Privacy researchers tested 13 leading LLMs against 11 strict protection criteria. Here's how your favorite AI assistant fared ...