Researchers published Evo 2, a foundation DNA model trained on genomes from over 100,000 species, and demonstrated the model’s ability to identify disease‑causing mutations and design genome‑length sequences for simple organisms. The Nature paper positions Evo 2 as a generative tool for sequence design and functional inference across diverse taxa. Authors showed Evo 2 can learn regulatory and coding patterns that generalize across evolutionary distance, enabling accelerated hypothesis generation for synthetic biology and comparative genomics. Evo 2 is an AI foundation model for genomics: it learns statistical patterns in DNA sequences to predict function or propose edits. The approach could shorten cycles for engineering pathways, but it will require experimental validation and regulatory oversight for deployment in applied settings.