Secure EU AI Platform

The Secure EU AI Platform enables batch processing of sensitive and personal data on isolated, exclusive HPC nodes through end-to-end encryption and other security measures. It enables the processing of critical data in accordance with GDPR Art. 9.

Cables in a server rack

Target users

  • SMEs, startups, and large enterprises, particularly in manufacturing and engineering, that must comply with legal or contractual data protection requirements
  • Public institutions processing sensitive data
  • Critical infrastructures for which the security measures of classic HPC are not sufficient

Description

The Secure EU AI Platform (an external service provided by KISSKI) enables batch processing on HPC hardware with maximum security. Its secure compute environment provides a strong foundation for AI applications and is particularly suited for secure model training and optimization for HammerHAI users. It enables users process data while meeting security requirements, helping reduce both the cost and deployment effort often associated with fully isolated computing environments.

This is made possible by:

  • Processing on completely isolated nodes (all ports disabled, no admin access possible)
  • End-to-end encryption of data, software containers, and batch scripts
  • Cryptographic man-in-the-middle protection at the user, node, and job levels
  • Secure key exchange via a key management system

The Secure EU AI Platform also includes:

  • Consulting on the design and implementation of secure workflows
  • Support for researchers in testing, porting, and debugging their batch scripts

Applications

  • Processing of pseudonymized medical image data (CT, MRI)
  • Analysis of sensitive patient data (e.g., genome sequences, clinical studies)
  • Fine-tuning of large language models with sensitive corpora (e.g., internal company documentation)
  • Machine learning/deep learning on critical data sets (e.g., personal data)
  • Monte Carlo methods on internal data for financial or risk modeling
  • Simulations on critical data (e.g., energy sector)

How to get access

  • Check whether your use case is compatible with the Secure EU AI Platform (see above).
  • Contact us with a description of your workflow.
  • Set up a test environment in collaboration with us to ensure that there are no technical complications with your workflow and the Secure EU AI Platform (using only non-sensitive, synthetic test data).
  • Please note: We strongly recommend that you create the following documents to comply with your own GDPR standards:
    • Data processing agreement (DPA)
    • Record of processing activities (RPA)
    • Operating concept (OCP)
  • Users must set up the open-source Secure Client in their local, trusted environment and allow the client to connect to the front-end nodes of the KISSKI HPC.

Publication: https://doi.org/10.1109/CCGrid54584.2022.00118

Documentation: https://docs.hpc.gwdg.de/services/secure-hpc

Secure Client: https://github.com/gwdg/secure-hpc

To use this service, go to:

https://kisski.gwdg.de/en/leistungen/2-13-02_secure-hpc-hammerhai/

Service provided by

Logo of the GWDG (Gesellschaft für wissenschaftliche Datenverarbeitung mbH Göttingen). The abbreviation is stacked on top of the written-out name. To the left of this wordmark five blue squares of different sizes can be seen touching each other with their rounded corners.
KISSKI (KI-Servicezentrum für sensible und kritische Infrastrukturen) logo featuring its symbol and wordmark.

Contact

Jakob Dieterle

GWDG

Lars Quentin

GWDG

Funding

The acquisition and operation of the EuroHPC Al-optimised supercomputer is funded jointly by the EuroHPC Joint Undertaking, through the European Union’s Digital Europe Programme, as well as by the German Federal Ministry of Research, Technology and Space (BMFTR) and the Baden-Württemberg Ministry of Science, Research and the Arts.

Views and opinions expressed are those of the author(s) only and do not necessarily reflect those of the European Union or the EuroHPC JU. This project has received funding from the European High Performance Computing Joint Undertaking under grant agreement No. 101234027. It is co-funded by the European Commission, the German Federal Ministry of Research, Technology and Space (BMFTR), the Baden-Württemberg Ministry of Science, Research and the Arts, the Bavarian State Ministry of Science and the Arts and the Lower Saxony Ministry of Science and Culture.

Logo of the Federal Ministry of Research, Technology and Space of Germany. A black eagle icon (coat of arms of Germany) on the left, a vertical strip in the colors of the German flag (black, red, yellow) to the right of it and even further to the right the wordmark "Bundesministerium für Forschung, Technologie und Raumfahrt".
Logo of the Federal Ministry of Research, Technology and Space of Germany. A black eagle icon (coat of arms of Germany) on the left, a vertical strip in the colors of the German flag (black, red, yellow) to the right of it and even further to the right the wordmark "Bundesministerium für Forschung, Technologie und Raumfahrt".