Might be as simple as opening a website and reading some text. Technically also "downloading" - but why not use curl for any http-based reading of data.