FLock.io Partners with Walrus to Power Decentralized, Privacy-Preserving AI Training
Palo Alto, USA, July 29th, 2025, Chainwire
Bringing Walrus’s decentralized data layer and SEAL encryption to FLock.io, unlocking secure, community-owned AI development.
FLock.io, the first decentralized AI training platform, will build on Walrus as its core data layer to bring secure, privacy-preserving AI model development to the Sui ecosystem.
This partnership connects Walrus’s robust decentralized infrastructure, including data storage, availability, programmability, and access controls, with FLock.io’s innovative approach to AI development. FLock.io utilizes Federated Learning and blockchain to enable communities to build, train, and own AI models without centralizing sensitive data. Their ecosystem includes AI Arena for competitive model training, FL (Federated Learning) Alliance for privacy-focused collaboration, and Moonbase for decentralized model hosting. Together, these components provide an end-to-end stack for open, community-governed AI development.
The initial phase of this integration centers on strengthening FLock.io’s FL Alliance with Walrus and SEAL. Walrus provides the decentralized broadcasting and storage layer for model parameters and training outputs across federated learning nodes. SEAL, a decentralized secrets management system, enforces gated access and data confidentiality, ensuring that only verified federation members can access or contribute encrypted gradients and data.
“Previously, our data solutions either lacked decentralization or encryption, creating onboarding hurdles,” said Jiahao Sun, Founder & CEO of FLock.io. “Walrus changes that. It gives us a fully decentralized, encrypted storage solution for gradients, which means we can confidently expand our FL Alliance and bring more secure AI training to our users.”
Looking ahead, this partnership aims to deliver a true agentic AI experience on Sui through open collaboration and hands-on development. The next phase will involve fine-tuning an open-source foundation model, optimized for agentic interactions within the Sui environment. The teams aim for this to lead to a functional prototype, comparable to a “Copilot for the Sui Blockchain,” capable of Move-native code generation, smart contract assistance, and context-aware agentic reasoning.
“Walrus provides the global data layer for builders, giving them full control over their data and new ways to create value. Our collaboration with FLock.io showcases the power of Walrus for decentralized AI, enabling a truly secure and programmable foundation for cutting-edge AI model development,” said Rebecca Simmonds, Managing Executive of the Walrus Foundation.
By combining secure data infrastructure with innovative federated learning, this partnership opens new avenues for builders to create more collaborative, private, and powerful AI models, creating value and control for developers in the new data economy.