GAEA is a decentralized protocol for emotional data collection and AI training. It leverages users' idle network nodes to collect multimodal emotional data (e.g., voice, facial expressions, physiological signals) under privacy protection. The data is stored decentrally, quality-scored, and formed into high-value datasets for training AI models with emotional understanding. Users earn points by contributing data and network resources, which can be redeemed for future benefits. Its core mechanisms include data collection, quality assessment, incentive distribution, and a model feedback loop.