Chinese Publisher Recalls Novel After AI Response Found Printed in Text
Jiangsu Phoenix Fine Arts Publishing House has agreed to recall and destroy physical copies of a novel after readers found a typical AI prompt response printed directly in the text. The publisher confirmed that the mistake occurred during editorial processing rather than in the author's original writing. The mistake highlights the risks of incorporating generative AI tools into editorial workflows without adequate proofreading. It provides a concrete example of how unreviewed AI artifacts can slip past human quality control into commercial print media. The printed artifact asked whether the user needed sentence fine-tuning to better match a supernatural atmosphere, a line completely absent from the online web novel version. The publisher stated that multiple editors and proofreaders handled the manuscript, and promised to release a re-edited edition.
## BACKGROUND
Large language models (LLMs) often include conversational framing in their outputs, such as offering to make further edits or adjustments to tone. When editors or writers copy text generated by AI without careful reviewing, these conversational prompts can inadvertently remain in final published work.