Secure Processing Environments: AEPD Guide to Protecting Sensitive Data
Este contenido está disponible únicamente para socios BAIDATA.
Regístrate o accede con tu cuenta para verlo.
The secondary use of personal data, especially in areas such as scientific research and health, poses a challenge: making it easier to access and analyse valuable information without compromising people’s privacy.
Secure Processing Environments (SPE) are key to promoting research with sensitive data, because they make it possible to analyse the data in a controlled space without the data ever leaving it. To address this challenge, the Spanish Data Protection Agency (AEPD) and the Hellenic Data Protection Authority (HDPA) have produced a joint guide that analyses how to design these environments, what risks they present and what technical and organisational measures can help ensure secure processing of information.
A secure processing environment is a controlled space in which personal data can be processed for authorised purposes, limiting the risks of unauthorised access, disclosure or extraction of information.
A good example is the reading room of an archive: you can consult the material, but nothing leaves without authorisation. The idea is to bring the user to the data, not the data to the user.
Its aim is not only to protect data against cyberattacks, but to establish a set of safeguards that make it possible to carry out analyses and operations on sensitive information without unnecessarily exposing the original data.
To achieve this, SPEs combine three dimensions:
The guide insists that the security of these environments must be approached in an integrated way. It is not enough to have a protected technological infrastructure: it is also necessary to ensure that the conditions of use, governance and control over the data are appropriate throughout the entire processing cycle.
SPEs are based on the Data Governance Act (EU 2022/868) and the European Health Data Space Regulation (EU 2025/327). The latter requires that the secondary use of health data take place within an SPE managed by a health data access body. Researchers work with the information, but cannot download personal data: only anonymised and reviewed results leave the environment, and all activity is recorded in an auditable way. An SPE does not replace the GDPR or change who is the controller and who is the processor.
The guide distinguishes between legal control over the environment and de facto control, that is, the practical ability to limit, monitor and stop operations. Its golden rule is that what is legally prohibited should also be technically prevented. This is key when a technology provider is involved, because neither a contract nor an external certification is enough on its own. It recommends avoiding dependence on a single provider and distributing critical functions so that no one controls the entire environment. It also asks you to look out for signals such as exposure to legislation from outside the European Economic Area or the possibility of the provider accessing the encryption keys.
One of the aspects the guide addresses is the identification of the risks associated with processing personal data in these environments.
Among the issues to be taken into account are the possibility of unauthorised access, the disclosure of information during processing, the risks arising from third-party providers and the extraction of results that may reveal personal information.
This last aspect is especially important. Even if the original data remain within a controlled environment, the results of an analysis may also contain sensitive information or allow people to be identified.
Therefore, protection must not be limited to the storage of and access to the data: it must also extend to the operations carried out and to the results that leave the environment.
The guide proposes combining several tools.
It also warns that repeating queries with small variations can be used to reconstruct individual data, so outputs should be reviewed and the number of queries limited.
On the technical side, it proposes starting from the assumption that nothing and no one is trustworthy by default, with minimum permissions, strong authentication and an environment isolated from the outside (no USB or shared clipboard).
How can a data space change the way researchers access protected statistical information? The INVEUS project (Simple exchange of statistical data between researchers and Eustat) offers a very practical example.
Driven by MIK S. Coop. together with Eustat, the Basque Statistics Institute, INVEUS addresses a specific need: making it easier for researchers to access statistical information that, because of its confidentiality, cannot be published openly.
Until now, accessing certain statistical microdata involved an in-person procedure heavily conditioned by security measures. Researchers had to request an appointment, travel to Eustat’s offices and, after passing the security checks, work on an isolated computer with no internet connection. There they could carry out their analyses and, once the results had been reviewed, print them to take away.
The data space makes it possible to transform this procedure and enable remote work, while maintaining control over the protected information.
BEFORE
In-person access and an isolated environment
AFTER
Remote research through the data space
The infrastructure developed makes it possible to integrate different sources of information and carry out analytical operations within a controlled environment. Researchers can incorporate their own data, from surveys, private companies or other sources, and find in Eustat’s catalogue statistical information that complements their research.
The service facilitates the combination of both sources and the running of advanced analyses. At the same time, it incorporates connectors and control mechanisms to ensure that downloads and the extraction of results respect the applicable legal restrictions, including the GDPR and statistical legislation.
The proposal also pays attention to user experience: it aims to offer a simple interface, accessible to researchers without advanced technical knowledge, and to integrate processing into tools that are common among this group.
The project also envisages evaluating the possibility of extending this infrastructure to other scenarios, with larger volumes of data, additional analytical needs or unstructured information.
INVEUS has also contributed to the recognition of its promoters within the BAIDATA ecosystem. In 2024, MIK S. Coop. and Eustat were distinguished at the BAIDATA Awards, in the Use Cases category, for their innovative implementation of data use cases in a cooperative environment and for their role as a reference in the opening and sharing of public statistical data.
The guide offers several clear messages for hospitals, public administrations, data providers and technology platforms:
This article is part of our series on privacy and data spaces. If you need support to design an interoperable, sovereign data architecture that lets you share and analyse sensitive data with guarantees, consult the technical guides in the BAIDATA Library or take part in our upcoming Deep Dives webinars.
Not yet part of our ecosystem? Discover the benefits of joining BAIDATA to lead the sovereign data economy. Request membership information here.