In the 21st century publishing world, I can tell you from experience that even before LLMs, book authors who didn’t step in and do things themselves—or hire a freelancer with part of their advance—could often end up seeing surprisingly little effort go into things like design and publicity copy. But a new, anonymously-sourced report from Wired suggests AI is taking things from bad to worse. Wired talked to “more than two dozen” anonymous publishing industry employees, and found that three of the big five publishers are allegedly using AI and failing to notify the authors, let alone the public. In part, this is publicity-related. Publishers are reportedly generating PR material and cover copy with tools like ChatGPT and Claude. They’re also, according to Wired’s sources, using image generators to create cover art. This comes at a time when AI and books seem like forces arrayed against one another. The big five publishers jointly filed a lawsuit in May against Meta, alleging copyright infringement. Meanwhile, used booksellers are seeing their inventory snapped up by what they believe to be AI companies looking to literally destroy books—often rare ones. While that’s going on, literacy is in an all-out crisis. By one measure, students are the worst they’ve been at reading in 25 years. A study on people’s self-reported habits found that from 2003 to 2023, reading for pleasure declined about 3% per year. 44% of American men in particular didn’t read a book in 2025. But rather disturbingly, these measures of low reading and literacy coincide with good news for the publishers and their shareholders. The Association of American Publishers reports growth in revenue of 16.1% from 2021 through the present. Indeed, there are hints that the publishing industry’s AI-related woes are just a reflection of its broader, bottom-line-oriented approach across the board. For instance, in one-high-profile case earlier this year, the Macmillan imprint Minotaur Books scrapped a promising deal estimated at $2 million—following a 14-party bidding war— because there were questions as to whether the manuscript, called _Call Me, I’ll Hide the Body_ , was AI-generated. But AI didn’t necessarily trigger quality concerns. To quote the entertainment industry publication Deadline, “The problem: publishers and movie companies may not be able to register copyright and come away with a clear title chain if there are certain AI elements in the mix.” So the new Wired story paints a picture of a publishing industry where executives are just generally enthused about AI—allegedly exploring a wide array of uses, like expanding translation, generating promotional materials, and monitoring employee productivity. Executives also aroused suspicion that they’re generating their own internal documents with AI tools. An agent says they’ve started making publishers agree in advance not to run entire unpublished manuscripts through LLMs in order to generate promotional text—something that had apparently become an issue. “Ostensibly, that was meant to protect authors from tech companies, but it also applies to the company buying and publishing the book,” the anonymous agent said. Anonymous employees say their bosses are subscribing to AI tools, and holding brainstorming sessions, in which employees are tasked with finding ways to make them useful.