European researchers want to study social media's harms but can't get the data
Despite the EU's Digital Services Act granting researchers access to social media data, academics say platforms are obstructing their efforts to study disinformation and systemic harms, as seen in Romania's 2024 election.
Intelligence analysis by Llama
EU's Digital Services Act was meant to let vetted academics study platform harms, but two years in, researchers describe a process of denials, narrow risk definitions, and impossible security demands. Romania's 2024 election is a case study.
Imagine if a school cafeteria kept all its lunch menus secret, and grown-ups who wanted to check if kids were eating healthy couldn't look. That's what's happening in Europe: laws say researchers should be able to look at what social media shows people, but the apps are making it really hard. So when something bad happens, like fake posts that change an election, nobody outside the company can figure out how it worked.
Analysis
The Romania Precedent
The 2024 Romanian presidential election has become a defining case for why researcher access to platform data matters. According to the article, TikTok accounts that had spent years posting about manicures and fashion abruptly pivoted to promoting Calin Georgescu, a fringe candidate whose hard-line immigration and anti-Semitic messaging was viewed 120 million times before the first round. Georgescu jumped from single-digit polling to 23 percent, a result later annulled by Romania's Constitutional Court. Declassified intelligence alleged the campaign was coordinated by Russia and that Georgescu benefited from preferential TikTok treatment. Adriana Iamnitchi, who leads disinformation research at Maastricht University, told WIRED her team applied for TikTok API access weeks before the vote and was denied. She believes access could have helped identify who was profiting from the operation. The episode crystallized for many researchers the gap between what EU law promises and what platforms deliver.
The Law Was Never the Bottleneck
The Digital Services Act explicitly allows vetted researchers at credible institutions to obtain platform data for studying systemic risks, ranging from illegal content to threats to fundamental rights. The problem, researchers told WIRED, is implementation rather than scope. Application processes vary by platform but typically require data to be stored on infrastructure that cannot be compromised, including machines physically disconnected from the internet, a resource most universities do not have. Iamnitchi described platforms' narrow interpretations of what counts as a systemic risk and the lack of reproducibility in the API data they do provide, a violation, she argued, of basic scientific norms. DSA40, a German initiative tracking applications, found that of 46 cases, 20 were approved and 14 rejected, with stark divergence between platforms: TikTok approved 11 of 13 applications, while X rejected 11 of 23. Coordinator L. K. Seiling said the current setup actively disincentivizes researchers from using the pathway at all.
Eroding the Public Window
The researcher access debate sits on top of a broader trend toward data restriction. Meta shut down its public tool CrowdTangle and replaced it with the Meta Content Library, which requires approval. X, formerly Twitter, paywalled its API, forcing academics to pay hundreds of dollars a month, according to Duncan Allen of Democracy Reporting International. TikTok caps researcher accounts at 1,000 daily API requests, which Allen said makes it impossible to study anything at scale. The cumulative effect, researchers warn, is the steady growth of what Iamnitchi called black holes in society's understanding of how platforms recommend content or handle reports of sensitive material. TikTok told WIRED it has given more than 1,500 research teams access to its tools and considers its program DSA-compliant, while Meta said CrowdTangle covered only a fraction of its public data. The platforms frame their efforts as compliant, but the empirical gap between what researchers can study today and what they could study five years ago is widening, and elections like Romania's show what that gap costs.
Key points
- EU's Digital Services Act grants vetted researchers platform data access, but two years on, researchers describe a system of denials, narrow risk definitions, and impractical security demands.
- Romania's annulled 2024 presidential election is the textbook case: 120 million views of pro-Georgescu TikTok content preceded a 23 percent first-round result later linked by declassified intelligence to Russian coordination.
- Maastricht University researcher Adriana Iamnitchi was denied TikTok API access weeks before the vote; she believes data could have identified the network's operators.
- DSA40 tracked 46 applications: 20 approved, 14 rejected, with TikTok approving 11 of 13 and X rejecting 11 of 23, illustrating stark platform-by-platform divergence.
- Platforms have simultaneously restricted public tools (Meta's CrowdTangle, X's paywalled API) and capped researcher quotas, narrowing the evidence base available outside the companies themselves.
If the European Commission enforces the DSA more strictly and issues clearer guidance on what counts as a qualifying research application, platforms could be compelled to lower security barriers and expand quotas. The DSA40 data shows approval is possible, and TikTok's high approval rate suggests at least one major platform is willing to cooperate. Stronger compliance would restore the independent evidence base regulators and courts need to evaluate platform harms.
Without enforcement pressure, platforms have little incentive to make data genuinely accessible, and the trend toward walled-off research environments will continue. The Romania episode shows the cost: researchers learned about the pro-Georgescu network only after Romanian intelligence declassified findings, not through their own API access. If this pattern holds, future influence operations will be documented by intelligence agencies and journalists rather than by independent academics, weakening the empirical foundation of platform regulation.
