In 2019, I met a senior engineer who left Facebook and returned to China to work on a B2B/B2G project based on Federated Learning technology. I introduced him to a Sequoia China investor, and the project quickly secured a $3 million angel round from Sequoia. At that time, I started paying attention to "Federated Learning" technology, feeling that it had great potential when combined with blockchain. The core of Federated Learning is that multiple servers can collaboratively train models without sharing raw data. Data remains on local devices, only model updates (weights/gradients) are shared, which protects user privacy and meets data computation and sovereignty requirements. This is also one of the reasons I am interested in @flock_io; it is not only an AI × Crypto project but also uses "Federated Learning" as its core technology. Recently, Flock has made two significant moves worth noting. The first is a partnership with the Hong Kong Generative AI Research Center (HKGAI). The two parties are collaborating to develop a new generation of AI solutions for government and public sectors, focusing on compliance, reliability, and privacy-first smart governance systems to enhance the accuracy and efficiency of public services. HKGAI is an AI project supported by the Hong Kong government, and this collaboration reflects the government's embrace of AI technology in the "digital oil" era. Flock uses Federated Learning and anonymous encryption mechanisms to ensure the baseline of data privacy and security. Encouraging government involvement is a strong proof of compliance and practicality, and it also provides Flock's business with practical use case support and more data sources. The other news is that last week, Coinbase added $FLOCK to its listing roadmap, and it should be listed soon. $FLOCK is the native token of the Flock project, around which the business model is built. For example, task creators can pay FLOCK tokens to initiate training tasks, and the platform takes a certain percentage as a service fee, while the rest goes to model and computing power providers.