Empirically Measuring Data Localization in the EU
Alexander Gamero-Garrido, Kicho Yu, Sumukh Vasisht Shankar, Sachin Kumar Singh, Sindhya Balasubramanian, Alexander Wilcox, David Choffnes · 2025
Read Empirically Measuring Data Localization in the EU by Alexander Gamero-Garrido, Kicho Yu, Sumukh Vasisht Shankar, Sachin Kumar Singh, Sindhya Balasubramanian, Alexander Wilcox, David Choffnes (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-09019.txt.
Introduction
Alexander Gamero-Garrido and co-authors measure whether web data stays inside Europe as law intends in this 2025 study. European privacy rules restrict sending personal data to countries judged less safe. But the internet's routing systems were never built to respect borders, so real compliance is hard to check.
The team combined browser probes with network probes across more than a thousand networks in nineteen countries. They traced which servers and routers handled European requests and where those machines sit. On average, only a small share of servers sat in countries deemed unsafe, and the share was even smaller for known trackers. The authors conclude that providers mostly follow the policy, with some exceptions.
Catalog record
| CREATOR | Alexander Gamero-Garrido, Kicho Yu, Sumukh Vasisht Shankar, Sachin Kumar Singh, Sindhya Balasubramanian, Alexander Wilcox, David Choffnes |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 67 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-09019
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Empirically Measuring Data Localization in the EU Alexander Gamero-Garrido, Kicho Yu, Sumukh Vasisht Shankar, Sachin Kumar Singh, Sindhya Balasubramanian, Alexander Wilcox, David Choffnes Abstract . EU data localization regulations limit data transfers to non-EU countries with the GDPR. However, BGP, DNS and other Internet protocols were not designed to enforce jurisdictional constraints, so implementing data localization is challenging. Despite initial research on the topic, little is known about if or how companies currently operate their server infrastructure to comply with the regulations. We close this knowledge gap by empirically measuring the extent to which servers and routers that process EU requests are located outside of the EU (and a handful of “adequate” non-EU countries). The key challenge is that both browser measurements (to infer relevant endpoints) and data-plane measurements (to infer relevant IP addresses) are needed, but no large-scale public infrastructure allows both. We build a novel methodology that combines BrightData (browser) and RIPE Atlas (data-plane) probes, with joint measurements from over 1,000 networks in 19 EU countries. We find that, on average, 2.3% of servers serving users in each EU country are located in non-adequate destination countries (1.4% of known trackers). Our findings suggest that data localization policies are largely being followed by content providers, though there are exceptions. Volume: YYYY X DOI: XXXXXXX.XXXXXXX 1. Introduction Regulators around the world have taken various approaches at mitigating potential harms resulting from unfettered collection of user data at a large scale by Internet companies, usually for the purpose of targeted advertising. Data localization—broadly defined as the principle that requires user data to be stored in the jurisdiction where the user is physically located—is one such regulatory approach and a key component of the EU’s privacy landmark regulation enacted in 2018: The General Data Protection Regulation (GDPR). A central intent behind (GDPR) data localization is to prevent transfers of user data to jurisdictions with weaker privacy protections, as such transfers expose (EU) users to unlawful commercial surveillance of user data by both commercial advertisers and foreign government agencies, as was argued in the landmark Schrems decisions ( Union, 2015 ; Parliament, 2020 ) . In this paper, we explore the question of whether content providers serving users in Europe comply with data localization regulations by locating their servers in the EU (or a small number of third countries that the EU regulators have approved data transfers to ( Commission, 2022a ) ). In answering this question, we address a key technical challenge. Since data localization is based on physical boundaries, three types of information are simultaneously needed to audit compliance with data localization, and no existing platform nor methodology can procure all three: ( i ) Which domains are most frequently visited by European users, including many additional domains fetched automatically by user-loaded websites. ( ii ) The location in the network of all relevant domains, that is, their IP addresses. ( iii ) The physical location of the servers from which these domains are loaded. While ( iii ) is the key information needed to audit data localization practices, in practical terms it is difficult to obtain it directly from ( i ) domain names; rather, it is easier (though still challenging ( Katz-Bassett and Smaragdakis, 2021 ) ) to obtain the physical location of IP addresses, a necessary resource to connect to the Internet an actual server referenced by each domain. Our auditing framework on data localization compliance is built with three components, each obtaining the necessary information described above. First, we identify all popular domains in each EU country using queries from BrightData ( BrightData, 2023a ) , a proxy service that allows us to run a headless browser and therefore fetch complete web pages from EU-based devices. Second, we launch active measurements from RIPE Atlas ( Atlas, 2021 ) to identify the IP addresses serving each domain. Third, we use RIPE IPMap, ( Katz-Bassett and Smaragdakis, 2021 ) ) a database that translates IP addresses to a physical location (IP geolocation), to identify IP addresses that are potentially located in countries outside the EU. We confirm a subset of these as likely GDPR violations using active measurements and speed-of-light constraints. We validate our method using servers with known locations, finding no false positives. We find that the vast majority of servers responding to requests from EU users are located in the EU or adequate third countries. Neighbors of the EU, Russia and Turkey, are the most commonly observed destinations in non-adequate countries, besides the US. These servers primarily serve requests in Finland and Romania, respectively. We find smaller numbers of servers and known trackers in other non-adequate countries further away from the EU, primarily in Asia. News websites are the leading cause of these potential GDPR violations, as they are the most common type of site that loads trackers in non-adequate countries. We find evidence of tracking activity taking place in a non-adequate country using cookies as well. We also find significant differences across EU regions: Southern and Eastern EU countries are more likely to be served content by known trackers located in non-adequate countries when compared to Western and Northern EU countries. Combined, these findings suggest that Internet companies are likely inclined to host content in close proximity of users, which leads to by-default compliance with data localization regulations; however, there are exceptions where rigid technical constraints may force Internet operators to serve EU users from non-compliant jurisdictions. In these circumstances, the imposition of physical constraints on the Internet infrastructure is likely to be more challenging than currently recognized by both Internet operators and regulators alike, which is especially relevant as proposed regulations on data localization emerge in more countries ( Council, 2022 ; McKinsey, 2023 ) . 2. Policy Background Data transfers to jurisdictions with weaker privacy protections expose users in the initial jurisdiction to potential harms such as leaks of personal information to foreign governments or advertisers. Despite these potential harms, little is known about whether or how Internet companies that serve EU users operate their infrastructure to comply with this provision of the GDPR (Chapter V). Specifically, the GDPR in Article 45 mentions that “A transfer of personal data to a third country or an international organisation may take place where the [European] Commission has decided that the third country, a territory or one or more specified sectors within that third country, or the international organisation in question ensures an adequate level of protection.” ( Consulting, 2025 ) Generally speaking, the article refers to several principles that must be followed in such countries receiving data transfers from EU persons, such as “elements like rule of law, respect for human rights and fundamental freedoms, as well as whether or not data subjects’ rights are effective and enforceable, the existence and effective functioning of an independent data protection authority in the non-EEA country and the international commitments the country or international organisation has entered into.” ( Board, 2025 ) These adequacy decisions are published on a European Commission (EC) website ( Commission, 2022a ) and the decisions must be reviewed and renewed periodically by law. At the time of writing, the EC has determined that transfers between the EU and other European Economic Area countries (Norway,