Tick-borne viruses remain a substantial zoonotic risk worldwide. The advent of next-generation sequencing has significantly broadened our knowledge of viruses. However, large amounts of sequences in public datasets remain unannotated which could represent unrealised or undocumented viruses. This study aimed to assess whether these novel and other understudied viruses likely represent zoonotic risks and thus deserve further investigation. We ranked the human infection potential of 136 known tick-borne viruses using a genome composition-based machine learning model. We found five high-risk tick-borne viruses (Langat virus, Lonestar tick chuvirus 1, Grotenhout virus, Taggert virus, and Johnston Atoll virus) that have not been known to infect human and two viral families (Nairoviridae and Phenuiviridae) that contain a large proportion of potential emerging tick-borne viruses. Through bioinformatic analyses, we presented 83 unannotated contigs exhibiting high identity with known tick viruses in public meta-genomic and -transcriptomic datasets. These putative novel viral contigs were classified into three RNA viral families (Alphatetraviridae, Orthomyxoviridae, Chuviridae) and one DNA viral family (Asfaviridae). After manual checking, these 83 contigs were reduced to 5 putative novel Alphatetra-like viral contigs, 5 putative novel Orthomyxo-like viral contigs, and 1 Chu-like viral contig which clustered with known tick-borne viruses, forming a separate clade within the viral families. This study added to the knowledge of tick virus diversity and highlighted the importance of surveillance of newly emerging tick-borne diseases.