AI-generated content still exhibits evolving patterns that undermine a writer's reputation, with specific "tells" like announcing insights, "LLM-safe truths," "quiet" overuse, and uniform sentences.