Plot twist Publishing is grappling with that increasingly complicated question as major deals collapse over suspected AI use, manuscripts face machine-detection tests and the industry struggles to decide where assistance ends and authorship begins.
Peter PARKS/agence france-presse
Share
Artificial intelligence (AI) is forcing the book publishing industry to confront a fundamental question: How much AI can an author use before a work is no longer considered their own?
Major book deals have recently unraveled over allegations of generative AI use, exposing an industry struggling to establish common rules for authors, agents and publishers, Wall Street Journal reports.
In July, agents for debut author Jerry Falade pulled support for crime novel “Call Me, I’ll Hide the Body” after saying they could not verify that the manuscript had been wholly written by their client. Falade has denied allegations of AI use and called them racially motivated.
Hachette Book Group also canceled the US release of Mia Ballard’s “Shy Girl” in March following allegations that AI was used during the writing process. Ballard denied the claims, saying an acquaintance who edited the book had used AI.
The controversy has placed agents and editors on the front lines. Some reject authors who use AI even for research, while others use AI-detection tools to screen manuscripts despite questions about their reliability.
The scale of AI-assisted publishing may already be significant. A study led by Stony Brook University professor Tuhin Chakrabarty found “substantial” AI assistance in 20 percent of Amazon ebooks in its data set.
Publishers themselves have yet to settle on an industrywide standard. An executive at one of the Big Five publishers said the company decided against adding specific AI language to contracts because the technology and its use are evolving rapidly.
The Authors Guild has introduced a “human authored” certification for writers who attest that their books were written by humans, although critics question whether an honor-based system is enough.
For an industry historically built on trust between authors, agents and editors, AI is turning that trust into another vulnerability.
“The standards around it don’t exist right now,” literary agent Amanda Annis said. “There’s just so much gray area that hasn’t come out in the wash yet.”
