We’re excited to partner with @WalrusProtocol and @SuiFoundation to bring decentralized, privacy-preserving AI to the @SuiNetwork ecosystem. This integration brings Walrus’s decentralized data layer and SEAL’s encryption into FLock’s federated learning stack, enabling secure, community-owned model training.

2/ FLock trains AI models using federated learning, where participants collaboratively train without sharing raw data. To scale this securely, we needed decentralized infrastructure for storing, broadcasting, and encrypting model updates. With Walrus and SEAL, we get both.

3/ Walrus becomes the underlying infrastructure for FLock’s FL Alliance - our network of collaborative training nodes. It provides decentralized broadcasting and storage for model gradients, parameters, and training outputs. No centralized servers, no single point of failure.

4/ SEAL adds programmable encryption and access control to the process. It ensures that only verified participants in a training round can view or contribute gradients. Encrypted data is stored and retrieved securely, without needing to trust any individual server or party.

5/ Walrus and SEAL replace the missing pieces in building truly decentralized, secure federated learning pipelines. This unlocks a key initiative: growing the FL Alliance. We can now onboard more builders into federated learning with built-in privacy and decentralization.

6/ Together with Walrus and SEAL, we’re building the stack for secure, programmable, and community-owned AI. Looking ahead, we’re working with @SuiFoundation to fine-tune an open-source foundation model optimized for agentic AI on Sui. More to come. 💧