Teams

The teams funded under Safeguarded AI, by technical area, as ARIA’s own funded projects page groups them. Each title opens the project’s card there, which names the team and says what the work is.

TA1.1 Tooling (Theory)

• Safety: Core representation underlying safeguarded AI Closed

  Ohad Kammar, University of Edinburgh

• Axiomatic Theories of String Diagrams for Categories of Probabilistic Processes Closed

  Fabio Zanasi, University College London

• SAINT: Safe AI ageNTs – Formal certification of AI agents in the world

  Alessandro Abate, University of Oxford Team

• ULTIMATE: Universal Stochastic Modelling, Verification and Synthesis Framework

  Radu Calinescu, University of York

• Quantitative Predicate Logic as a Foundation for Verified ML

  Ekaterina Komendantskaya + Robert Atkey, University of Southampton/Heriot-Watt University; University of Strathclyde

• Learning-Theoretic AI Safety Closed

  Vanessa Kosoy, Association of Long Term Existence and Resilience (ALTER)

• Probabilistic Protocol Specification for Distributed Autonomous Processes Closed

  Nobuko Yoshida, University of Oxford

• Philosophical Applied Category Theory

  David Corfield, Independent Researcher

• Double Categorical Systems Theory for Safeguarded AI

  David Jaz Myers, Topos Research UK

• Monoidal Coalgebraic Metrics

  Filippo Bonchi, University of Pisa

• Doubly Categorical Systems Logic: A theory of specification languages

  Matteo Capucci, Independent Researcher

• True Categorical Programming for Composable Systems Total Closed

  Jade Master + Zans Mihejevs, Glasgow Lab for AI Verification (GLAIVE)

• Unified Automated Reasoning for Randomised Distributed Systems Closed

  Alexandra Silva, University College London + Cornell University

• Employing Categorical Probability Towards Safe AI

  Sam Staton, University of Oxford

• Syntax and Semantics for Multimodal Petri Nets

  Amar Hadzihasanovic, Tallinn University of Technology

• Event Structures as World Models

  Alex Kavvos, University of Bristol

• Supermartingale Certificates for Temporal Logic

  Mirco Giacobbe, University of Birmingham

• Hyper-optimised Tensor Contraction for Neural Networks Verification Closed

  Stefano Gogioso, Hashberg Ltd

• A Computational Mechanics Approach to World Models via Multi-Scale Compositional Abstractions

  Fernando Rosas, University of Sussex

• String Diagrammatic Probabilistic Logic: Specification and modelling languages

  Pawel Sobocinski, Tallinn University of Technology

• Profunctors: A unified semantics for safeguarded AI

  Nicola Gambino, University of Manchester

• Modal Types for Quantitative Analysis of Higher-Order Probabilistic Programs

  Vineet Rajani + Dominic Orchard, University of Kent

TA1.2 + 1.3 Tooling (Platform)

TA1.2

Building the platform.

• From string diagrams to GPU optimisation

  Paolo Perrone + Nikolaj Jensen, Adjoint Labs Limited

• Obsidian Systems

  Colin Hobbins, Obsidian Systems

• CatColab: collaborative modeling, specification, and verification using double categorical logic and systems theory

  Evan Patterson, Topos Institute

• Data-Parallel Proof Checking for Monoidal Theories

  Professor Fabio Zanasi, University College London; Paul Wilson, Hellas AI

• Automated Reasoning Technologies for AI Safety Verification

  Mirco Giacobbe, Luca Arnaboldi + Pascal Berrang, Zeroth Research

TA1.3

Designing its user-facing experience.

• GAIOS

  Peter van Hardenberg, Ink & Switch

• Safeguarded Collaboration with AI Agents in a Type-Theoretic Computational

  Cyrus Omar + Andrew Blinn University of Michigan

• UHURA: UX for Human-centric User-Responsive AI

  Dei Vilkinsons + Ciaran Morinan, HASH

TA1.4 Tooling (Sociotechnical Integration)

• Law-following AI

  Cullen O'Keefe, Director of Research, Institute for Law & AI

• Field Building for Better Formal Models of Society

  Joe Edelman + Ryan Lowe, Meaning Alignment Institute

• AI-enabled Governance Models for Advanced AI R&D Organisations

  Alex Petropoulos, Centre for Future Generations

• Privacy-preserving AI Safety Verification

  Pascal Berrang + Mirco Giacobbe, University of Birmingham; Yang Zhang, CISPA Helmholtz Center for Information Security

• Aggregating Safety Preferences for AI Systems: A Social Choice Approach

  Markus Brill, University of Warwick

• Deliberative AI Specifications and Infrastructure

  Aviv Ovadya, AI & Democracy Foundation

TA2 Applications

• SAILS: Safeguarded AI for Logistics and Supply chain management

  Dei Vilkinsons + Ciaran Morinan, HASH

• Safeguarded AI-Enabled Biopharmaceutical Manufacturing

  Mirco Giacobbe + Leonardo Stella, University of Birmingham

Archive

Projects from the TA3 and the first TA2 funding calls, all of them closed.

TA3 Real-World Applications

• SAGEflex: Safeguarded AI Agents for Grid-Edge Flexibility Closed

  Thomas Morstyn + Jakob Foerster, University of Oxford

• SAINTES: Safe and scalable AI decisioN support Tools for Energy Systems Closed

  Dawei Qiu + Zhong Fan, University of Exeter

• SAFER-ADS: Towards Safety Assurance of Frontier AI for Automated Driving Systems Closed

  Simon Burton, University of York

• Transforming Clinical Trial Design Using Safe and Reliable AI Closed

  Nijat Hasanli, Lindus Health; Nobuko Yoshida, University of Oxford

• Digital Custodians for Ageing Infrastructure Closed

  Nathan Korda, Mind Foundry

• Safeguarded AI for Energy Savings in Radio Access Networks Closed

  Marco Fiore + Paul Patras, Net AI

• Towards Large-Scale Validation of Business Process Artificial Intelligence (BPAI) Closed

  Nobuko Yoshida + David Parker, University of Oxford

TA2 Machine Learning (Phase 1)

• Cognitive Emulation: Our Path to Safeguarded AI Closed

  Connor Leahy, Conjecture

• SHIELD: Safeguarding High-Impact AI for Enhanced Manufacturing Development Closed

  Mohammed Begg, The Manufacturing Technology Centre

• Recursive Safeguarding Closed

  Younesse Kaddar, Recursive Safeguarding Limited