Amazon.com began requiring authors who want to sell books through its e-book program to tell the company in advance whether their work contains artificial intelligence-generated content.
Matt Slocum/AP
Hide caption
Toggle the caption.
Matt Slocum/AP
Amazon.com began requiring authors who want to sell books through its e-book program to tell the company in advance whether their work contains artificial intelligence-generated content.
Matt Slocum/AP
When super-influential tech journalist Kara Swisher came out with her new book, Book burning, there were reports of his seemingly AI-generated autobiographies suddenly appearing on Amazon. Swisher responded immediately New York Times’ The hard fork The podcast, “I Sent. [Amazon CEO] Andy Jesse took a note and said, ‘What’s up?’ You’re costing me money,” Swisher said.
But while Swisher was successful in removing offensive books from Amazon, the issue of AI-generated scam books has been a widespread concern for authors, many of whom are not on email terms with Amazon’s CEO. are
“Scammed books have been a problem on Amazon for years,” says Mary Rasenberger, CEO of the Authors Guild, an advocacy group for authors. But she says the problem has multiplied in recent months. “Every new book seems to have a companion book, a book trying to steal sales.”
Mary Arana is an author who spent years researching and writing her book. Latinos: A Portrait of America’s Largest and Most Understudied Minority. The book was released in February. The day after its release, she went to Amazon to see how it was doing. “There was another cover under the cover of my book,” Arana says. “The cover read ‘America’s Largest and Most Underestimated Minority. Latinoland Summary.’
Arana sent NPR a photo of the Amazon search results. The book says it was written by Clara Bailey. A review of Bailey’s work shows that Bailey published a large number of these so-called summaries and offered them for sale on Amazon. NPR asked an Amazon spokesperson about Bailey but did not receive a response. And the company didn’t turn up anyone for an interview when asked about AI-generated books, in general. Since NPR’s inquiry, Bailey’s books have been removed from Amazon. Bailey’s publication history still appears on Goodreads.
AI-generated biographies, summaries and even copycat books produce low-quality writing that makes it easy to flag them as AI creations, says Jane Friedman, author and publishing industry analyst. She says that there is a general quality of writing. “It feels like no human wrote it,” she says. “Humans would – ironically – do better than be evil.”
Amazon spokeswoman Lindsay Hamilton sent out a statement outlining recent steps the company has taken on the AI front. Last year, the company implemented a policy requiring all publishers using Kindle Direct Publishing to provide information about whether their content is AI-generated. There is also a limit to the number of titles published in a day.
“We both prevent books from being listed as well as remove books that don’t follow these guidelines, including content that creates a poor user experience,” the statement said. “. “When patterns of abuse warrant it, we also suspend publisher accounts to prevent repeated abuse.”
Publishers posting these books take advantage of increasingly sophisticated AI tools that can quickly produce low-quality “scam” books, Rasenberger says. “By the time Amazon finds out about them, they’ve already made some money and they move on to something else,” she says.
But the problem with AI-generated books can hurt more than the author’s sales numbers.
“It’s reputational damage,” Friedman says.
Last year, she wrote a blog post outlining her experience with books being published that she was supposed to write but didn’t write. “Even a beginner will read it and say, ‘This person is not going to give me any useful information,'” she says. Friedman says she makes most of her money through paid newsletters and classes she offers and that books can be viewed by potential customers. “And then they go to find some other better resource.”
And while it’s possible to flag AI-generated writing right now, Rasenberger and other authors are envisioning a future where it won’t be so easy.