Given the answer to their question on how it's written is quite obvious in the provided link, I'd say if they are using an AI it doesn't appear to be parsing the content and maybe only the titles?