AI-Generated Books Flood Amazon as Publishing Industry Struggles to Detect Chatbot Authors

The new premium product: books written by people

Earlier this year, author Steven Rosenbaum took a phone call that turned his world upside down. The book he had spent three years researching and writing, he was informed, contained quotations fabricated by AI tools. "When I learnt there were factual inaccuracies in The Future of Truth, my first reaction was: that's impossible," Rosenbaum tells the FT two months after the discovery. "It had gone through a traditional publishing process with multiple rounds of editing and review." Rosenbaum had been clear about the use of AI tools in his nonfiction book -- ironically, a study of the future of truth and technology -- but had used them only to research and organise material, including summarising source documents and interviews. "What I failed to recognise was that, in some cases, it transformed those summaries into what appeared to be direct quotations," he says. "The danger isn't that AI writes bad prose. The danger is that it writes persuasive prose attached to facts that never existed." The rapid development of AI is changing the way the $140bn publishing industry commissions, edits and distributes books. Unauthorised use of copyrighted material by AI companies has also become a crucial concern. Last week, Anthropic agreed a $1.5bn settlement with authors and publishers after using over 480,000 published works to train its large language models without payment or permission. But it is also the deployment of the technology in the very creation of literature, and the threat that AI could replace human authors, that is worrying publishing -- a business that has operated in similar ways for decades if not centuries. The book trade is deeply split over how to handle a technology that some publishers and authors vow not to touch but which others identify as a huge opportunity that is already changing reading and listening habits. "Every week we see a book that we suspect involves AI," says one senior book executive. "The grey area between generic writing and AI writing is difficult to judge. We don't know how much we unknowingly publish." Agents say they too are being swamped with AI-generated prose. "We see it show up in people that are submitting books to us," says Christy Fletcher, co-head of UTA Publishing, who represents authors such as Daniel Mason, Eric Ries and Maggie O'Farrell. "It's really transparent. You now have to train yourself to ask: does this read like it was written by [Anthropic chatbot] Claude?" In March, Hachette, the French publisher, was forced to pull Shy Girl, a horror novel by Mia Ballard, from the shelves in the UK and cancel its US launch following allegations it had been partly written by AI. Ballard has denied the claim. The novel had built up a following as a self-published book before being picked up by Hachette. "Everyone liked it," says one book executive. "After Shy Girl, every editor is nervous." Even now, Ballard's book has an average of over three stars on Good Reads, a book reviews website. One author, talking off the record, voices the industry's ultimate fear: "What everyone is really afraid of is that the audience might not care." 'Hallucinations aren't typing mistakes' Executives at other publishers say it is just a matter of time before they are hit by a Rosenbaum or Ballard-type scandal and are scrambling to try to set rules as the technology advances. "AI has seeped into absolutely every nook and cranny of software," says Madeline McIntosh, co-founder and chief executive of the independent publishing house Authors Equity and a former US boss of Penguin Random House. "Unless what you are saying is 'you may only use pen and paper', the fact is you are touching AI -- and AI is touching your work," she adds. Existing guidance from publishers to authors broadly suggests that AI can be used to assist the creative process -- researching, fact-checking and sometimes even editing and refining -- but not used to actually generate copy. But the extent to which AI can assist in writing a book varies depending on the publisher, the editor and the agent. "There are many, many shades of grey at this point," says McIntosh. Fletcher describes a "schism", especially between fiction, where there is "more dogma" over AI use, and non-fiction, where its value for research is often prized. "What level of AI use is acceptable?" asks another executive in the sector. "What does research constitute? What does 'written with AI' mean? What percentage needs to be written if so? It's a slippery slope." Even among authors, there are divergent views. "We've got 12,500 members, and each of them is going to have slightly different red lines," says Anna Ganley, chief executive of the UK's Society of Authors. Publishers are now inserting AI clauses into contracts. Pan Macmillan says its standard contracts require authors to guarantee the work is entirely their own, original and human-authored, which also ensures copyright is protected. "If an author or illustrator wants to experiment with new AI tools, we're completely open to having a conversation about how to do that responsibly," says a spokesperson for the publisher. That might include using AI to aid research, check facts, edit grammar, polish text, summarise source material or speed up the writing process. But the author's "judgment, voice and intellectual responsibility must remain at the heart of the book". It has provided guidance for its editors on how to spot possible AI use, and as yet no book has failed to pass its tests. Bloomsbury also prohibits the use of generative AI in its publications, though its policy for authors says AI "may offer helpful solutions in some contexts, particularly when used to support, rather than replace, human creativity". Others are taking a more hardline approach. Microcosm Publishing, a Portland-based independent publisher and distributor, says that its AI policy is "no, thank you . . . not even spelling and grammar edits. Not reading level. Not continuity, and not fact-checking. Not adding citations. These apps will do far more than you ask them to." Rosenbaum is painfully aware of this risk of AI mission creep. "Hallucinations aren't typing mistakes," he says of the perception by an LLM of a pattern or object that does not actually exist. "They're authoritative-sounding inventions. That's a completely different editorial challenge . . . leaving authors and publishers to shoulder the responsibility for detecting hallucinations after the fact." The AI doom loop Many in the industry believe they can detect AI writing. But executives admit that with careful editing, an experienced writer could still pass off AI prose as their own. Poor style on its own is not enough to identify AI usage. "Lots of books are badly written," one executive shrugs, adding that the inadvertent tiny imperfections and deliberate creative breaking of the rules are what can make people feel the humanity behind the writing. Some even worry about a doom loop for future generations, in which younger readers become so accustomed to reading AI-generated copy, with all its flaws, that they start to write in that style too. Publishers have deployed AI-detecting technology but are generally dissatisfied with the results. "We have tried AI-testing tools but don't always use them. The threat is seen as enough for now and they aren't very good," one chief executive of a publisher says. The Authors Guild in the US has warned that it would be a material contract breach to cancel a contract or pull a book from sale on the basis of accusations or use of detection tools alone. Given the difficulties in defining rules and policing them, and the risk of errors, McIntosh of Authors Equity says transparency has become key to the conversations that they are now having with authors. "In our contract, we and the author both need to disclose to each other what our use of AI was," she says. "If an author has used AI to help support their research for a nonfiction book, that's fine. But I want to know that they did, because that means that we now need together to apply another layer of human research to make sure that the facts are actually true." UTA's Fletcher says that her author clients often want reassurances about how AI is being used by publishers, particularly that "editing is not using AI, cover design is not using AI". She adds: "They really care that . . . it's not being handed over to software." Pan Macmillan says editors can use secure AI tools to quickly summarise a long manuscript, list character details or pull out strong quotes to help pitch a book. They cannot use it for editing or deciding what to accept or reject. Many publishers also consider AI useful in sales and marketing tasks, such as automatically refreshing metadata so that books appear more prominently online, and in the printing and warehousing processes. Bonnier Books has run author and agent roundtables on AI in an effort to find collective solutions. "The only way through possibly the greatest transition the industry has seen in modern times is open conversations," says Sarah Posner, publishing innovation manager at Bonnier's UK arm. "We need to work together." Chatbots vs books The challenges of AI extend beyond determining who or what wrote a book. The technology is also beginning to change what readers buy and how publishers operate. Sales of more functional non-fiction books in areas such as self-help, cooking, biography and finance are being hit as more people turn to chatbots for practical advice. Non-fiction sales fell 6 per cent last year in the UK alone, according to Nielsen data. While this relates to a variety of issues, including the popularity of podcasts and YouTube videos that cover similar topics, publishers think there is a direct correlation with the rise of chatbots. "People will want to just parse a bestseller book," says one senior executive. "They'll ask ChatGPT 'tell me all the things I need to know about [decluttering guru] Marie Kondo'." Another executive says certain non-fiction genres, such as history and true crime, skew towards male readers "and men also love ChatGPT". The result, according to industry executives, is an even greater need for authors such as Patrick Radden Keefe, whose brand of accessible investigative journalism has proved a hit with critics and readers. "Where we're seeing strength is people who have really strong singular voices, and those readers are identifying with those individuals," says Fletcher, the agent. Authors can also see the use of AI beginning to affect fiction categories. Nilanjana Roy, an FT Weekend columnist and author of The Wildings and The Hundred Names of Darkness, says it is "pointless trying to hold AI literature at bay", adding that a "younger generation will experiment" and established writers such as Vauhini Vara and Rie Kudan "have already played with AI/LLMs to generate fairly fascinating texts". But she suggests that as large volumes of AI-enabled texts threaten to overwhelm agents, editors and publishers, "you have the ironic but also delicious prospect of an AI assistant sifting through AI manuscripts". The influence of AI is most visible in the self-publishing market, which is going through a boom that many publishers link to the use of tools that make writing a book much easier. New books with ISBN numbers in the US rose by a third last year to over 4mn, with more than four-fifths of these self-published. Many such authors are able to use AI tools to generate a book a day with little use of their own creativity. Amazon's Kindle Direct Publishing, the biggest self-publishing platform, asks its authors to self-report AI-generated content, but they are not required to disclose AI-assisted content. Amazon tells the FT that it "continues to enhance our protections against non-compliant content", pointing to rules such as limiting the number of titles authors can create to 10 books per format each week. It takes "proactive and reactive measures to prevent, detect and remove content that violates . . . guidelines, whether AI-generated or not". AI-created synthetic voices can also be used to produce almost instant narrations of books for social media platforms, removing the need to buy and download an official audiobook, according to one top executive, who says synthetic voice tools are now so good that they can copy the voices of established authors, who would typically be asked to narrate their own novels for audiobooks. 'Human authored' Trepidation about the impact of AI on publishing is not universal. Some sector executives believe that after an initial surge of AI-generated slop, there will be more demand for quality human writing. "Books aren't going anywhere," says Posner, of Bonnier Books. "In fact, I see people reaching more and more for real human experiences." The industry has survived a number of other disruptions, from self-publishing to ebooks, that were considered existential threats when they appeared. Tom Weldon, chief executive of Penguin Random House UK, part of the world's largest English-language publishing group, says books have "proved remarkably resilient through every technological shift because they offer something unique". "AI can be powerful in supporting creatives in many ways, but should never replace the imagination, judgment and lived experience of an author," he adds. But there is now greater pressure on publishers to find what McIntosh of Authors Equity describes as the "work that is exceptional, not the lowest common denominator . . . because [what] AI does really, really well is the lowest common denominator". She refers to the "thumbprints of humanity" that make a book stand out, a concept being practically applied by publishers in the UK and US amid efforts to clearly label content as human or AI-assisted. Ganley, of the Society of Authors, says transparency "is key, because whether it's organic literature or AI-assisted literature or AI-generated literature, we think there should be upfront clear labelling so that then the consumer can decide what they want to purchase". Both the SoA and The Authors Guild in the US have launched kitemarking schemes that can be used on human work to distinguish it from AI-generated books. Authors can register their books to use a "Human Authored" logo on their back cover, signifying only de minimis tinkering with text by AI such as spelling and grammar checks. But Rosenbaum's ill-fated book carried just such a disclaimer. He says that he is now working with a fact-checker to review the manuscript word by word before publishing a new version. He notes that his experience has completely convinced him of the need for new "editorial safeguards for a world where AI is part of the publishing workflow". But he adds that he is "not walking away from AI".

How A.I. Books Sneak Their Way Into Stores

Big booksellers are taking different approaches to dealing with the deluge of titles generated by artificial intelligence. Jess Michaels, a historical romance writer, has released 117 novels since she started writing full time in 1999, most of them self-published. She's built a devoted readership, and has had 10 USA Today best sellers. But lately, sales have been so dismal that she might stop publishing. "The glut of the A.I. work makes discoverability very difficult," Michaels said. "You're talking about directly pirating my work, putting it into an engine so that you can produce a 'Jess Michaels-type book,' and then glutting the market with it. "I cannot tell you how soul-crushing it is," she added. "It essentially erases actual authors." It didn't take long for chatbot writing to spill into books after artificial intelligence models capable of spitting out sophisticated prose arrived several years ago. But many in the book world held out hope that fiction would be the last fortress to fall -- that readers wanted stories full of human emotions and could tell when they weren't written by humans. A few years into the A.I. era, there is sobering evidence that a significant and growing portion of the novels being sold are written with or by chatbots, and that readers are buying them. A new study conducted by computer scientists using A.I. detection software to analyze thousands of self-published novels on Amazon, found that a substantial number of those novels contained chatbot-generated prose. "In genre fiction, people are using A.I. to write these books, and readers are not able to tell," said Tuhin Chakrabarty, a professor of computer science at the State University of New York at Stony Brook, who led the research. After setting aside books that weren't selling, the team looked at a random sample of 14,419 self-published novels released on Amazon between January 2023 and March 2026 using A.I. detection software from Pangram, which many computer scientists consider to be one of the most reliable programs of its kind. The study found that in 2,168 of those books, more than 50 percent of the prose was A.I. generated, and that 970 of the novels were more than 90 percent A.I.-generated. The researchers estimate that the sales of all novels in their sample including at least 25 percent A.I.-generated text totaled in the tens of millions of dollars in the last three years. The findings, which have not yet been peer-reviewed, seem to confirm what many in the book business feared: A.I. writing is now an inescapable fixture of the retail landscape. Amazon doesn't prohibit the sale of A.I.-generated books on its site. The company asks self-published authors to disclose to Amazon whether or not they used A.I. to create their books, and limits the number of books each author can upload to the site to 10 per week. But writers are not required to flag their A.I. use for readers. And the 10-book limit still allows for the sale of A.I.-assisted works at a scale and speed that would be practically impossible for even a prolific human to match. A spokesman for Amazon, Josh Pflug, said the company has clear content guidelines regarding the sale of books on its site. "We take proactive and reactive measures to prevent, detect, and remove content that violates those guidelines, whether A.I.-generated or not," he said. Other researchers have also found an alarming number of books containing A.I. writing on Amazon. A study from the National Bureau of Economic Research that came out earlier this year found a correlation between the spike in the number of books that contain A.I.-generated prose and the rapidly growing number of e-books on the site -- a finding that Amazon disputes. "The claims about title growth in our store are not consistent with what we are seeing," Pflug said, although he declined to provide internal figures. The researchers used Pangram to analyze text from around 60,000 books sold on Amazon. From that sample, about 53 percent of the books published in 2025 contained A.I.-generated text, compared to 23 percent in 2023. The study found A.I. use to be higher in nonfiction categories like travel and outdoor sports, and comparably low in fiction. It also showed that the average rating of new books on Amazon has dropped as A.I. use has spread. "If all of these new books are much worse than what we already have, then there's little to be gained by consumers," said Imke Reimers, an associate professor of strategy and business economics at Cornell University, who coauthored the study. Both studies could add fuel to copyright cases that authors and publishers have brought against A.I. companies. In a barrage of lawsuits against OpenAI, Anthropic, Google and xAI and Meta, publishers and writers have argued that the surge in A.I.-generated works has flooded the market, diluting some of their sales. Over the past several months, a string of A.I.-related scandals has unsettled the literary world. And the explosion of A.I. writing has put distributors and retailers in an awkward spot. Amazon's stance toward selling A.I.-generated titles has generally been lenient, but other retailers have tried to take a more nuanced approach toward stocking them. In May, James Daunt, the chief executive of Barnes & Noble, caused an uproar in the book world when he said in an interview on the "Today" show that he has "no problem" selling A.I.-generated books if they are labeled. His remarks were widely condemned on social media by readers and writers, who said he was opening the door to selling slop at a major bookstore chain. Daunt later clarified his remarks in a statement he shared with The New York Times, noting that Barnes & Noble takes "active measures to exclude all A.I.-generated books from our online catalog and never knowingly order any for stocking in our stores." He added that while "the store will sell A.I.-generated books if there is clear demand, we think it very unlikely that there will be customer demand for A.I.-generated books." There are at least some A.I. novels for sale on Barnes & Noble's website, including "Fighting Love's Flame," a steamy romance about an Alaskan smokejumper and a documentary filmmaker who falls for him, and "Tumbleweeds & Tequila," about a woman who is stranded in the desert and rescued by a hot ex-Navy Seal. Both books were published by Future Fiction Press, a new publishing company that produces novels using A.I. The books' overviews on the Barnes & Noble website disclose that they were "produced using author‑directed A.I. tools." Other booksellers are drawing a harder line against A.I. generated works. "Are you going to allow scammers to rip off your customers? The answer has to be no," said Andy Hunter, the founder of Bookshop, an online book-selling platform that bills itself as an independent business alternative to Amazon. Hunter said Bookshop asks Ingram and Draft2Digital, companies that distribute books to the site, to filter out A.I.-generated works. But he worries there is no reliable way to fully block them. With more than 800,000 self-published books available for purchase on the site, it's inevitable that some slip through the cracks. Other online retailers say they have been flooded with A.I. slop and are struggling to stay ahead of the torrent. "We're on the receiving end of an avalanche, an avalanche that is only going to get bigger," said Michael Tamblyn, the chief executive of Rakuten Kobo, an e-reader producer and e-book retailer. Last year, Kobo's self-publishing platform rejected 45 percent of the books submitted. Of those, 85 percent were not accepted because Kobo determined them to be A.I.-generated. "You can say, This probably isn't a publisher, this is three LLMs in a trench coat," Tamblyn said. Other times, it's a harder call. Kobo is now evaluating whether A.I. detection software, along with human review, can assess titles. But it's becoming more difficult to catch scammers, who are using chatbots that produce more sophisticated prose, and can also use A.I. agents to create multiple accounts on publishing platforms. Kris Austin, the co-founder and chief executive of Draft2Digital, which distributes independently published books to retailers, libraries and subscription services, said that the company saw an explosion of high volume uploads and low quality books over the last 18 months -- a surge partly driven by A.I. He worries about how that will affect self-published authors, many of whom are struggling to find an audience in a flooded marketplace. "This rise of low-quality slop, it's threatening what the indie community has built," he said.

Is This What Comes After AI Slop?

Daggermouth could be the first bestselling novel partly written by a chatbot. 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 traces the unlikely romance between the president's son and an assassin hired to kill him. Since H. M. Wolfe uploaded Daggermouth to Amazon as a Kindle ebook, it has been a viral hit. "Did I finish Daggermouth, or did Daggermouth finish me?" a TikTok book influencer gushed. In February, Simon & Schuster paid seven figures to secure the publishing rights to the book and its sequel, which is due out next year. Daggermouth has spent months on USA Today's best-seller list and is ranked No. 1 in Amazon's "science-fiction romance" genre. It also tops a less prestigious list, one compiled by academics studying AI writing. Among thousands of self-published Amazon ebooks that they identified as containing "substantial AI text," none was more popular than Daggermouth. Until recently, AI fiction has been easy to dismiss as slop. On Amazon, new-book releases nearly tripled from 2022 to 2025. (That's probably not because humans suddenly dedicated themselves en masse to the written word.) Many are blatant copycats of existing titles, seemingly intended to trick readers into buying them. Their sales figures are generally meager; the reviews, indignant. A suspicious number of them feature protagonists with the same names. Last year, readers of one fantasy novel noticed a line that seemed to be responding to a prompt asking a chatbot to write in the style of another author. But if a book as beloved as Daggermouth were written with a chatbot, it would raise an awkward question: Could AI fiction be getting ... good? In a new study, researchers randomly selected more than 14,000 Kindle ebooks and ran them through Pangram, an AI-detection tool. One in five came back as more than 25 percent AI-written; Daggermouth returned 60 percent, according to an underlying data set that the authors shared with me. Those numbers, they acknowledge, tell us only so much. The researchers grouped together text that Pangram had flagged as either "AI-generated" or "moderately AI-assisted" (the latter of which can mean that AI was used to edit or polish human writing, or vice versa). And although Pangram has a low rate of false positives, it is not perfect. Read: America has a Pangram problem So the study's authors, led by Tuhin Chakrabarty, a professor of computer science at Stony Brook University, devised another check. They scanned each book's text 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 human-written ones. Daggermouth had many of them. For instance, the following sentence appears near the end of a multipage sex scene in Chapter 27: "They collapsed together, a tangle of sweat-slicked limbs and racing hearts." That same sentence, word for word, also shows up in another self-published ebook by a different author. Or consider this line from Chapter 38 of Daggermouth: "Her pulse thundered in her ears, drowning out the murmurs of the crowd." It's also in an ebook by another author who has published more than 400 different titles on Amazon in 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 wholly untrue," Wolfe said in a statement sent by her lawyer. "I wrote this book myself. I have been outspoken about my opposition to generative AI and what it's doing to writers, artists, and the creative community. I don't believe it belongs in the writing process." Referencing Pangram's limitations, Wolfe added that "accusations like these have real consequences for authors, and they should not be made on the basis of technology that is known to be unreliable." Susannah Lawrence, a spokesperson for Daggermouth's publisher, Simon & Schuster, told me in an email that the company stands 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 the "rare expressions" that Daggermouth shares with other recent ebooks. Chakrabarty, who researches AI writing, told me that Daggermouth has "too many telltale signs" of AI for him to buy that explanation. He cited Chapter 18, "Thank You," as an example: His research team broke it into 12 chunks for Pangram to evaluate, and the tool classified all parts as AI-generated. (In my own spot checks, the tool classified the first few paragraphs as human-written and the rest of the chapter as 100 percent AI.) I ran this by Mohit Iyyer, a professor of computer science at the University of Maryland who wasn't involved with the study. An expert on AI-language detection, Iyyer told me that even accounting for Pangram's fallibility, it's "almost statistically impossible" for an entirely human-written book to register as 60 percent AI. He added that the study's broader methodology, which connects AI writing to ebook sales, is sound. Whether or not you believe that Wolfe deserves the benefit of the doubt, the same questions about how Daggermouth was written apply to a growing number of other hits on Amazon. Chakrabarty's study found that books written with "substantial" AI assistance made up 20 percent of Amazon's ebook catalog, 12 percent of sales, and 10 percent of best sellers in their genre. Yet Daggermouth stands out for more than just its popularity. The second-best-selling ebook in the data set is Inked Athena, by Nicole Fox, an author who has 418 titles listed on Amazon. Daggermouth, in contrast, appears to be one of just 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 craft. Iyyer said that using AI in a way that results in a novel that's actually enjoyable to read takes "a huge amount of effort by a human author." Read: The future of writing is a lot like hip-hop Another difference: Jennifer Prokop, who reviews books for Kirkus and co-hosts Fated Mates, a popular podcast about romance novels, told me that people in her world keep talking about Daggermouth. "Nicole Fox, I've never heard a single soul talk about her," she said. (Fox did not respond to a request for comment.) Prokop sees mechanical prose cropping up mostly in self-published ebooks, for which the dominant platform is Amazon's Kindle Unlimited subscription service. Just because an ebook is charting on Kindle doesn't mean people are devouring it. Grifters can now use AI to crank out titles by the dozen and then use all kinds of tricks to entice readers, such as buying hundreds of bogus five-star reviews and paying click farms to juice their metrics. "If you can get 10,000 people to read 10 pages of your book, and you have 300 books you're putting out," she said, you can make real money. "But you're not actually writing romances people want to read." (Amazon did not respond to a request for comment.) If Daggermouth is AI, then maybe it's just an outlier -- a book that caught the perfect wave of social-media hype. (A series of controversies surrounding the book, including accusations of the author's AI use and a backlash to the cover art for a special hardcover version, seemed to galvanize its fans and bring it even more attention.) The notion that AI writing requires a perfect confluence of events to top the charts is a comforting thought. But it might not be true. "Everyone thinks they can tell AI writing," Christine Larson, a journalism professor at the University of Colorado at Boulder who studies the effect of technology on creative work, told me. "But I don't think we can tell so much." Privately, several authors and editors have told her that they're using the technology to accelerate or improve their output. "I totally buy into the idea that a human working with AI can write something that is compelling," she said. "I have absolutely no doubt that we will see more of this in the future." If she's right, then the question is not about whether we'll see AI best sellers. It's about whether anyone involved with them will admit it. Coral Hart, a South African author who uses AI to help her publish hundreds of books a year under different pen names, told me that she works hard to keep her pseudonyms secret after she received death threats and thousands of one-star reviews following a New York Times story about her. Hart claimed that she has had successful authors join the classes that she teaches on AI writing, though she declined to name them. "Until they stop the bullying, the witch hunts, and the ousting, I don't think authors are going to be transparent, and I don't blame them," Hart said. "I'm certainly not going to." The taboo against the use of AI in literature is understandable: Among other things, it's wrong to take credit for a book that you didn't write yourself. Shunning authors suspected of employing AI might work if there were any way to tell for sure -- which there isn't. That shaming ritual puts the moral burden on individual authors instead of on the mammoth corporations responsible for the system in which they're trying to eke out a livelihood. It also tends to collapse any distinction between pure AI slop and human writing that is assisted by AI. The upshot is a state of affairs that feels untenable: AI writing can be lucrative, but only if it's kept secret. Eventually, something's got to give.

I searched Amazon for AI-written books and it was hard to spot them

There are more AI-authored books on the market than you'd expect When ChatGPT first launched, most of the conversation centered on AI-generated student essays, deepfake images and the threat of automated fake news. Books, however, felt different. As someone who reads and reviews half a dozen books a month for ReaderViews, I've always found a certain comfort in the traditional publishing process. Bringing a book to market has historically required editors, designers, proofreaders, and, most importantly, an author who actually sat down and wrote it. That barrier to entry gave books an inherent authority that standard websites and fleeting social media posts often lacked. Then I read a recent New York Times investigation into AI-generated biographies appearing on Amazon. These included books about living public figures that were riddled with factual mistakes and seemingly produced with zero human oversight. I wondered: Could I actually spot one in the wild? I opened Amazon and started searching. Most of the listings looked completely normal The most surprising part of the whole experiment was at first glance, nothing looked unusual. All the books with AI-generated content featured professional-looking covers, convincing descriptions, reasonable price points, and even star ratings. Nothing immediately screamed "AI." If I hadn't intentionally gone looking for them, I probably would have assumed they were entirely legitimate. Once I dug a little deeper, however, the facade started to crack -- and the first thing that tipped me off was the author pages. I clicked on the name behind one biography and landed on the page of someone called John Crane Miller, described as a "seasoned biographer and cultural analyst" with a stock-photo portrait. Miller had published ten biographies in a single week, all about journalists and public figures, and when Kashmir Hill left a public review on Amazon asking to speak with him, his books quietly vanished. He wasn't the only phantom. A study by Originality.ai found entire catalogs built by authors who appeared to be fabricated identities, one of whom had published 74 books in five months on topics ranging from porn addiction to solo parenting to toxic workplaces. No single person writes like that. These weren't prolific indie authors with a strong work ethic; they were fake names attached to automated output. Marketing copy was a dead giveaway Blurb after blurb leaned on the same broad promises and eerily familiar phrasing -- "delve into," "comprehensive guide," "unlock the secrets of" -- without ever explaining what made the book worth reading over any other book on the same topic. After half a dozen of these, I could practically predict the next sentence before I read it. The marketing copy wasn't just generic; it was interchangeable, as if the same prompt had been run through the same model with only the subject swapped out. The biographies were the most troubling category, and the wave of books about tech journalist Kara Swisher illustrated why. Multiple unauthorized biographies of Swisher appeared on Amazon around the release of her memoir "Burn Book," written by authors with no apparent journalism credentials and no indication they had ever spoken to Swisher or anyone in her orbit. Political scientist Ian Bremmer encountered the same phenomenon: books bearing his name as a subject, produced without his involvement, that simply scraped and summarized what was already freely available online. This is, of course, exactly what chatbots are built to do, that is, synthesize existing information into fluent, confident-sounding prose. The result is a product that looks and feels like a real book but contributes nothing a reader couldn't get from the source material itself. Beyond books Substack recently launched an AI detection feature in partnership with a company called Pangram, letting readers scan any post to see an estimate of how much was written by a human versus generated by AI. Substack CEO Chris Best framed it as a way to keep the platform from becoming "like LinkedIn," overrun with hollow AI content, but even Substack acknowledges the tool isn't perfect. The Atlantic found that Pangram's accuracy is far from bulletproof, and critics worry about false positives flagging legitimate human writers. Still, the fact that a major publishing platform felt compelled to build this feature at all says something about how urgent the problem has become. Amazon, for its part, has no equivalent. The takeaway As someone who absolutely loves to read, I found this to be a very interesting experiment. My takeaway is that now, instead of trusting a listing at face value, I take a few extra minutes to look up the author outside of Amazon to see if they have a credible publishing history elsewhere, and I specifically hunt down professional reviews rather than relying on standard star ratings. I always preview the sample text to check if the writing feels robotic or relies on repetitive phrasing, and for nonfiction, I actively look for evidence of real human interviews and original citations. None of these steps individually guarantee a book was written by a human, but together, they make it much easier to separate thoughtful, researched work from automated content designed just to game search results. Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds. Subscribe to Tom's Guide on YouTube and follow us on TikTok. Finally, you can visit our dedicated Tom's Guide Savings Squad hub for expert help on getting the best products for less.