Loading: Could Classifying AI Research Prevent Public Harm?