Public concern about harmful policing is surging. Governments are paying historic amounts for law enforcement liability. Has police behavior changed? Or is society responding differently? Traditional data sources struggle with this question. Common metrics such as lawsuits and payouts conflate the prevalence and severity of policing harms with the responses of legal actors such as lawyers, judges, and juries. We overcome this problem using a new data source: liability insurance claims. Our dataset contains 23 years of claims against roughly 350 law enforcement agencies that contract with a single insurer. We find that while lawsuits and payouts have trended upwards over the past decade, insurance claims have declined. We generate and test multiple explanatory hypotheses. We conclude that, in our sample, police behavior is not getting worse; rather, public responses to policing harms are intensifying. Data selection, our analysis shows, strongly influences results in policing research.