Researchers at Cold Spring Harbor Laboratory developed BATMAN, an AI model trained on the expansive BATCAVE database of over 22,000 TCR-peptide interactions, achieving superior accuracy in predicting T cell receptor cross-reactivity. This advance aids efforts to engineer TCR therapies with reduced off-target effects, critical for cancer immunotherapy safety. The study, published in Cell Systems, reveals that comprehensive mutational scanning and Bayesian inference enable effective prediction of TCR binding landscapes, a major step for designing targeted immunotherapies and addressing challenges in immune receptor specificity.