Entrena-Serrano, C., Degeling, M., Romano, S., & Çetin, R. B. (2025). TikTok’s Research API: Problems Without Explanations. arXiv [cs.CY].
Summary
This paper offers an early empirical evaluation of TikTok’s Research API, which the platform expanded in Europe in July 2023 to satisfy the Digital Services Act’s (DSA) requirement that Very Large Online Platforms (VLOPs) facilitate data access for independent researchers. Testing the API in practice, the authors argue that it fails to deliver the transparency and accessibility the DSA envisions: persistent limitations, inconsistencies, and gaps undermine its utility for studying systemic risks. The broader claim is that VLOP compliance gestures often fall short of the DSA’s substantive transparency goals, making platform-provided access mechanisms an unreliable foundation for accountability research.
Key Contributions
- Provides one of the earliest empirical assessments of TikTok’s DSA-driven Research API expansion.
- Documents concrete shortcomings and inconsistencies in a platform-provided research access mechanism.
- Feeds ongoing policy and academic debate over what “adequate” researcher data access should look like under the DSA.
Methods
- Empirical, hands-on testing of access to TikTok’s Research API in the European context.
- Analysis of the API’s limitations, inconsistencies, and gaps measured against DSA transparency requirements (detailed methodology not available from the abstract).
Findings
- Despite the July 2023 expansion, the Research API continues to exhibit substantial limitations and inconsistencies.
- These shortcomings constrain researchers’ ability to study transparency and systemic risks on the platform.
- Formal compliance with the DSA does not translate into practical, reliable data access.
Connections
This work sits within the growing literature on the DSA’s data-access regime and the reliability of platform-provided research APIs, connecting to broader critiques of how VLOPs implement transparency obligations such as Rieder2025-ju and Rieder2026-pp. It also complements empirical audits of the data infrastructures researchers must rely on, including Bruns2026-yv and Murtfeldt2025-wu.