From Decentralised Data and Discovery to Federated AI

ESPRESSO Workshop
on
Sovereign Intelligence

London, 13 November 2026

Artificial intelligence increasingly depends on access to large, diverse and continually evolving sources of data. Today, much of that data and the infrastructure used to train and operate AI systems is concentrated within centralised platforms. An alternative is to ask whether useful intelligence can be created across distributed resources while individuals and organisations retain meaningful control over their data, models and participation. 

We use Sovereign Intelligence as a starting point for exploring this question: how can information be discovered, shared and learned from across independently controlled data and compute resources, while preserving the rights, policies and agency of the people and organisations that contribute them? 

A range of technologies and policy developments approach different parts of this challenge. Decentralised data infrastructures and platforms such as Solid explore new models of data control and sharing. Federated learning frameworks enable models to be trained across distributed data without first bringing those data into a common repository. At another layer, UK initiatives around sovereign AI and compute address questions of national capability and access to AI infrastructure. Recent work on data intermediaries further raises questions about delegated data rights, trust and governance within data ecosystems. 

The EPSRC ESPRESSO project approaches this landscape through the problem of information discovery. ESPRESSO is investigating mechanisms through which data controllers can retain authority over access to their resources and over the metadata exposed to third parties, with initial work focusing on personal and health-data environments. 

The second ESPRESSO workshop will bring together researchers, technology practitioners, policymakers and members of the public to explore the emerging idea of Sovereign Intelligence and, in particular, four questions: 

  • What do we mean by Sovereign Intelligence, and what must remain sovereign? 
  • How can information discovery and machine learning operate across independently controlled data resources? 
  • Where can decentralised, federated and centralised approaches complement one another—and where do they offer genuine alternatives?
  • What technical, economic, governance and policy conditions would make Sovereign Intelligence infrastructures viable? 

DATE AND VENUE: 

  • 13 November 2026, CLO B01 Theatre, Birkbeck University of London, London, UK 

REGISTRATION LINK: 

  • Participation is free of charge but requires registration. Kindly fill out the form [TBC] to confirm your participation. 

ORGANISING COMMITTEE: 

  • Mohammad Bahrani, University of Southampton 
  • Faria Ferooz, University of Southampton 
  • Mohamed Ragab, Birmingham City University 
  • Thanassis Tiropanis, University of Southampton  
  • Alexandra Poulovassilis, Birkbeck, University of London  
  • Age Chapman, University of Southampton  
  • George Roussos, Birkbeck, University of London  

WORKSHOP PROGRAMME:

ESPRESSO Workshop on Sovereign Intelligence
From Decentralised Data and Discovery to Federated AI

Welcome
09:00 – 09:30 Registration and breakfast
09:30 – 09:40 Welcome and opening remarks

Framing Sovereign Intelligence 
09:40 – 10:00 Decentralised Data and Infrastructures 
10:00 – 10:20 Federated Learning and Distributed AI 
10:20 – 10:40 Decentralised Search and Sovereign Intelligence 
10:40 – 11:00 Coffee break

Technology and Policy Lightning Talks and Demonstrations
11:00 – 12:30 Speakers to be confirmed

 
12:30 – 13:30 Lunch 

Directions 
13:30 – 14:15 Parallel roundtables

  • Architecture and technology
  • Governance and sovereignty 
  • Adoption, economics and public value 

14:15 – 14:45 Roundtable reports and closing panel: What should happen next?

14:45 – 15:00 Closing remarks 

15:00 Onwards Coffee and networking