This working paper introduces the Risk–Danger Fallacy, a diagnostic framework that clarifies a central misapplication of algorithmic or actuarial risk tools in child welfare systems: treating probabilistic risk estimates as if they are evidence of present danger. Drawing from empirical research, federal policy analysis, and constitutional due process doctrine, the paper articulates core principles distinguishing probability from proof and traces the causal chain through which predictive harm emerges. It proposes practice, judicial, and policy reforms to restore evidentiary discipline and prevent wrongful family separations driven by systemic misuse of predictive risk technologies. The framework is intended as a conceptual working model and does not constitute legal advice or a diagnostic instrument for individual cases.
Megan H. McCarthy (2026) studied this question.