One evening last fall, a middle-aged couple, heads bent in conversation, approached the table where I sat signing books. “Go ahead,” the woman urged. Speaking softly, her husband explained that he had Parkinson’s disease. Although he hadn’t read my novel The Weight of Ink, it was his understanding that it featured a character with Parkinson’s who chose to die by assisted suicide rather than endure her decline. He had some objections.
I was horrified—not because I’d written something that offended him, but because I hadn’t. Helen Watt, the character with Parkinson’s in The Weight of Ink, does not, in fact, end her own life. I didn’t know for certain where his misunderstanding had originated, but I had a guess. Just a few weeks earlier, I’d asked ChatGPT a few questions about my novel; it responded to certain queries with this false rendition of Helen’s plotline. Ever since, I’d worried that the misrepresentation might somehow have a real-life impact. Now, it seemed, it had.
Any working writer can expect to be called out for their choices; critique is part of the terrain. As a novelist, I’ve been chided for details including a character’s self-defeating relationship decisions and an incorrect date for a 17th-century full moon. But that evening in late 2025 was the first time I’d been called to account for something I hadn’t actually written. And even as I explained to the couple before me that what they’d seen online was incorrect, I wondered: How many other people had been similarly misled?
The proliferation of artificial intelligence has raised new fears about the long-term viability of writers’ already tenuous profession, as the technology infiltrates prize-winning writing, scrapes books for learning material, and perhaps erodes readers’ attention spans. But alongside that upheaval, another subtler sort of destruction is under way. AI-enabled meddling is turning the reader-writer connection into a hall of fun-house mirrors, and threatening to break the trust that is essential to that bond.
In the Monty Python sketch “Bookshop,” a customer requests a progressively more absurd list of copycat Charles Dickens works. Does the shop have a copy of Knickerless Knickleby? Or perhaps titles by one Charles Dikkens (“that’s Dikkens with two K’s”)? The customer’s insistence on the legitimacy of these books gradually drives John Cleese’s bookshop-clerk character into a frenzy. What was delightfully preposterous decades ago is now uncomfortably familiar. With a blizzard of literary fakery overtaking the world of books, discerning what’s real requires vigilance on the part of readers.
Read: The worst thing about AI writing
In 2017, when I published The Weight of Ink, there was no other book by that title in the English language. This remained the case until the fall of 2025. Since then, at least five other novels with titles identical to mine have become available on Amazon, three of them published in the space of three weeks. Several other books with nearly identical titles also are listed for sale. I don’t know whether, or to what extent, AI was used in creating these books whose content otherwise seems to bear little or no resemblance to my novel; I found only one that contains an up-front disclaimer acknowledging AI authorship. But as Will Oremus noted in a recent Atlantic article, new-book releases on Amazon skyrocketed after 2022, around the time that AI models became widely accessible to the public. “That’s probably not because humans suddenly dedicated themselves en masse to the written word,” Oremus wrote. Similarly, it’s hard to believe that this profusion of Weight of Inks is the result of an abrupt, hyperspecific surge of entirely human creativity.
In some ways, I felt lucky to have just my title copied; in late 2024, my friend the novelist Daphne Kalotay discovered that her author page on Amazon featured a brand-new novel that she hadn’t written. This book, which Kalotay described as “an AI hallucination based on descriptions of Russian Winter,” her 2010 novel, had been paired with a title and cover from a completely unrelated book from decades earlier and was being sold under her name. “What was most unsettling,” Kalotay told me, “was how difficult it was to prove it was garbage.” In the end, the Authors Guild, an advocacy organization of which Kalotay is a member—and which has launched a certification program designed to distinguish human-authored books from those generated by AI—succeeded in getting Amazon to delete the book from her author page.
This sort of scam is now everywhere in publishing. Generating book-length texts using AI costs almost nothing, and Amazon lets individuals upload a maximum of six books a week, allowing for endless churn. Anyone selling AI-generated texts need only mislead a few customers per book—whether by using an existing work’s title or an author’s name—in order to turn a profit. “Amazon spends a lot of time and effort trying to prevent these books from getting posted,” Mary Rasenberger, the CEO of the Authors Guild, told me, “but it’s always a game of catch-up.”
The fakery cuts both ways: Writers are also being bombarded by AI-generated phishing attempts in the guise of emails from appreciative readers. Such emails are so pervasive that, like many of my peers, I now sometimes hesitate to even open my reader mail. Each of these notes begins personally, only to morph into an offer of paid marketing services. A recent example begins: “Dear Rachel, This feels scary to send, but I’m trusting my heart today.” The subsequent description of my work’s ostensible impact (“Your words found me at a time when I was quietly falling apart”) is followed by a pitch for a marketing plan to promote my novella, I Was Here. These promotion schemes are, of course, more fakery; the Authors Guild maintains a running list of such scams, in which writers who sign up see their money vanish.
Read: Deepfakes are coming for your bank account
Although fraudulent emails in writers’ inboxes usually purport to be from readers, some are sent under the names of published writers. (Some are also brazenly posthumous; a friend of mine received compliments on her memoir from “Charlotte Bronte.”) Earlier this month, a writer I don’t know forwarded me an email he’d received from someone claiming to be me. In that email, “Rachel Kadish” reached out to him “author to author” to express interest in his work. Below my alleged signature on the email were links to my actual website, as well as to my Amazon and Goodreads pages.
I reached out to “Rachel Kadish” to ask them to explain their purpose in using my identity. The reply I received—an ingratiating and incoherent piece of clearly AI-generated text—acknowledged the impersonation before going on to explain that the ruse had all been for my benefit. (“My intention was to connect with fellow authors and introduce them to your work, as I believe authors discovering and supporting one another can be valuable.”) The best part was the sign-off: “Going forward, I will communicate under my own name and will not use your identity to contact other authors. Sincerely, Rachel.”
This is a new, bleak kind of fraud, different from the literary deception of earlier eras, in which scandals at least had something to teach us. Consider The Education of Little Tree: Published in 1976 as the true autobiographical story of a Cherokee boy, the book was revealed in 1991 to be the work of a former member of the KKK. Or Binjamin Wilkomirski’s Fragments: Memories of a Wartime Childhood, whose exposure as a fake gave rise to the term Wilkomirski syndrome for the phenomenon of non-Jews posing as Holocaust survivors. These scandals revealed not only the writers’ lies but also the degree to which readers’ desire for certain stories can blind us. There was human folly and frailty to learn from. A few recent scandals, involving prize-winning writing that was suspected of being AI-generated, likewise force us to stare into the abyss of our own clichés and expectations. But we can glean no meaning from a marketing bot spitting out derivative novels or slathering one writer with praise under another’s name.
When readers and writers can no longer reliably connect, the literary world loses some of its richness and potential. Charles Dickens (that’s D-I-C-K-E-N-S) famously reexamined his use of anti-Semitic tropes in response to a reader’s rebuke, resulting in a notable change in his depiction of Jews; though the gentle Mr. Riah of Our Mutual Friend is arguably as two-dimensional as the villainous Fagin of Oliver Twist, his depiction undeniably represented progress. Arthur Conan Doyle’s choice to kill off Sherlock Holmes in the 1893 story “The Final Problem” sparked such grief and outrage from his readers that, even years later, publishers were offering Conan Doyle serious money to revive the great detective—bait that Conan Doyle eventually decided to take in 1903, resulting in three more collections of Holmes stories and one more novel.
Read: How Sherlock Holmes broke copyright law
Most reader-writer interactions, of course, don’t make literary history. Yet even passing exchanges can have a powerful effect. Every writer can recall a moment when a reader’s pointed question made them rethink a choice—or else pissed them off enough that they gleefully doubled down at the next literary opportunity. Every reader, for their part, can name a book that outraged them, or challenged or even rewrote their understanding of the world.
“Narrative is radical,” Toni Morrison said in her 1993 Nobel Prize lecture, “creating us at the very moment it is being created.” This is what’s most fearsome about AI’s intrusion into storytelling: It separates us from those opportunities for growth and empathy in a world where these are desperately needed.
That man and woman who approached me that evening last year to confront me about my character’s alleged suicide were speaking from the depth of their personal experience. They deserved the chance to find out whether I’d written my characters with a thoughtfulness worthy of their trust. And I deserved the chance to have a human conversation, even a painful one, about a story I’d actually written—not a figment of the imagination of a machine.