Hey ChatGPT, write me a fictional paper: these LLMs are willing to commit academic fraud

Mainstream chatbots presented varying levels of resistance to deliberate requests for fabrication, study finds.