ARPA‑H awarded multi‑million dollar contracts to industrial‑academic consortia developing AI and human biology‑based drug safety models. Inductive Bio won up to $21 million to lead DATAMAP (Digital Acceleration of Toxicity Assessment), partnering with Amgen, Baylor, Cincinnati Children’s and others to generate organoid, ex vivo and microphysiological data focused initially on DILI and cardiotoxicity. Separately, a Ginkgo‑led PREDICTS consortium received up to $31.7 million to build in silico and ex vivo safety prediction platforms. Both programs will train AI models on data drawn from advanced human model systems rather than traditional animal testing and will work with regulators on qualification and context‑of‑use validation. Inductive’s team emphasized regulatory engagement with the FDA; Ginkgo’s effort will supply transcriptomic and cell‑painting datasets to power predictive algorithms. Clarification: DILI = drug‑induced liver injury. These programs aim to reduce late‑stage safety attrition by building human‑relevant preclinical predictors that can de‑risk candidates earlier in development.