Artificial intelligence has moved through one industry after another. Finance, marketing, software development, law – each has had its version of the same conversation: What do we use this for, what do we keep human-centred, and who decides? And now, due to recent events that we’ll discuss in this article, publishing is in the middle of that conversation.
There are, naturally, complications surrounding the use of AI in publishing, partly because publishing doesn’t operate as just a “business”. The publishing industry is one that combines business with creativity and passion. There is a lot to be said about the implicit trust that exists between an author and a reader. And when readers move to pick up a book, they expect authenticity.
When AI enters industries like logistics and tech, the stakes feel different. When it enters the publishing industry, it feels personal. Maybe that’s because it is.
What’s Actually Happening in Publishing
The technology itself is no longer new. ChatGPT launched in late 2022. Since then, AI tools have become part of daily working life for millions of people across many fields. Writers have had access to the same tools as everyone else. And while some haven’t used them, many have.
Adoption of AI has been uneven. Some large publishers use AI tools for tasks like market analysis, manuscript sorting, plagiarism detection, and ad targeting. And we can see the appeal for streamlining these processes with the support of technological assistance.
However, a growing number of authors and editors are utilising generative AI tools for the core tasks related to the creation of the stories that readers hold dear, including developmental editing, defining plot structures, and even generating the writing itself. A Gotham Ghostwriters and WOBS LLC. survey of 1,190 writing professionals and 291 fiction writers found that 61% reported using some form of AI tools, whether in their writing, editing, or research, with 26% reporting using it daily. (Note this figure is not inclusive of fiction authors, who were analysed separately for the survey.)
At the same time, many writers won’t touch these tools, with 22% citing that they never use AI in their work. Only one in five fiction author state that they use AI daily.
In terms of the publishers, some have drawn hard lines. Several literary journals now require submission forms that ask writers to confirm their work is human-made. And many publishing houses require a statement of AI use and intent when authors submit their manuscripts for consideration.
However, the “yay” and “nay” responses to AI in publishing vary, and they don’t follow a simple pattern. You’ll find technophiles in small presses and AI sceptics in major publishing houses. What tends to divide people is a set of deeper questions about what writing is for, and who it belongs to.
The Key Concerns
The debate around AI in publishing keeps returning to three core issues:
1. AI Training Data
Most large language models learned from text scraped from the internet, including books, articles, and creative work that authors never agreed to share. Several class action lawsuits are working through United States courts right now. The Authors Guild has been vocal, and so have individual writers, with approximately 10,000 authors contributing to the recentDon’t Steal This Book protest.
The legal picture is not settled; the ethical picture is even less so. When a model learns to mimic a style by processing thousands of pages of someone’s work without payment or consent, the question arises as to whether this is a blatant act of intellectual theft or if it’s simply as faster-paced approach to the way human writers learn from reading.
2. Trust in Voice and Authenticity
Many writers would describe their voice as something that developed over years, sometimes decades, shaped by both their grasp of the language(s) they write in and their life experiences. So, while there is of course an economic argument, the more ethical concern lies in the fact that AI output, even when technically “competent”, flattens something. Due to issues like repeated sentence patterns, overuse of buzzwords, and often monotonous rhythm, it produces text without flow and without the author’s passion.
This view is perhaps subjective, but subjectivity matters in literary publishing. Readers form relationships with authors, and trust is a key part of their engagements.
3. Lack of Standardised Disclosure
Should a novel generated with AI assistance carry a label? What about a book where the author used AI to create an outline but wrote every sentence themselves? What about one where they rewrote 80% of an AI draft? Or one where the publishing house used AI tools to edit the work?
While some guidelines exist, nobody has agreed on a threshold. Publishers, book retailers, authors, and readers all want different things, and the industry has not moved fast enough to set norms.
The Recent Scandal: Shy Girl
Earlier this year, Hachette Book Group pulled Shy Girl, a horror novel by Mia Ballard, from UK shelves and cancelled its planned US release after an investigation concluded that large portions of the text were AI-generated. The book had been self-published, gained traction on social media, and was acquired by Hachette’s Orbit imprint.
Suspicion built slowly from social media users on Reddit and Goodreads, flagging the prose for hallmarks of AI generation. Then, a YouTube essay titled I’m pretty sure this book is AI slop drew millions of views.
Ballard denied personal responsibility, claiming an editor she hired during self-publication had used AI without her knowledge. She described the fallout as devastating to her mental health and said her name was “ruined for something [she] didn’t even personally do”. She has since taken her social media accounts offline.
The case raises questions the industry hasn’t resolved. AI detection tools are imperfect: Researchers have found that none achieve full reliability, and false positives can destroy careers. The book’s cancellation looked like a verdict, but Hachette made no definitive public statement on the underlying facts. So, whether Ballard is a fraudster or a victim of a rogue editor remains genuinely unclear.
What is clear, however, is that the publishing industry’s trust-based model is not built to handle this kind of ambiguity.
How Publishers Are Responding
The Shy Girl case is not an isolated incident. The Gotham Ghostwriters and WOBS LLC. survey found that 51% of fiction writers make use of AI, with 23% using it every day. For now, only 11% report using it to generate text content that is published without significant editing.
For traditional publishers, the challenge has increased in recent years. Some are revising contracts to explicitly prohibit undisclosed AI use. Others are investing in more thorough vetting of self-published acquisitions. Overall, most reputable publishers do not accept generative AI fiction, and books created primarily using AI are not protected under copyright law. In addition, some initiatives are being introduced to bring more trust to the reader, such as the Society of Authors in the UK launching a logo to mark books as “human authored”.
At a minimum, most publishers request that authors provide a statement of their use of any AI tools in the development of their manuscripts when submitting them for review. At Vindigo Press, we have our own AI policy on our submissions page.
What Needs to Happen Next
While detection technology will improve, so will AI’s ability to produce convincing prose. So as publishers (and authors), what we can control are our own standards and how openly we communicate them:
- The publishing industry needs a clearer, standardised framework.
- Contracts need explicit language around AI use and disclosure.
- Submission processes need to keep up with detection capabilities.
It’s important that the industry applies these standards consistently across publishers, retailers, and platforms, so that readers can make informed choices regardless of where they buy books.
This article reflects the views of our editorial team and is intended to open conversation. We welcome responses from authors, agents, and readers.
Want to be featured on our blog?
We’re inviting contributions from across the publishing community! If you’ve written a blog post or article on a book-related topic and are keen to share, feel free to send it to us to be considered for publication on our website.
- 500–1000 words
- Relevant to the industry
- Light editing may apply
Send your contributions to media@vindigopress.com for a chance to be featured!