This research used a novel deep learning algorithm, LucaProt, to significantly expand the known RNA virosphere. LucaProt integrates sequence and structural information for superior accuracy in identifying RNA viruses, even highly divergent ones. The study uncovered a substantial number of previously unknown RNA virus species and supergroups across diverse global ecosystems. The findings highlight the immense, still largely unexplored diversity of RNA viruses and underscore the power of AI in biological discovery. Verification methods, including RT-PCR, supported the RNA nature of many newly identified supergroups. Ecological analyses revealed varying virus diversity and abundance across different environments.