This summer, the literary agencies Europa Content and Hodgman Literary shook the publishing world when they withdrew from sale a manuscript for the highly anticipated crime novel “Call Me, I’ll Hide the Body.” The book, written by Nigerian author Jerry Falade, had already generated plenty of online buzz. It had secured multimillion-dollar deals from major publishers in both the United States and the United Kingdom and was slated to be released in 2028.
It was also, the literary agencies said, unclear whether the book had been written by Mr. Falade or by an artificial intelligence chatbot.
The controversy is at once new and familiar, reflecting both the dramatic changes that the proliferation of AI has wrought on book publishing and echoing other skirmishes this year in which authors have been accused of using AI. This past spring, Hachette Book Group, a leading American publisher, canceled the publication of “Shy Girl,” an eagerly awaited horror novel by Mia Ballard, after rumors that Ms. Ballard had written the book using AI overtook the internet.
Why We Wrote This
Authors increasingly use AI tools to craft their work. As book publishers struggle to set guidelines for AI use, readers vent their concerns about transparency and quality. Some observers say a new social contract is needed between readers and authors to instill confidence.
Mr. Falade has publicly denied the accusations, arguing that Black authors’ work has faced more scrutiny for potential AI use than that of white authors. Ms. Ballard, who is also Black, also denied the accusations in an email to The New York Times, writing that someone she had hired to edit her novel had used AI but that she had not. But both controversies reflect a growing rupture of trust between readers, writers, and publishers, as AI use – alleged or proven – continues to rock the literary world.
“Until AI, there has been the expectation that no matter what is written in a book, It [was] authored by a person, or people,” says Vauhini Vara, a journalist and author who has written about technology for more than two decades. “There needs to be new terms in the social contract between reader and writer.”
Other AI-related controversies have consumed the literary world recently. In May, the Commonwealth Short Story Prize, a prestigious award for unpublished fiction, faced allegations that three of this year’s winning stories had been at least partially written by AI. Earlier that month, The New York Times reported that “The Future of Truth,” a nonfiction book about how AI is changing perceptions of truth, contained quotes that were misattributed or concocted by AI.
Those troubles, in an industry that is perhaps among the most vulnerable to AI’s capabilities, have raised questions not only about how many authors might now be using AI for some or all of their writing, but also about how thoroughly publishers check the books they release.
AI companies themselves may hold solutions. In 2024, The Wall Street Journal reported that OpenAI created a technology – known as watermarking – that allows content generated by ChatGPT to be detected with 99.9% accuracy. The company has not released the technology because it might decrease the number of ChatGPT users, the Journal reported, and because of ongoing internal debates about how widely to issue it and its reliability.
This month, Anthropic, the maker of the chatbot Claude, said that it will add similar watermarks to all of Claude’s outputs. Users will not be able to opt out, the company said. The move follows Anthropic’s decision to sign onto the European Union’s code of practice on transparency of AI-generated content. Several other AI companies, including OpenAI and Microsoft, have also signed the agreement.
When is AI OK?
Controversies over AI use have struck the publishing industry amid ballooning work and thinning staff. The number of books released by publishing houses in the United States increased by 10% between 2022 and 2025, according to data obtained by Publishers Weekly, rising to 642,242. Yet the number of book publishing jobs fell by 28% between 1987 and 2025, according to the Bureau of Labor Statistics.
Given those pressures, a wholesale ban on AI is unreasonable, says Kathleen Schmidt, a publishing strategist who has worked in the industry for more than 30 years. She uses AI for menial tasks, and doesn’t mind if authors do, too.
“I’m not saying that there aren’t instances where it should not be used – like, you should not use it to write an entire book,” she says. “Can you do it to brainstorm ideas? Sure.”
That seems to be the line that many readers draw, says Rebecca Joines Schinsky, chief of staff at Book Riot, an independent publication covering book publishing in North America. “I think a lot of people probably fall into a middle space of, ‘It’s cool to use it for research, but I don’t want to read any sentences in your book that were written by AI.’”
Yet for some readers, the creative process is sacred, from start to finish. In a 2025 YouGov survey that asked U.S. readers whether AI use was acceptable in five circumstances – ranging from “ideation” to “outlining” – majorities disapproved of AI use in all five scenarios. Sixty-one percent of readers said they would feel at least “somewhat” or “much” less fulfilled if they discovered a book they’d read was written by AI. Still, only 28% of readers said that AI use is unacceptable in all of the circumstances presented by the survey.
For Ms. Vara, the journalist, the problem boils down to honesty. When authors hide their AI use, the trust between readers and writers can erode, she says.
“There needs to be some kind of disclosure that makes it possible for readers and authors to have this mutual understanding,” she says. Her recent nonfiction book “Searches,” which examines AI’s impact on humanity through her own experience using it, contains several AI-generated passages – but they’re all labeled. “For me, the problem arises when AI-generated material appears in text, whether it’s a book or something else, and there’s no way for us to know.”
The promises and pitfalls of AI detectors
In both the “Shy Girl” and Commonwealth prize controversies, social media users pointed out writing habits they saw as proof of AI use – for example, an overreliance on metaphors, only some of which made sense. In “The Serpent in the Grove,” the short story that set off the Commonwealth conflagration, a line reading, “She had the kind of walking that made benches become men,” drew heavy scrutiny.
Some readers also ran the texts through online AI detectors. One such tool, Pangram, found that “The Serpent in the Grove” was 100% AI-generated. (The Monitor got the same Pangram result.)
The story’s author, Jamir Nazir, told The Observer, a British newspaper, that his writing process exclusively uses the text-to-speech function on his Android phone. The Commonwealth Foundation said in a public statement that Mr. Nazir and other accused writers denied using AI to write the story, and that the foundation had found no evidence of AI use after reviewing “working drafts, time-stamped documents and notes.”
But there’s no foolproof method of uncovering AI-generated content. Human writing may well contain errors or literary faux pas. Pangram’s creator says it has a false-positive rate – the rate at which it wrongly identifies a human text as AI – of just 0.01%. Yet a 2023 Stanford study found that AI-detectors were biased against non-native English speakers, who might have more limited vocabularies, and many AI researchers contend that such detectors are unreliable. (Pangram was not included in the 2023 study.)
Tara Radvand, a doctoral student at the University of Michigan who has studied AI detectors, says they do work – depending on the context. They’re better at identifying AI-fabricated fiction than nonfiction, she says, and reliability improves as text length increases. Nevertheless, she points out that even a false positive rate of 1% is high for an industry like publishing, where an accusation can send a career up in smoke.
Publishers’ AI rules remain murky
Authors debate whether – and how – they should be using AI in their work. Recently, Olga Tokarczuk, a Nobel Prize-winning Polish author, generated an uproar when she said she uses AI in her work. (She later added that she uses it only “as a tool for faster preliminary research.”)
Anthony Horowitz, an English author, made headlines this spring when he said he used the technology “all the time,” though he nevertheless stressed the need to “judge what you’re being told, and also look for a second source.” Coral Hart told The New York Times that last year she had published some 200 romance novels using Claude, an AI chatbot created by Anthropic.
In a 2025 poll of more than 1,200 authors by BookBub, a book discovery service, 45% of respondents said that they currently use generative AI to “assist with their work,” compared with 48% who said they do not use it and don’t plan to. Of those who do use AI, more than half said they use it to assist with writing.
Some publishers have sought to establish rules around the technology’s use, but few, if any, prohibit it outright. Hachette, for example, lists “use cases not requiring disclosure,” including “researching elements of a book,” but says that authors must attest that their work is their own. Penguin Random House recently made waves when it prohibited AI companies from using authors’ copyrighted work to train AI models without permission.
Still, publishers’ rules around AI use remain murky in many cases, says Ms. Schinsky, from Book Riot. Few publishing houses have made their screening methods public, she says.
For those readers who want a decidedly human experience, many authors remain happy to oblige.
Nick Harkaway, a British sci-fi author who describes himself as a lover of technology, nevertheless remains skeptical of the cognitive costs associated with using a machine to assist in thinking, and with the environmental impact of AI systems. “Ultimately, it doesn’t appeal to me because I love what I do,” he says, of using AI to write books. “I don’t want a computer to do my job.
“If I didn’t want to write, I’d do something else,” he says.
Editor’s note: This story, originally published on Aug. 12 has been updated on Aug. 13 to clarify the details of a 2023 Stanford study about AI-detectors and non-Native English speakers.
