Over the past year, researchers have found clear evidence of the use of AI text generators in several scientific papers. And yet when it comes to the illicit scholarly use of AI chatbots, we may have only seen the tip of the iceberg. Authors have been caught publishing works with legitimate-looking scientific references that were entirely invented by AI. Others were shown to have left snippets of text from ChatGPT's user interface pasted in their work. One maths study was even retracted after it