What the EU's content-moderation database actually discloses, and what it doesn't

The DSA Transparency Database is a public, machine-readable record that the European Commission built under Article 24(5) of the Digital Services Act to collect every "statement of reasons" that online platforms issue when they remove or restrict a piece of user content. It began accepting submissions from the largest platforms in September 2023, and a Commission-backed analysis of a single 24-hour snapshot found 57.4 million such statements filed on July 31, 2024 alone — a scale that has not translated into clarity about what is actually being taken down.

The database exists because the DSA requires it, not because platforms volunteered it. Article 17 of the regulation obligates any hosting service to give an affected user a "statement of reasons" whenever it removes, demonetizes, or restricts visibility of their content. Article 24(5) then requires online platforms — a narrower category of hosting services that both store and publicly disseminate user content, such as social networks, marketplaces, and app stores — to forward a stripped, non-personal version of every one of those statements to the Commission's public repository. General hosting services that don't disseminate content publicly are exempt.

What does a statement of reasons actually contain?

Each statement records the type of restriction applied — removal, demotion, demonetization, or account suspension — along with the legal or terms-of-service ground cited, the facts and circumstances behind the decision, and whether the platform used automated means to detect or decide the case. Providers must strip personal data before submission. The European Commission's own guidance describes the required fields but does not commit platforms to a uniform level of detail, which is where the disclosure starts to thin out.

Very Large Online Platforms — services with more than 45 million monthly users in the EU — were the first required to report, starting when the database launched. Smaller online platforms, apart from micro and small enterprises, had to begin submitting by February 17, 2024. The database organizes the incoming statements into daily files, and the Commission has said it plans to keep adding analytics and visualization tools; the underlying dashboard code is published on GitHub.

Why do researchers say the categories obscure more than they reveal?

An October 2025 report from the Information Technology and Innovation Foundation examined the database's own category system and found that nearly 90 percent of statements were filed under a single label — "scope of platform service" — which the report called incredibly vague because it bundles violence, nudity, hate speech, and misinformation into one bucket without differentiation. A statement tagged that way tells a researcher that something was removed, but not what.

The same analysis flagged three further gaps. The database does not distinguish whether a takedown removed content only for EU users or globally, so a decision driven by European law can look identical to one that erased content for users elsewhere. It also does not separate "illegal" content, which has a legal definition a court could test, from "harmful" content, which is a platform's own subjective judgment — collapsing the two into one combined category. And it does not indicate whether a given removal enforced EU law or simply the platform's private terms of service, a distinction that matters for anyone trying to measure the DSA's actual legal reach versus ordinary platform housekeeping.

What would fix the disclosure gap?

The same report proposed four specific reforms: retire the catch-all "scope of platform service" category in favor of more granular labels, split "illegal or harmful speech" into two distinct fields, require platforms to flag whether a removal was grounded in EU law or in their own terms, and mandate disclosure of whether an EU-triggered removal was applied globally. None of these changes require a new law — they would mean amending the database's data-submission schema, which the Commission controls administratively.

The gap between volume and clarity is the central tension in how the DSA's transparency mandate is playing out in practice. The regulation succeeded at compelling platforms to report content-moderation decisions at a scale no prior transparency regime had achieved — tens of millions of statements a day, filed by services from Amazon to Snapchat, searchable by the public. What it has not yet forced is disclosure detailed enough for an outside researcher to reconstruct what any single removal was actually about, which is the harder second half of transparency regulation: not just requiring platforms to speak, but requiring them to say something specific.

FAQ

• Who has to submit statements of reasons? Online platforms that store and publicly disseminate user content — social networks, marketplaces, app stores. General hosting or cloud services that don't disseminate content publicly are not covered.
• Does the database name individual users? No. Providers are required to strip personal data before a statement is submitted to the public database.
• Can the public download the raw data? Yes. The database supports search, summary dashboards, and full dataset downloads through its public interface.
• Is the vague-category problem a violation of the DSA? The report's authors frame it as a design shortcoming in the database's schema, not an allegation that platforms are breaking the law by using the categories the Commission itself defined.

For a related media news perspective, read British Royal music featured on Gossip Stone TV reality TV show by Debbie Wingham.