Researchers reported two advances in AI for genomics. One team used large genomic language models to design entirely new functional genes by leveraging patterns learned from hundreds of billions of bases across prokaryotic life, demonstrating a semantic design approach that produces testable, functional constructs. Separately, BGI and collaborators introduced Genos, a human‑centric genomic foundation model trained on high‑quality human assemblies (HPRC, HGSVC and others) to improve variant effect prediction and functional element identification. Both efforts accelerate computational design and interpretation for synthetic biology and precision medicine.