top of page

Jane Endacott | Finance Writer

Can You Prove You Wrote Your Own Book?When Publishers Stop Trusting Authors

Jane Endacott
Aug 11
4 min read

Updated: Aug 18

Jerry Falade’s debut crime novel, Call Me, I’ll Hide the Body, sparked a 14-way auction in the U.S. and landed a two-book deal reportedly worth more than $2 million, along with a separate U.K. deal and interest in screen rights. Then questions arose about whether AI had been involved in writing the manuscript. Falade’s agent withdrew the book and ended his representation after saying aspects of Falade’s explanation had changed. Falade denies using AI to write the novel and says he used it only for research.


Mia Ballard’s Shy Girl was originally self-published. The novel was acquired by Hachette, published in the U.K. and scheduled for a U.S. release before allegations of AI-generated writing prompted Hachette to review the book, cancel its U.S. publication and discontinue the U.K. edition. Ballard denies using AI to write the novel and has said an editor she hired may have used it without her knowledge.


I’m not here to decide whether Falade or Ballard is telling the truth, or whether their publishers and agents were justified in walking away. Publishers have every right to enforce their contracts, and evidence about plagiarism, authorship or provenance can emerge long after a book has been acquired. What interests me is the larger problem these cases expose.


If AI-related allegations can cost an author a book deal or a career, what exactly is the standard for “AI writing”? 


What evidence establishes that the standard was violated? And what opportunity does an author have to answer the accusation? Generating a novel from a prompt is an obvious case. The boundaries get less obvious from there. The Authors Guild distinguishes between AI-generated text and limited uses such as research, spelling and grammar checks.


There is also a business reason for caution. Publishers invest in intellectual property and need confidence that the rights they acquire are protectable. The U.S. Copyright Office says copyright extends only to human-authored expression. Human modification of AI-generated material may qualify when the contribution is sufficiently creative, but the Office applies that standard case by case. There is no simple rule for how much rewriting is enough. That uncertainty helps explain why a publisher might want a bright-line restriction on AI-generated prose. 


The problem is that the bright line for the publisher can become fog for the author. What must be disclosed? What counts as generated text? What if an author uses AI for research, structure, or editing? What if an editor uses it? The Authors Guild's model contract language recommends disclosure of AI-generated text and suggests negotiated limits, but it is guidance, not a universal industry standard.


The timing leaves an obvious question about due process. Hachette described its review as “thorough and lengthy,” but publicly announced its decision the same day The New York Times approached the company with evidence, and it remains unclear when that review began, what evidence it considered, or what opportunity Ballard had to respond. Jennifer Weltz, president of the Jean V. Naggar Literary Agency, told Publisher’s Weekly that the Ballard controversy showed that clearer communication and transparency about AI use are needed across publishing to avoid similar disputes. When the consequence is cancellation of a book and potentially lasting damage to an author’s reputation, transparency about that process matters. 


The American Association of Literary Agents says authors have a responsibility to fully disclose how generative AI was used because agents need that information to evaluate originality, copyright, contractual obligations, and compliance with publisher policies. If authors are expected to disclose whether and how they used generative AI, they should also be able to understand the standards by which that use will be judged, what evidence can establish a violation, and what process exists to challenge that conclusion. Without that, authors are being asked to comply with rules whose enforcement remains opaque.


Publishers Weekly quoted researcher Rachel Noorda asking how innocent authors will prove the "humanness" of their writing as accusations and cancellations become more common. Authors can spend years producing manuscripts before an agent or publisher invests anything. An authorship accusation can jeopardize a contract, years of uncompensated labor, and a writer's reputation. Traditional publishing asks authors to surrender control in exchange for institutional benefits. If I am going to spend years producing a book, hand control of it to another institution, warrant that it is my work, and depend on that institution for my career, I need confidence in the process that determines whether I violated that warranty.


Agents and publishers involved in these controversies have emphasized trust, honesty, and transparency, which are reasonable expectations in any business relationship. But trust cannot run in only one direction. If institutions demand transparency from authors without offering the same in return, writers have reason to question whether those relationships truly serve their interests.


Written with AI assistance; edited and fact-checked by the author.


Photo Credit: Igor Omilaev from Unsplash.


Sources

 
 
 

Comments

bottom of page