Google DeepMind published AlphaGenome, an AI model that predicts functional effects of noncoding DNA across up to one million base-pair sequences, and released the tool alongside a Nature paper. Early external users and peer reviewers are urging independent validation and head-to-head comparisons against emerging models, raising questions about benchmarking datasets and reproducibility for large genomic language models. DeepMind says AlphaGenome can map noncoding variants to gene-expression effects and shared the model with researchers; reviewers are asking for standardized evaluation against established datasets and clinical variant catalogs. The release advances AI-driven interpretation of regulatory DNA but highlights the community’s demand for transparency, open benchmarks, and experimental validation before clinical applications can be contemplated.