I wonder why not write an article for people with correct information. Then have the LLM create 5 articles with slightly wrong information with generated plausible URLs.