In of the most puzzling and high-profile cases in the world of AI, a $2m two-book publishing deal with Minotaur, part of Macmillan US, has been withdrawn by the author’s agent because of concerns over the use of AI.
A 14-way auction was brokered by primary agent Marc Gerald at Brooklyn-based Europa Content, while Sandy Hodgman at Hodgman Literary is understood to have received multiple six-figure offers following a competitive multi-way auction in the UK.
Gerald told the Bookseller: “The book was stunningly good, everyone fell in love with it. It’s a book that appeals to absolutely everyone.” But an editor raised concerns that the book “had signs that AI might have been used. We went through the manuscript [with the author] in extreme detail, discussed the responsibilities of an author, and ultimately continued with the submission”.
However, the process of trying to authenticate the book continued, and following a meeting on 29th July “aspects of [the author’s] story changed”, Gerald said, and the agency decided to “sever ties, as we need to have total faith”.
He added: “We feel terrible for putting this into the universe and the firestorm it has caused. It’s very much a moving target. The use of AI could look very different in 10 years, though we certainly hope not. The book itself is genius – however it came to be drafted, it’s amazing.”
The author told the Bookseller: “The story, characters, scenes, humour, structure and voice came from me, and I have time-stamped drafts, written notes and dated WhatsApp messages documenting the book’s development. I am also concerned that AI-detector scores are being treated as conclusive evidence. I tested passages from books published in 2012 and 2017, and the software classified them as 98 and 99% AI-generated, despite their predating the widespread availability of generative AI. Within six hours of the
rumours, I lost US, UK, translation, and screen opportunities without being given a proper chance to explain my writing process.”
An editor who wishes to remain anonymous, and a bidder for the novel, told the Bookseller: “I read the entire manuscript and found that it had the same flaws other debut novels I’d acquired had: pacing issues, clunky writing, maybe some try-hard metaphors. But nothing that suggested it had been written with AI, and neither did any of my sales, marketing or publicity colleagues who also read alongside me.
“What concerns me is whose responsibility it is to determine that the writing is AI-created or AI-assisted? And how do we determine that? By using AI-detectors that are a. not entirely accurate and b. themselves AI. It also brings into question the ethics of uploading an author’s work into an AI model. I’m not sure what the answer is here, nor am I sure what the future looks like when the very creation of a novel is now in question.”
