Explaining the Wrongs of Undisclosed LLM Use: Reply to Sahebi and Montefiore

Sahebi and Montefiore (2026) provide a compelling and timely criticism of undisclosed LLM use in writing. Their central point, put briefly, is that undisclosed LLM use epistemically deprives an audience of key information about the authentic competencies of the LLM user. But in this reply, I argue that their explanation for the wrongness of undisclosed LLM use only cleanly handles a proper subset of cases of undisclosed LLM use, and has trouble with other examples. On that basis, I suggest two additional ways in which undisclosed LLM use is likely to be morally wrong: it wrongfully fakes a kind of signal of caring, or it wrongfully fakes the very identity of the author. I consider a potential reply based on a thread of argument in Sahebi and Montefiore’s original paper, but suggest that it either does not go far enough or risks undermining their overall thesis. Along the way, I also gesture at the need for innovative normative models, such as care ethics and second-personal ethics, to explain the wrongness of undisclosed LLM use.