“factual accuracy in large language models remains an area of active research”