Deepfakes, an increasingly prevalent form of information attack, pose serious threats to security and privacy. Deepfake voice attacks, in particular, have the potential to cause widespread disruption, creating an urgent need for an effective detection system. In this research, we propose the Trident of Poseidon - a novel set of triad training strategies aimed at enhancing the generalizability of deepfake voice detection models. Our solution comprises three key components: (1) Supervised Contrastive Learning, (2) Hard Negative Mining by Audio Re-synthesizing, and (3) Effective Proactive Batch Sampling. Together, these enable the model to learn more robust features. Our extensive experiments demonstrate that our approach outperforms existing methods in both in-domain and out-of-domain testing scenarios, making significant strides toward securing digital media against deepfake voice attacks. Furthermore, we conducted a deeper analysis to explore whether deepfake voices can be categorized into families. By identifying the factors that contribute to the formation of a deepfake voice family, we can better organize a deepfake voice corpus, thereby reducing the effort needed to combat the arms race challenge. Finally, to promote practical utility and community-wide adoption, we have made our solution publicly available as a web application available on deepfake.aisrc.technology, where users can utilize this tool to test for potential deepfake voices.