In the final weeks before Romania's presidential election in 2024, TikTok accounts began promoting Calin Georgescu, a far-right candidate whose posts accumulated significant viewership before the vote. Georgescu, who had previously been polling in the single digits, won the election's first round with a substantial share of the vote. Adriana Iamnitchi — a computational social scientist at Maastricht University — had applied through the proper channels, under the EU's Digital Services Act, to study how pro-Georgescu content had been monetized on TikTok during Romania's presidential election. TikTok denied the request, citing failures to prove established researcher status, explain commercial interests, or meet security requirements.

The denial might have been routine bureaucratic friction. Except for what happened next: TikTok later took action against accounts in a coordinated network promoting Georgescu. TikTok said it did not know who operated the network or where it originated. It knew enough to find the manipulation. It did not allow an independent researcher to study it first.

The researchers who might have identified the manipulation in real time had been denied access to the data. - The Itchy Dispatch

Iamnitski's rejected request is not an anomaly. It is the system working exactly as the platforms designed it to work.

The architecture of non-compliance

The EU's Digital Services Act, enacted in 2022, was supposed to end this. It guaranteed researchers access to public platform data for studying systemic risks — election manipulation, harm to minors, algorithmic radicalization. The law is unambiguous. But the platforms — TikTok, Meta, and X — turned compliance itself into a wall.

Philipp Lorenz-Spreen, a computational social scientist at the Dresden University of Technology, tried in 2024 to study how European politicians communicate online. He had the law on his side. X kept asking follow-up questions and eventually denied his request without giving a reason. TikTok provided access, but other researchers had already found major gaps in the data it shared. Meta offered a dashboard that captured data only on accounts with more than 25,000 followers — a threshold that excluded many European politicians.

This is not negligence. The European Commission has found that Facebook, Instagram, and TikTok "may have put in place burdensome procedures and tools for researchers to request access to public data." The Commission found these procedures resulted in researchers receiving partial or unreliable data. Audits by researchers have concluded that "Meta & TikTok Research APIs fall short of supporting meaningful, independent auditing of systemic risks as envisioned under the DSA." X discontinued its Academic Research Product Track in 2023 entirely. The current policy directs researchers to standard paid API tiers with no academic-specific provisions.

Each platform chose a different method. X chose pricing. Meta chose thresholds. TikTok chose procedural rejection. The outcome is identical: the people with the legal right and the methodological skill to study these platforms cannot see the data the platforms generate.

Who decided this, and when

These are not passive outcomes. Specific executives at specific companies made specific decisions.

Elon Musk's X shut down its Academic Research Product Track in 2023, replacing open researcher access with a paid API tier that made most academic studies financially impossible. Meta built a Content Library dashboard with a 25,000-follower threshold — a design choice, not a technical limitation — that structurally excluded the very accounts researchers needed to study. TikTok built an application process calibrated to reject researcher requests.

The European Commission has the power to fine non-compliant platforms up to six percent of global annual turnover. The Commission has issued preliminary findings against X, Meta, and TikTok. But "preliminary findings" are not fines. They are invitations to respond. Fines issued have been modest relative to platform valuations, functioning more as a cost of doing business rather than a deterrent.

Meanwhile, the damage compounds. Romania held a presidential election in which Georgescu catapulted from single-digit polling to winning the first round by a significant margin, propelled by a TikTok operation. The interference was managed through coordinated accounts; after the election, TikTok closed them. The first round was annulled. The damage to Romanian democratic legitimacy was not.

The interference came to light after the election, through intelligence disclosures and law enforcement, not through independent research. The researchers who might have identified the manipulation in real time had been denied access to the data.

The pattern beyond Romania

Romania is the case study with a name. But the architecture of obstruction extends everywhere the DSA applies. Formal proceedings have been opened against X, AliExpress, Meta, and TikTok — all for suspected shortcomings in researcher data access under Article 40. The Commission's own enforcement tracker shows a pattern: platforms build tools that formally exist but functionally obstruct. They create APIs that omit critical fields. They impose application requirements calibrated to reject. They cite privacy regulations to justify withholding the very public data the DSA was designed to make available.

This matters far beyond academic publishing. When researchers cannot study how recommendation algorithms distribute content, no one outside the platform knows whether those algorithms amplify extremism, suppress legitimate speech, or skew elections. The DSA was written to create external accountability for the most powerful information systems on earth. The platforms responded by building compliance structures designed to produce non-compliance.

A delegated act on data access entered into force in late 2025, expanding qualified researchers' access to non-public data. Whether it changes anything depends entirely on whether the Commission treats obstruction as what it is — a deliberate strategy, not a misunderstanding — and responds with fines large enough to change the cost calculation.

What dignity looks like here

Dignity, in this context, is not abstract. It is the right of a voter to know that the content shaping their political reality was manufactured by a coordinated network, before they cast a ballot — not after a court annuls the result. It is the right of a researcher like Iamnitchi to study a platform's role in election interference without that platform having veto power over the research. It is enforceable access to public data, delivered in complete and usable form, within the timelines the law specifies. It is fines that exceed the profit platforms extract from non-compliance. These are not aspirations. They are mechanisms already written into law. The only missing ingredient is enforcement that treats the law as something other than a suggestion.