Catch bad users and bad content. Adaptive ML for user-level threats. LLM infrastructure for content violations. Two specialized tools from one T&S team.