Language models are a type of AI that generate human language, but they have limitations and can make unintentional mistakes. Despite their usefulness, AI language models are not infallible. In this article, we discuss the challenges of creating a crossword puzzle with AI and the potential for errors and assumptions.