Opinion: The Data Black Hole: Big Tech's War on Independent Research
AI-generated image · Bay Street Wire
Under the guise of security and policy, social media giants are systematically blocking the data access required to hold their algorithms accountable.
The Digital Services Act (DSA) was designed to pull back the curtain on Big Tech, granting vetted researchers access to the data necessary to study systemic risks. But as my latest roundup reveals, the giants aren't just ignoring the law—they are actively sabotaging the independent research needed to hold their algorithms accountable.
**The Romanian Warning Sign** According to reporting from Ars Technica, the real-world stakes of this data stonewalling are immense. In the lead-up to Romania's November 2024 presidential vote, TikTok accounts previously dedicated to fashion and manicures suddenly began promoting Calin Georgescu, a politician known for anti-Semitic tropes and hard-line immigration views. His content reached 120 million views, helping him win the first round of the election with 23 percent of the vote before the results were later annulled.
Adriana Iamnitchi, chair of computational social sciences at Maastricht University, attempted to investigate how this content was monetized through hidden influencer marketing. Despite applying for API access under the DSA on October 28, 2025, TikTok denied her request. The platform claimed Iamnitchi failed to meet security requirements or prove her status as an established researcher. While TikTok later admitted to removing over 27,000 fake accounts in a coordinated network promoting Georgescu and the Alliance for the Union of Romanians (AUR), Iamnitchi argues that API access could have identified who actually profited from the campaign.
**A Strategy of Obfuscation** Researchers are finding that the implementation of the law is where the sabotage happens. Duncan Allen, a research officer at Democracy Reporting International (DRI), tells Ars Technica that platforms have systematically dismantled public access tools. Meta replaced CrowdTangle with content libraries, and X has paywalled its API data, forcing academics to pay hundreds of dollars monthly. TikTok has further hampered large-scale study by capping the number of posts researchers can pull daily.
Even when researchers attempt to follow the legal path, they hit a wall of impossible requirements. Iamnitchi notes that platforms often demand data be stored on infrastructure that cannot be compromised—such as machines physically disconnected from the internet—resources that most universities simply do not possess.
**The Numbers of Denial** Data from the DSA40 Collaboratory, a German initiative led by coordinator L. K. Seiling, highlights the inconsistency of these approvals. Out of 46 tracked applications, 20 were approved and 14 were rejected. The disparity between platforms is stark: TikTok approved 11 of 13 applications, while X rejected 11 of 23. Seiling warns that these numbers likely underrepresent the actual rejection rate, concluding that the current setup is designed to disincentivize researchers.
Even for the lucky few who gain access, the data is often unreliable. Iamnitchi points out that API data is frequently impossible for colleagues to reproduce, violating a basic requirement of scientific research. While a TikTok spokesperson told Ars Technica the company has granted access to over 1,500 research teams and approved 130 EU applications in late 2025, the reality on the ground is a growing "black hole" of information regarding how these platforms recommend content and manage sensitive reports.