Abstract Artificial intelligence (AI) is increasingly deployed in oncology and pathology to support diagnosis, prognosis, molecular subtyping, and clinical documentation. Although these systems often achieve high predictive accuracy, their integration into clinical workflows exposes a structural mismatch with existing legal frameworks. U.S. medical law is organized around accountable, reason-giving clinicians, yet many clinically deployed AI systems operate as opaque decision engines that cannot articulate the rationale underlying their outputs. This article argues that treating such systems as mere “tools” obscures a growing justification gap resulting from poor explainability and lack of reasoning of AI models: predictive and prognostic artificial intelligence increasingly shapes clinical outcomes without bearing legal responsibility or providing explanations required for malpractice defense, insurance coverage determinations, or regulatory review. Drawing on malpractice doctrine, insurance law, FDA regulation, and professional licensure regimes, this article demonstrates that clinicians cannot fully delegate diagnostic justification to AI systems, even when those systems materially influence clinical decisions. As a result, clinicians remain legally accountable for decisions they do not entirely control. The article reframes clinical autonomy not as an ethical preference, but as a legal necessity arising from justification-centered accountability structures. It concludes by outlining regulatory and institutional approaches for clarifying responsibility in AI-mediated oncology and pathology without undermining clinical judgment.
Leyva et al. (Sat,) studied this question.