From Frye to Daubert, and Beyond: Admissibility of AI-Assisted Forensic Evidence After Washington v. Puloka (2024)

The rapid integration of AI-assisted tools into criminal investigation and adjudication presents significant challenges to traditional evidentiary doctrines. Courts are increasingly confronted with forms of proof whose inferential processes are statistically mediated, partially opaque, and resistant to conventional modes of adversarial scrutiny. As a result, existing admissibility frameworks—developed primarily for human-driven scientific evidence—are being asked to regulate machine-driven epistemic processes in ways for which they were not originally designed. This Article examines how courts can respond to these challenges through an analysis of Washington v. Puloka (2024), a recent decision addressing the admissibility of AI-assisted forensic evidence. Although grounded in U.S. evidentiary doctrine, particularly the evolution from the Frye test to the Daubert standard, the Article argues that the issues raised by AI-generated evidence transcend jurisdictional boundaries. Rather, they operate as a stress test for foundational procedural values common to modern criminal justice systems, including scientific reliability, transparency, procedural fairness, and the right to meaningful adversarial challenge. The Puloka ruling illustrates how courts may exercise principled gatekeeping by scrutinizing the scientific foundation and empirical validation of AI systems, evaluating the variability and contextual dependence of error rates, examining the disclosure of training data and algorithmic methodology, and assessing whether the defense can meaningfully test and contest the evidence. At the same time, the decision highlights the need to guard against undue prejudice arising from the perceived objectivity and authority of algorithmic outputs. Drawing on these insights, the Article proposes a structured, Daubert-informed analytical framework intended to inform and support investigators, criminal justice practitioners, and courts in evaluating the admissibility of AI-assisted forensic evidence. The model contends that technological sophistication alone cannot substitute for evidentiary reliability or constitutional fairness, and that the legitimacy of AI-assisted proof ultimately depends on its continued susceptibility to meaningful judicial oversight and adversarial testing.