An underground industry of AI-assisted romance writers is flooding Amazon with hundreds of titles a year, unsettling veterans and reshaping the economics of genre fiction.
The Invisible Juggernaut
Last February, a romance novelist who goes by Coral Hart began producing books at a pace that would have seemed implausible even in the most caffeinated corners of pulp fiction.
Using Anthropic’s Claude AI, Hart said she wrote — or generated — more than 200 romance novels in a single year. She published them on Amazon under 21 different pen names. None became breakout sensations. But together, she said, they sold roughly 50,000 copies and earned her six figures. During a Zoom interview with The New York Times, she completed a book in 45 minutes.
“If I can generate a book in a day, and you need six months to write a book, who’s going to win the race?” she asked.
Hart, who was already an established writer of explicit romance before adopting artificial intelligence tools, has since positioned herself as both practitioner and evangelist. She launched a business, Plot Prose, where she claims to have taught more than 1,600 aspiring authors how to use AI to produce fiction at industrial scale — including writers who had once publicly opposed the technology.
Certified Cyber Crime Investigator Course Launched by Centre for Police Technology
Yet even as she encourages others to “be shameless,” Hart has taken precautions. “Hart” is a pseudonym she uses for teaching AI courses. She publishes books under other pen names and keeps her legal identity attached to separate publishing and coaching work. She has cited the stigma surrounding AI use in creative fields as reason enough to compartmentalize her career.
Prompting the Machine
A significant portion of Hart’s instruction involves navigating chatbot guardrails — the safety systems designed to prevent the generation of explicit or risky content. According to reporting by The Times, she teaches writers how to coax bots past those constraints and how to refine the tone of their output.
She advises compiling an “ick list” — a catalogue of overused words and phrases the AI tends to default to — and instructing the system to avoid them. She encourages detailed prompting, including highly specific descriptions of sexual scenarios, to ensure the prose aligns with readers’ expectations in niche subgenres.
“Be shameless,” she tells students.
Romance publishing, particularly in its more explicit corners, has long relied on high volume. Readers’ appetite for serial content interconnected tropes, familiar character archetypes, rapid releases has created a business model that rewards speed. AI, for some, is simply an accelerant.
Hart’s method treats the machine less as ghostwriter and more as production engine: a generator capable of drafting plots, scenes and dialogue that can then be lightly edited and packaged for digital storefronts. The result is not necessarily singular works designed for critical acclaim, but a steady stream of consumable titles optimized for algorithmic visibility.
A Swamp Teeming With Titles
Veteran authors describe the impact differently.
Marie Force, a best-selling romance novelist, told The Times she was alarmed to learn that her novels had been used without permission to train AI systems, including Claude. She worries that the sheer volume of AI-assisted titles is distorting the marketplace.
“It bogs down the publishing ecosystem that we all rely on to make a living,” she said. “It makes it difficult for newer authors to be discovered, because the swamp is teeming with crap.”
On Amazon’s self-publishing platform, where discoverability already hinges on keywords, release frequency and reader reviews, a surge of rapidly produced titles has heightened long-running tensions over quality and saturation. The term “AI slop” has entered common usage among critics who argue that algorithmically generated fiction is crowding digital shelves.
For emerging writers without access to marketing budgets or loyal audiences, the proliferation of inexpensive, quickly assembled e-books presents a structural challenge. Where once the bottleneck was finding a publisher, now it is cutting through a deluge of content.
Yet the economics are difficult to ignore. Even modest sales, multiplied across hundreds of titles, can yield meaningful income. The strategy does not require a single runaway hit — only steady, cumulative performance.
Prolific Then and Now
Literary history contains its own legends of productivity. At the height of his career — and, by some accounts, his amphetamine use — the science fiction author Philip K. Dick wrote around 30 novels in two decades, along with hundreds of short stories. Works like “Do Androids Dream of Electric Sheep?,” “The Man in the High Castle” and “A Scanner Darkly” reshaped the genre and, eventually, popular culture.
Such output once inspired awe. How could so much emerge from a single mind?
In the age of generative AI, the question has shifted. Speed and quantity no longer signal singular obsession or stamina; they may simply indicate computational assistance. Chatbots can draft in seconds what once required months of drafting and revision.
For some writers, that represents democratization — a tool that lowers barriers and expands creative possibility. For others, it signals an erosion of craft and a marketplace destabilized by abundance.
Hart frames it as a race defined by efficiency. If one writer can produce a book a day and another needs half a year, the competitive calculus changes. Whether readers will ultimately distinguish — or care — remains uncertain.
What is clear is that romance publishing, long shaped by voracious demand and rapid cycles, has become a testing ground for how artificial intelligence may reorder creative labor. In the space between pseudonyms and prompt engineering, a new model of authorship is quietly scaling — one 45-minute novel at a time.
