Is it legitimate to use AI to write a book?

The premise is unsettling and attention-grabbing: a young woman in precarious financial circumstances agrees to live as if she were the pet of a lecherous rich man. The story unfolds in Shy girl, a horror novel self-published on Amazon by an unknown young author, Mia Ballard, in early 2025. After gaining traction on TikTok, Hachette Book Group bought the rights and published it in the United Kingdom in November 2025. A controversy then erupted on social media. A literary analysis of the book by Pangram, one of the tools most used by scholars, claimed that 70% of the novel had been written by artificial intelligence. The wave of outraged readers helped push Hachette, which says it defends “creative originality,” to withdraw the book after some 1,800 copies had been sold. Ballard’s attempt to defend herself — saying the editor she hired to revise the novel was the one who used AI, not her — did little to help. Having lost credibility as a writer, her career now appears to be irreparably stalled.

Polarization of opinion is forcing publishers to take a stance. Some argue that if publishers stopped putting out bad literature, AI would not be a problem; but on some Reddit forums the most radical skeptics say they no longer read literature published after 2020, an unfair stance toward a new generation of writers.

The worrying thing is that it’s impossible to know with certainty whether a novel was actually written by a machine, despite AI detectors boasting false-positive rates of only 1%. “If you write in a language in which you are not native, your personal false-positive rate can be much higher, which exposes you to unjust accusations,” Tim Requarth, a neuroscientist at New York University and author of the Substack newsletter The Third Hemisphere, explains in a café. “The harm of those accusations falls hardest on people who are already more likely to have their credibility questioned.” Continuous exposure to AI in the ecosystem we live in is, moreover, homogenizing modes of expression. “AI sounds increasingly like a human, but humans are sounding increasingly like AI.”

According to Canadian writer Stephen Marche, authors today feel they have only two options: to not use AI (which he considers backward-looking) or pretend they don’t use it. Acknowledging its use sparks controversy. That was the case with Nobel laureate Olga Tokarczuk, who after saying she had relied on AI to develop her latest novel immediately had to clarify that she had used it only as a tool to research and verify facts more quickly, not in terms of creativity.

“Everyone is or will be using AI simply by searching on Google. It’s excellent for data processing and as a research assistant; it quickly provides documentation such as what music Japan’s bourgeoisie listened to in the 19th century,” writer Miguel Ángel Hernández says by phone. “For me the red line is using it for creative writing. I don’t find it ethical or sensible, because it works with what has already been written, with conventions, and it tends to homogenize sentence structure. It flattens language and generates a style. It pulls sentences into a universe that no longer belongs to the author.”

Others say the problem is not co-authorship with machines but a lack of transparency. Under the European AI Regulation approved in 2025, AI-generated texts must be labeled as such. There are also certifications or logos that claim to guarantee authors are human; the problem is there is no reliable verification beyond the author’s word.

Marche, who wrote his first story with AI in 2017 for Wired, published Death of an Author in 2023, a novel he developed using AI tools such as ChatGPT, Sudowrite, and Cohere beyond mere cut-and-paste. “My prompts were very elaborate and I then edited the responses with Sudowrite; or I wrote with a preset model I created in Cohere,” he explains by email. Marche considers himself 100% the author of the work, since for him AI is a resource “like a camera is for a photographer.” His novel became the first AI-assisted book reviewed in The New York Times, which called it a feat. Yet three years later it has become an uncomfortable precedent. “After that, the literary world practically made it impossible to publish works created with AI,” Marche continues, stressing that both young writers and editors are increasingly using AI. “The literary world has created a situation in which it refuses to acknowledge the power of this new technology while it erodes the very foundation of writing.”

Italian philosopher Andrea Colamedici did something similar to Marche when he used AI to create Hypnocracy: Trump, Musk, and the New Architecture of Reality, which he published under the invented name Jianwei Xun, deceiving an entire industry. In that light, it’s worth asking what Emilio Carelli, director of L’Espresso, posed amid the controversy: “If the book’s theses are correct — or at least have managed to spark an intense cultural debate — what does it matter that they were written by artificial intelligence?”

The debate has also reached the theater: Molière Ex Machina is a play sponsored by Sorbonne University in collaboration with Théâtre Molière Sorbonne, Obvious Collective, and Mistral AI, written with AI assistance and playing with the idea that Molière might have authored it had he lived another year. Pierre-Marie Chauvin, vice president for Arts, Sciences, Culture and Society at the university, says on its website that the goal is to “demystify AI,” because “it does not replace the human being, it becomes a co-creator.” But after its premiere at the Royal Opera of the Palace of Versailles on May 5, 2025, some audience members said it could have been written entirely by humans — so to what extent was collaboration with AI necessary?

Most publishers have yet to take a position on possible co-creations. “I have never encountered an author who confessed to using AI to write their book,” Judith Feher-Gurewich, founder of U.S. publisher Other Press, tells EL PAÍS by phone. “If they’re trying to invent a style experimentally, I might consider it. Honestly, I don’t know how I would react.” She adds: “But I publish unique voices. I’m convinced Antonio Muñoz Molina (one of her authors) would never use AI to help write; it would confuse him immensely. He follows his thought.”

According to Marche, “if AI is used in fascinating and powerful ways, it will be fascinating and powerful. If it’s used in a boring, hackneyed, lazy and corrupting way, the work it produces will be boring, hackneyed, lazy and corrupt.”

Sign up for our weekly newsletter to get more English-language news coverage from EL PAÍS USA Edition