Transparency reports describe measures, policies and indicators that signatories say they use to address disinformation. They record what companies report, but do not by themselves prove that those measures work. To interpret the figures, readers need to know which services, regions and periods they cover, and whether the data are complete, comparable and open to verification.
What the 2024 study examined
Published in July 2024 in Internet Policy Review, a study examined Google’s, Meta’s, Microsoft’s, TikTok’s and Twitter’s—now X’s—reporting obligations under the European Union’s strengthened Code of Practice on Disinformation. The researchers assessed the platforms’ initial reports, published in early 2023. Their findings concern those reports and that period; they do not automatically describe later company practices.
The authors judged compliance with reporting obligations to be partial overall. They found that qualitative information often lacked detail or relevance, while quantitative data were in several cases missing, incomplete or insufficiently robust. The study also identified statements the researchers considered doubtful or, in their assessment, contradicted by earlier evidence. Reporting gaps constrain what can be concluded about how the Code was applied in practice.
The research assessed report quality and the Code’s monitoring system. A company-reported figure describes, first of all, what that company reported; it is not automatically an independent measurement of whether disinformation declined.
What changed under the Digital Services Act
The regulatory framework changed after the period covered by the study. On 13 February 2025, the European Commission and the European Board for Digital Services approved integrating the voluntary Code into the Digital Services Act (DSA). From 1 July 2025, the recognised Code became part of the DSA’s co-regulatory framework and a reference point for its signatories’ commitments.
The Commission says that very large online platforms and very large online search engines that subscribe to the Code commit to report every six months on measures taken on the services covered. These categories include services with at least 45 million monthly users in the EU. Other signatories that are not platform providers report annually. Reporting frequency depends on the type of signatory.
The integration also links the Code’s commitments to DSA oversight mechanisms. For signatories in scope, compliance with commitments is subject to an independent annual audit; the Code can also serve as a relevant reference when assessing legal risk-mitigation obligations. These mechanisms strengthen accountability, but evaluating any report still depends on its scope, comparability and data quality.
Reports for the second half of 2025
In March 2026, the Commission described a reporting round covering 1 July to 31 December 2025. It includes reports from Google, Meta, Microsoft and TikTok, as well as contributions from fact-checkers, research and civil-society organisations, and representatives of the advertising industry. This is not the reporting round examined in the 2024 study.
The Commission says the reports address the implementation of anti-disinformation measures, changes to policies, tools and partnerships, ongoing crises, and election integrity. They describe actions signatories say they took and the topics they addressed. When reading them, it helps to distinguish reported activity, indicators and independent assessments.
Meta’s report for the second half of 2025 covers Facebook, Instagram, Messenger and WhatsApp in the European Economic Area, although some indicators have a global scope. Meta says it removed more than six million Facebook and Instagram ads during the period for violating its misinformation policy. That figure concerns ads and a specific policy; on its own, it does not measure the spread of disinformation or the effect of the intervention on people.
Meta also describes labels for ads created or significantly edited with its generative AI tools, media-literacy initiatives, and action against coordinated influence operations. These are activities reported by the company, not independent assessments of their effects. The report specifies the services, region and period to which its information applies.
How to read indicators: activity is not impact
A moderation indicator may count ads, accounts or content on which a platform took action. Its meaning depends on how the category is defined, the period and territory covered, and the counting method. More interventions do not automatically mean better results. To assess a figure, readers need to understand what problem it measures and whether it can be compared across services or periods.
It is useful to distinguish three levels: measures a company says it implemented; indicators describing its activities; and claims that the spread or effects of disinformation may have fallen. Reports offer the clearest view of the first level and may provide clues about the others. Assessment requires understandable methods and oversight that makes the data open to scrutiny.
The timing matters, too. The academic study examined reports published in early 2023 and found partial shortcomings in compliance and information quality. The reporting round described by the Commission in March 2026 covers the second half of 2025, when the Code was already integrated into the DSA. The 2024 findings cannot automatically be applied to later reports. Detail, relevance, completeness and robustness remain important criteria.
The reports can therefore help readers understand which measures platforms say they take and how they describe them. Determining whether those measures reduce disinformation requires intelligible indicators, comparable data and independent checks.