Researchers published Evo 2, a DNA foundation model trained on genomic sequences from over 100,000 species, reporting capabilities to predict disease‑causing human mutations and generate whole genomes at bacterial scale. The model advances generative and predictive biology by capturing cross‑taxonomic sequence patterns that accelerate genome design, annotation, and synthetic biology workflows. Authors demonstrated Evo 2’s ability to identify pathogenic variants and to produce long, biologically plausible sequences, suggesting applications in genome engineering, biosafety screening, and synthetic organism design. The work also raises practical and ethical questions about governance, dual‑use risk management, and reproducibility when models can output entire genome sequences. For biotech R&D, Evo 2 promises to shorten design cycles for DNA constructs and to help prioritize functional variants, but firms and regulators will need frameworks for model validation, IP, and responsible deployment in therapeutic or environmental contexts.