Most self-published authors can only dream of writing something as popular as Daggermouth. Set in a surveillance state ruled by a masked elite, the novel chronicles the unlikely romance between the president’s son and an assassin hired to kill him. Since HM Wolfe uploaded Daggermouth When the Kindle e-book appeared on Amazon, it was a viral hit. “Am I finished? Daggermouthor did Daggermouth Are you killing me?” gushed one TikTok book influencer. In February, Simon & Schuster paid a seven-figure sum to secure publishing rights for the book and its sequel, due out next year.
Daggermouth spent months on it USA todayIt is ranked No. 1 in the “Science Fiction Romance” category on Amazon’s bestseller list. It also tops a less prestigious list compiled by scientists studying AI writing. Among thousands of self-published Amazon e-books that they said contained “substantial AI text,” none was more popular than Daggermouth.
Until recently, it was easy to dismiss AI fiction as sloppy. On Amazon, new book releases almost tripled from 2022 to 2025. (This is probably not because people have suddenly embraced the written word en masse.) Many are blatant copycats of existing titles, apparently designed to entice readers to buy them. Their sales figures are generally poor; the reviews, outraged. A suspicious number of them feature protagonists with the same name. Last year, readers of a fantasy novel noticed a line that appeared to respond to a request for a chatbot to write in the style of another author. But when a book is as popular as Daggermouth written with a chatbot, it would raise an uncomfortable question: Could AI fiction be… good?
In a new study, researchers randomly selected more than 14,000 Kindle e-books and ran them through Pangram, an AI recognition tool. One in five came back as if they were more than 25 percent AI written; Daggermouth According to an underlying data set the authors shared with me, the return was 60 percent. They admit that these numbers tell us only so much. The researchers grouped texts that Pangram had labeled as either “AI-generated” or “moderately AI-powered” (the latter of which can mean that AI was used to edit or polish human writing, or vice versa). And while Pangram has a low false positive rate, it’s not perfect.
Therefore, the authors of the study, led by Tuhin Chakrabarty, professor of computer science at Stony Brook University, developed another check. They searched the text of each book for “rare expressions” – strings of at least five words in a specific order – that also appear in other suspected AI books, but almost never in books written by humans. Daggermouth had many of them. For example, near the end of a multi-page sex scene in Chapter 27, the following sentence appears: “They collapsed, a tangle of sweaty limbs and racing hearts.” The same sentence appears word for word in another self-published e-book by another author. Or consider this line from chapter 38 of Daggermouth: “Her pulse thundered in her ears, drowning out the murmur of the crowd.” It is also included in an e-book by another author who has published more than 400 different titles on Amazon over the past three years.
The study has yet to be peer reviewed, and to be clear, there is no definitive proof that Wolfe used AI. “The suggestion that I used generative AI to write Daggermouth is completely untrue,” Wolfe said in a statement through her attorney. “I wrote this book myself. I have been open about my opposition to generative AI and the impact it has on writers, artists and the creative community. I do not believe it belongs in the writing process.” Wolfe noted Pangram’s limitations, adding that “allegations like these have real consequences for authors and should not be made based on technology that is known to be unreliable.” Susannah Lawrence, a spokesperson for DaggermouthSimon & Schuster, the publisher of , told me in an email that the company was behind the book, noting that it went through “the same editorial and production process as our other published titles.” Neither Wolfe nor Lawrence addressed my questions about “rare expressions.” Daggermouth Shared with other current e-books.
Chakrabarty, who works on AI writing, told me this Daggermouth has “too many telltale signs” of AI to believe this explanation. He gave Chapter 18, “Thanks” as an example: his research team divided it into 12 parts for Pangram to evaluate, and the tool classified all parts as AI-generated. (In my own sampling, the tool classified the first few paragraphs as written by humans and the rest of the chapter as 100 percent AI.) I ran this by Mohit Iyyer, a computer science professor at the University of Maryland who was not involved in the study. Iyyer, an expert in AI speech recognition, told me that even taking Pangram’s fallibility into account, it is “statistically almost impossible” for a book written entirely by humans to register as 60 percent AI. He added that the study’s broader methodology, which links AI writing to e-book sales, was solid.
Whether you believe Wolfe deserves the benefit of the doubt or not, the same question arises as to how Daggermouth was written apply to a growing number of other hits on Amazon. Chakrabarty’s study found that books written with “significant” AI assistance accounted for 20 percent of Amazon’s e-book catalog, 12 percent of sales and 10 percent of bestsellers in their genre.
Still Daggermouth is not only characterized by its popularity. The second best-selling e-book in the data set is Colorized Athenaby Nicole Fox, an author whose 418 titles are listed on Amazon. DaggermouthIn contrast, it appears to be one of only two novels to Wolfe’s name in 2025, and its sequel is not due out until 2027 – an indication that Wolfe’s process likely involves significant amounts of genuine human craftsmanship. Iyyer said that using AI in a way that results in a novel that is actually fun to read requires “a tremendous effort from a human author.”
Another difference: Jennifer Prokop, who reviews books for Kirkus and is a co-host Companions in fatea popular podcast about romance novels, told me that people in her world keep talking about it Daggermouth. “Nicole Fox, I have never heard a single soul speak about her,” she said. (Fox did not respond to a request for comment.) Prokop expects mechanical prose to appear primarily in self-published e-books, whose dominant platform is Amazon’s Kindle Unlimited subscription service. Just because an e-book comes out on Kindle doesn’t mean people will devour it. Scammers can now use AI to launch dozens of titles and then use all sorts of tricks to lure readers, such as buying hundreds of fake five-star reviews and paying click farms to boost their metrics. “If you can get 10,000 people to read 10 pages of your book and you release 300 books,” she said, you can make real money. “But you don’t actually write romance novels that people want to read.” (Amazon did not respond to a request for comment.)
If Daggermouth is AI, then perhaps it’s just an outlier – a book that caught the perfect wave of social media hype. (A series of controversies surrounding the book, including allegations about the author’s use of AI and a backlash over the cover of a special hardcover version, seemed to excite fans and bring the book even more attention.) The idea that AI writing requires a perfect confluence of events to reach the top of the charts is a comforting thought. But it might not be true. “Everyone thinks they can narrate AI writing,” Christine Larson, a journalism professor at the University of Colorado at Boulder who studies the impact of technology on creative work, told me. “But I don’t think we can say that much.” Several writers and editors have told her privately that they are using the technology to speed up or improve their production. “I fully believe that a human working with AI can write something compelling,” she said. “I have absolutely no doubt that we will see more of this in the future.”
If she’s right, it’s not about whether we’ll see AI bestsellers. It’s about whether anyone involved with them will admit it. Coral Hart, a South African author who uses AI to publish hundreds of books every year under various pseudonyms, told me that she works hard to keep her pseudonyms secret after receiving death threats and thousands of one-star reviews after one New York Times story about her. Hart claimed that successful writers had taken the AI writing courses she taught, but declined to name them. “Until the bullying, witch-hunting and expulsion stops, I don’t think the writers will be transparent, and I don’t blame them,” Hart said. “There’s no way I’m going to do that.”
The taboo against the use of AI in literature is understandable: among other things, it is wrong to take credit for a book that you did not write yourself. Avoiding authors suspected of using AI might work if there was a way to tell for sure – which there isn’t. This shameful ritual places a moral burden on individual authors rather than the giant corporations responsible for the system in which they try to make a living. It also tends to erase any distinction between pure AI nonsense and human writing assisted by AI. The result is a situation that seems untenable: AI writing can be lucrative, but only if it is kept secret. At some point something has to give.
https://www.theatlantic.com/technology/2026/07/daggermouth-novel-bestseller-ai/688067/
