

Opportunity space
Trust Everything, Everywhere
Scaling Trust
Backed by nearly £50m, this programme sits within the Trust Everything, Everywhere opportunity space and is funding the frontier infrastructure and fundamental research for AI agents to coordinate securely on our behalf, across digital and physical worlds.
Meet the R&D Creators
Our starting cohort of Track 2 and 3 creators are the first explorers on the journey.
Track 2 is the open-source cyber-physical agent stack. On the digital front, we’re funding several integrated approaches to secure agentic interactions, and agentic tooling such as TEE sandboxes. On the physical front, we’re funding a team embedding watermarks within robotic movements, and a team providing autonomous manufacturing data for others to use.
Overarching the digital and physical is a first integration partner, building an end-to-end application by combining the outputs of other Creators and their own work.
Everything is open-source by default, so the stack stays inspectable, contestable, and open for others to build beyond. Integrated tools and agents will compete under open-ended tasks and adversarial pressure in our Arena competition.
This cohort is a starting point. New teams will join through our rolling call for the stack and theory tracks, with grants awarded quarterly, some filling gaps the portfolio exposes, others extending the map entirely.
Our Track 3 project will be developing privacy techniques suited to AI workloads using cryptography techniques.
Track 2 | Tooling
Centre for Cryptographic Trust Infrastructure (CCTI)
Martin Kleppmann, University of Cambridge; Martin Pompéry, SINE Foundation
Project HAMMER
Davide Crapis, Multiscalar Intelligence
Agentic Security Reasoner for Emerging Multi-agent Protocols
Tariq Elahi, University of Edinburgh
Manufacturing Grounding for Cyber-Physical Trust
Sebastian Pattinson and Douglas Brion, Matta
Dovetail: Automated interaction protocol design
Samuele Marro, Institute for Decentralized AI
ProtoSage and ProtoBench: Developing and evaluating agent capabilities in cryptographic protocol reverse-engineering, auditing, and verification
Peter Gilbert, Head of Research, RDI Foundation
Physical Watermarking: Robust policy certification for embodied multi-agent systems
Amanda Prorok, University of Cambridge