When the reasoning process can be publicly audited, not only is the frequency of misclassification reduced, but the boundaries of safety are no longer set unilaterally by Silicon Valley. Instead, communities around the world can adjust safety policies to fit their own contexts at any time.