370 Novels, 21 Pen Names: The AI Flood Hitting Publishing

One romance writer has produced 370 novels under 21 pen names since 2025 using AI, earning what she calls "high six figures" on Kindle Unlimited. Coral Hart, a former Mills & Boon author who moved to self-publishing full-time in 2015, disclosed the figures in an interview from her home in South Africa.

The scale is the story. A study led by Tuhin Chakrabarty of Stony Brook University, still awaiting peer review, estimates that roughly one fifth of recently self-published books contain at least 25% AI-generated text. Hart's output is rewarded by Kindle Unlimited's algorithm, which favors writers who release a new book at least monthly. She says she knows of authors who hire more than a dozen ghostwriters to keep pace; she can write 10,000 words a day by herself, and AI pushes her well beyond that.

Hart insists the craft still matters. "A bad writer or a bad storyteller is not going to get a good book out of AI," she said. "It's not taking the human out of the equation. It's taking the friction and the time and the business blocks out of the way." She now teaches courses on AI-assisted book production — and advises authors not to disclose their AI use unless the law requires it. After she spoke to the New York Times earlier this year, she was doxed and received death threats.

What is the money and legal exposure?

Last year, a class action covering around 500,000 titles won a $1.5 billion settlement from Anthropic over alleged use of writers' work without consent. Similar lawsuits are ongoing. US copyright law adds a commercial wrinkle: a book must be "human authored," and AI-generated text is considered public domain.

That gives publishers a hard financial reason to care, as Natalie Lakosil, a literary agent and chair of the copyright committee at the Association of American Literary Agents, explained: "If it's AI, they have no protection, they can't have exclusive rights." She believes the "human authored" requirement will survive court challenges — "the purpose of copyright protection is to protect human creation, to protect artists, and AI is not an artist" — while conceding the law must eventually accommodate writers experimenting with the technology.

How did the scandals unfold?

• March: Hachette withdrew the romance novel Shy Girl over evidence author Mia Ballard used AI. Ballard claimed an editor she hired may have inserted the AI text.
• This year: Commonwealth short story prize winner Jamir Nazir was accused after readers flagged phrases like "she had the kind of walking that turned benches into men." He denies it. Pangram, the detection tool, later concluded four regional winners since 2025 likely used AI.
• July: Agents withdrew from a $2m (£1.5m) debut deal with Jerry Falade over AI accusations. He denied using AI for anything beyond research — and has since signed with a new agent.
• Last month: Thélyson Orélien's novel was struck off the Goncourt longlist after AI-detection scans. He denied using AI but admitted plagiarism in earlier work; his publisher Boréal dropped him in late October.

Every writer at the center of these scandals has been an author of colour. Falade, who declined to speak on record, wrote that scrutiny seems reserved for non-native or "culturally unfamiliar" English: "apparently that level of scrutiny is reserved for a Black boy from Nigeria." By contrast, white author Steven Rosenbaum's The Future of Truth contained fabricated AI-generated quotes he had disclosed; he expressed regret, and the book was not withdrawn.

Detection is imperfect. Pangram is estimated to be at least 95% accurate with a near-0% false-positive rate, but human judgment is not flawless, and evidence suggests non-native and neurodivergent writers attract false suspicion more often.

Where does the industry stand?

A literary agent described mainstream publishing's atmosphere on AI as "the wild west." Seventy prominent authors signed an open letter urging Penguin Random House, Hachette, HarperCollins, Simon & Schuster and Macmillan to "never release books that were created by a machine." The publishers did not promise.

That silence may be strategic. Tess Callero of Europa Content asked whether the big five "see a future in which potentially these books could be monetised." Thad McIllroy, author of The AI Book Revolution, said publishers need "brave," extensive AI policies; he was surprised Hachette cancelled Shy Girl rather than issuing a disclosure, noting the rumors predated the acquisition and the text was heavily edited after purchase.

Hannah Bowman, chair of the AALA's AI special committee, warned that public detector-led scandals deny authors a fair defense. Agents now advise writers to save multiple drafts and avoid sending PDFs.

Stephen Marche, whose 2023 AI-assisted novella Death of an Author was reviewed by the New York Times and released only as an audiobook due to copyright concerns, argues the moral line is labelling: careful users are "totally excluded. But you know the scammers will find a way." A new French prize, the Prix Pascal de Littérature IA, launched last month for fiction writers working with AI.

One more Chakrabarty study, also awaiting peer review, found MFA students in blind tests tended to prefer fine-tuned AI passages over their colleagues' writing. Doug Wallace of Andrew Nurnberg Associates framed the industry's stance plainly: "We're in the human creativity business, not in the business of giving robots jobs." Whether that holds — and whether readers can tell the difference — remains the industry's unresolved bet.