Early Adopters

FRIDGE: Federated Research Infrastructure by Data Governance Extension

Artificial Intelligence (AI) models can help enhance research to solve complex problems in society, but this requires access to sensitive data on large-scale, shared AI supercomputers, like the UK’s AI Research Resource (AIRR).

Using sensitive data to develop AI models involves carefully isolating parts of large AI supercomputers as Trusted Research Environments (TREs) so that only approved researchers can safely access them for their research. FRIDGE aims to explore this possibility by creating a ready-to-use TRE that meets the SATRE specification (developed as part of the DARE UK Phase 1 Driver Projects and enhanced by TREvolution) on the AIRR. This will enable the secure use of sensitive data to develop AI models for research that benefits the public while ensuring strict information governance.

The TRE will be deployed across AIRR supercomputing systems, Dawn (Cambridge) and Isambard-AI (Bristol) and compatible with public cloud platforms, Microsoft Azure and Amazon AWS. Approved researchers will be able to access the TRE and analyse sensitive health data, which will help test and improve the TRE capabilities they use in a real-world setting.

By establishing a robust governance framework, FRIDGE will enable AIRR to host secure TREs, ensuring that large-scale AI research can operate within secure, compliant environments and providing a testbed for other TREvolution capabilities.

By the end of the project, FRIDGE will:

  • Deliver a SATRE and NHS standards-compliant, cross-platform TRE on AIRR for AI-driven research using sensitive data.
  • Enable secure isolation of AIRR computing resources, ensuring only authorised researchers access sensitive data under approved governance frameworks.
  • Publish openly accessible software, legal agreements, and policies, allowing other research projects to deploy sensitive data securely on AIRR or similar supercomputing platforms.
  • Establish AIRR TRE as a platform for testing and validating TREvolution capabilities within large-scale AI environments.

Project information

Lead organisation: The Alan Turing Institute
Principal investigator: Martin O’Reilly
Project duration: 14 months
Project partners: University College London, University of Cambridge, University of Bristol
Funding provided: £399,999
Primary contact email: moreilly@turing.ac.uk

Relevant links

Back to early adopters page

GET IN TOUCH

If you’re interested in learning more about our work, how it can benefit you, or how to get involved, click the button to get in touch with us using our contact form.

Contact us