Researchers characterize statistical separability in TP-CRIV for probabilistic AI models, estimating the verification budget needed for a target AUC from challenge and repetition counts
This work characterizes statistical separability in third-party challenge-response identity verification (TP-CRIV) for probabilistic AI models, relating the challenge-wise behavior of matching and non-matching provers to verification-level separability, explicitly describing how the numbers of independent challenges and repeated responses affect detection performance and enabling estimation of the verification budget required for a target AUC; the authors instantiate the characterization for LLMs using open-ended challenges, and experiments show matching-non-matching separation, close agreement between theoretical and empirical AUCs, and consistent estimates of minimum verification budgets.