Google DeepMind unveiled AlphaGenome, an AI model that analyzes up to one million base‑pair stretches of noncoding DNA and predicts how mutations affect gene expression. The work, described in Nature, integrates multiple genomic datasets to infer regulatory effects across long genomic regions and aims to help researchers prioritize functional variants in complex loci. AlphaGenome is positioned as a successor to DeepMind’s AlphaFold success, applied to the genome’s regulatory “dark matter.” DeepMind released a version of the model for research use, enabling external labs to benchmark variant‑effect predictions and to guide experimental follow‑ups. The model accelerates efforts to interpret noncoding variation, a key bottleneck in translating genome sequencing into mechanistic insights and therapies. Teams using AlphaGenome will test its value in variant prioritization, CRISPR target selection and functional genomics pipelines.