for every product decision. For a user in the US, it can feel like Brussels setting global rules, because platforms often apply one policy worldwide.

How to notice it #

  • Read your statement of reasons. It should say whether the decision was based on law or on the terms of service. If it is law, it should name it.
  • Search the DSA Transparency Database. It shows billions of moderation decisions by platform and ground.
  • Check who flagged. Trusted flaggers are listed publicly per country, with their funding in their annual reports.
  • Separate the fines from the rhetoric. For any DSA headline, find the Commission decision and see which articles it cites.
  • Watch Article 36. If a crisis measure is triggered, it will be public; that is the clause with the widest power.

Frequently asked questions #

Does the DSA make “disinformation” illegal? #

No. It requires the largest platforms to assess risks, including disinformation’s effect on civic discourse, and to mitigate them. Only for content illegal under national or EU law does a notice put the platform at risk of liability if it does not act.

Can the EU order a platform to delete my post? #

Not directly under the DSA. Orders against specific illegal content come from national authorities or courts under their own laws (Article 9). The Commission supervises systems, not individual posts.

Was X fined for its content moderation? #

No. The December 2025 fine was for the blue checkmark, the ad repository and researcher access. The content-related parts of the investigation were still open as of the latest reports I found.

Does the DSA protect free speech at all? #

It contains protections: statements of reasons, appeals, out-of-court review, and freedom of expression as a risk platforms must assess. Whether those protections outweigh the pressure to remove is the core of the debate.

The lesson #

The DSA is neither the censorship machine its critics describe nor the neutral rulebook its authors describe. It is a set of procedures with some broad clauses in the middle. The first fine was narrow and technical; the first public letter by a commissioner was broad and political. If you want to judge it, read Articles 17, 22, 34 and 36 yourself, then watch what the Commission actually decides, not what anyone says it decided.