Pierri, F., Araujo, T., Kruikemeier, S., Lorenz-Spreen, P., Abeele, M. M. V., Vandenbosch, L., Gonçalves-Sa, J., & Grabowicz, P. (2026). Research opportunities and challenges of the EU’s Digital Services Act. Communications of the ACM. https://doi.org/10.1145/3779412
Summary
This perspective piece examines the research opportunities and obstacles created by Article 40 of the EU’s Digital Services Act (DSA), a legal and technical framework enabling vetted researchers to access public and non-public data from very large online platforms (VLOPs) and search engines (VLOSEs). Written by researchers directly involved in EU policy discussions on the Article 40 Delegated Act (adopted July 2025), the authors argue that while the DSA is a landmark for platform transparency and accountability, significant legal, technical, and organizational barriers threaten its promise. They contend that the goals of researchers, citizens, governments, and platforms are frequently misaligned, that severe power and resource asymmetries favor corporate over academic research, and that practical bottlenecks—slow vetting, mismatched expectations, and a circular resource problem—could undermine effective data access. The piece is framed against a contrasting US trend toward reduced platform oversight.
Key Contributions
- Insider account of the practical challenges of implementing DSA Article 40 from researchers involved in the policy process.
- A structured typology of obstacles: misaligned incentives, power and resource asymmetries, and implementation bottlenecks.
- Three concrete recommendations: streamline access via standardized data-sharing frameworks and dataset categorization; support independent research to counter resource asymmetries; and close regulatory blind spots, notably around large language models.
- The argument that LLMs should fall within the DSA’s regulatory scope as systemic-risk-bearing intermediary services.
Methods
An informed perspective/opinion piece rather than an empirical study. It draws on the authors’ direct experience in EU policy discussions, an 18-month ERC-supported pilot program bringing together researchers, DG-CONNECT, and national Digital Services Coordinators (DSCs), and a roundtable with major platforms and the European Commission. It synthesizes prior literature and documented incidents to build its argument.
Findings
- The DSA identifies four categories of systemic societal risks: threats to democratic processes, fundamental rights, public health, and dissemination of illegal content.
- Platform representatives showed varying reluctance to direct data sharing, favoring alternatives such as an “independence by permission” model where platforms test researcher-supplied hypotheses.
- Industry-aligned research tends to produce platform-favorable outcomes; a cited example found Meta-collaboration research portrayed platforms as more effective at curbing misinformation, partly by omitting disclosure of temporary algorithmic changes during the 2020 US election.
- Data access requests can now be submitted through a dedicated EU portal, with first decisions expected in late February 2026.
- A “circular problem of resources” exists: data access is often needed to win grants, but without funding researchers lack the time and legal support to complete complex applications.
- Platforms leverage their own reach to frame transparency regulation as censorship, reaching tens of millions—an asymmetry academics cannot counterbalance.
Connections
This paper sits at the center of the platform-data-governance debate over researcher access and regulatory infrastructure, complementing work on data-access APIs and platform accountability such as Rieder2026-pp, Bruns2026-pn, and de-Vreese2026-zx. Its concern that LLMs constitute a regulatory blind spot connects to broader discussions of platform power and information-disorder oversight, while its critique of industry-aligned research and asymmetric platform reach resonates with studies of platform-shaped misinformation narratives like Lewandowsky2026-ob and Bak-Coleman2026-mk.
Podcast
A research-radio episode discusses this paper: Listen