DISTRIBUTE//AI
DISTRIBUTE//AI brings together machine learning engineers, cryptographers, mechanism designers, and systems builders to tackle one of the most consequential open problems in AI: how do we build AI that is not owned by three companies in one country?
This is a 48-hour collaborative hackathon on April 17–19, 2026, co-organised by Bloomsbury Technology, Abelar, and Unitaware, with a London HQ and 20+ simultaneous chapters across Eurasia and Africa. Every participant contributes to a single shared architecture: a decentralised, encrypted, incentive-aligned LLM built on blockchain. One codebase. One mission. 100+ builders across 15+ countries.
Why this hackathon?
Three companies control nearly 90% of the general-purpose AI market.¹ The models mediating access to knowledge, opportunity, and public services are trained on centralised infrastructure, governed by private interests, and inaccessible to the vast majority of the world's researchers and developers.
This is not a technical inevitability. The tools to change it now exist. Federated learning, homomorphic encryption, zero-knowledge proofs, and on-chain incentive mechanisms have each matured to production readiness. What is missing is a coordinated build that assembles them into a working system.
Governments are legislating fast: the EU AI Act entered into force in August 2024² and is already reshaping global compliance norms. With AI projected to become a $4.8 trillion market by 2033³, the window to establish decentralised alternatives before the current architecture becomes entrenched is narrow. That is what this hackathon is for.
Hackathon Tracks
CS / ML: Model Design LLM architecture for decentralised training. Federated learning, model partitioning, gradient aggregation.
Cryptography: Weight Security Encrypt weights, preserve inference. Homomorphic encryption, MPC, zero-knowledge proofs.
Mechanism Design: Incentives Tokenomics for decentralised compute contribution and anonymised inference. Design systems where participation is economically rational.
Engineering: Build Smart contracts, node infrastructure, API layer. Integration and deployment of the full stack.
Who should participate?
Machine learning engineers and researchers Cryptographers working in ZK, MPC, or homomorphic encryption Blockchain and smart contract developers Economists and mechanism designers with interest in token incentive systems Systems engineers with infrastructure or distributed systems experience Students and early-career researchers exploring decentralised AI
No cross-domain experience required. Deep expertise in your own track is enough.
What you will do
Over 48 hours, you will form a team, choose a track, and contribute working code, documentation, or research to a shared open-source repository. At the end of the event, teams present their contributions to judges and mentors from across the decentralised AI ecosystem.
All outputs are open-source. Everything shipped belongs to the community.
What happens next
All submitted work will be reviewed by expert judges. Top contributions are featured publicly and may be invited for further development or publication. Findings from this event will feed into ongoing research and policy work on decentralised AI governance.
Why join?
DISTRIBUTE//AI is co-organised by Bloomsbury Technology, a technical AI research lab, alongside Unitaware (a global community network with 100+ members across Eurasia and Africa) and Abelar. Partners include Gonka (decentralised AI inference, mainnet live) and Loyal (privacy-preserving AI on Solana).
¹ Brookings Institution Menlo Ventures, 2025. https://www.brookings.edu/articles/what-happens-when-ai-companies-compete-with-their-customers/ ² European Commission. EU AI Act, entered into force 1 August 2024. https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai ³ UNCTAD Technology and Innovation Report 2025. https://unctad.org/newsai-market-projected-hit-48-trillion-2033-emerging-dominant-frontier-technology