Before finalising the Scaling Trust programme, we ran a pre-programme discovery call : a round of short, fully open-source research projects — three months, up to £20,000 each — to sharpen the programme’s direction and start building its community.
Teams worked across four tracks: systematising knowledge (from “nature cryptography” to multi-agent security for embodied agents), prototyping agentic coordination challenges, analysing the ethics and risks of multi-agent systems, and growing the community in the UK and beyond.
PactMulti-agent
A choreographic language bringing game theory to agent protocol design.
Profit is the Red TeamMulti-agent
A profit-driven adversary that red-teams agents through economic interaction.
Mechanism Design Is Not EnoughMulti-agent
Incentives aren't enough — agents need intrinsic prosocial preferences to cooperate.
AgentMatrixMulti-agent
Automated benchmarking and securing of AI-agent interaction protocols.
NDAI ZoneFoundations
Can LLMs assess security evidence inside trusted execution environments?
Diagrammatic foundations for trustFoundations
Category-theoretic, substrate-independent foundations for cryptographic trust.
Living ProofFoundations
Mechanobiological cryptography: grounding trust in the physical organisation of living cells, where security rests on the non-invertibility of biological state transitions rather than computational hardness.
Verifiable supply-chain dataFoundations
Cryptographic methods for supply-chain data integrity.
Physical-Layer TrustCyber-physical
Measuring embodied interaction and trust in robot swarms.
LabTrust-GymCyber-physical
A benchmarking environment for trustworthy AI-agent interactions in the lab.
Trusted Molecular MemoryCyber-physical
Trust in hybrid biological–computational systems.
Untrustworthy Things EverywhereCyber-physical
Securing cyber-physical industrial infrastructure.
Who am I?Cyber-physical
Identity and authentication for physically-embodied AI agents.
SoTA communityCommunity
Building a UK community around cyber-physical security.