The power of text similarity in identifying AI-LLM paraphrased documents: The case of BBC news articles and ChatGPT