Researchers using RFdiffusion reported de novo antibody designs that bind user‑specified epitopes with atomistic accuracy; multiple designs were validated by cryo‑EM, marking a milestone in computational biologics. The Nature paper details how AI models can construct complete antibodies, including challenging loop regions, and yields structures that match predicted binding poses. David Baker and colleagues emphasized the advance while noting the field remains early stage: computational designs require experimental validation and optimization, but the work cuts development timelines and could alter discovery workflows for therapeutic antibodies. The breakthrough attracts industry interest: de novo AI antibody design promises to reduce reliance on traditional screening and could accelerate hit‑to‑lead cycles if real‑world manufacturing, immunogenicity and developability hurdles are addressed.