David Toro-Tobon and colleagues argue that “clinician in the loop” is shifting responsibility for AI safety from developers ...