Skip to main content
European data
data.europa.eu
Uradni portal za evropske podatke

How open are Europe's charity registers and what does this mean for AI?

Topic:

Sharing a data reuse use case: GiveRadar 

If you ask an AI assistant whether a charity is legitimate, you would expect it to check an official government register. After all, that's where you should be able to find out whether an organization is registered, whether it's still active, and how it manages its finances. 

Unfortunately, it's not always that simple. Official charity registers differ quite a lot between countries. Some provide detailed information that AI agents can easily access, while others barely publish anything at all. To understand how big these differences actually are, we researched 76 national charity registers for our Global Charity Register Transparency Index 2026. 

We assessed each country across seven dimensions, including whether a public register exists, whether financial filings and board information are available, and whether the data can be accessed and processed by machines. The results show some pretty significant differences. 

Only six countries achieved a perfect score of 100/100: Australia, Estonia, Ireland, Latvia, New Zealand and the United States. Germany scored just 49, Austria 39, and Spain 38. In 13 countries, the public has access to almost no information at all. 

And this becomes particularly interesting when looking at how AI agents can use public data. At GiveRadar, we've been working on making official charity register information more accessible. We currently combine government registers and tax authority data for more than 7.9 million nonprofits across 100+ countries. We've also made this information accessible to AI agents through a Model Context Protocol (MCP) server. This allows an agent to search for an organization and retrieve its registration status, available financial filings, and the official sources behind that information. But there's an important limitation: an AI agent can't verify information that the official register doesn't publish. Even with more advanced AI models, we can't solve a lack of public data. If a country doesn't publish financial information, board details, or reliable registration identifiers, there is simply less information available to verify an organization. Our Transparency Index was researched using AI agents following a published methodology, with an independent audit and human editorial oversight. We only included countries in the final ranking when every dimension could be verified through an official source. 

The complete dataset is publicly available under a CC BY 4.0 license, so anyone can explore the findings or use the data for further research: https://doi.org/10.5281/zenodo.21804904 You can read the full results here: https://giveradar.com/research/charity-register-transparency-index-2026/ 

I think these findings raise an important question for the European open data community. We're putting a lot of effort into making AI systems more reliable, but shouldn't we also be focusing more on making the official data they depend on publicly accessible and readable by machines? 

I'm curious to hear how others working with European public registers experience this. Are there particular countries or registers where accessing reliable information remains a challenge? 

Matt Timmermans (GiveRadar)