Book publishers have spent 2026 looking like the industry fighting hardest against generative AI: lawsuits against AI companies, books pulled over suspected LLM text. Wired reported on October 9 that, behind the scenes, at least three of the Big Five US publishers (HarperCollins, Simon & Schuster and Hachette) have been working AI tools into everyday publishing work "without public disclosures or author consent." The reporting rests on interviews with more than two dozen workers, all anonymous because they fear for their jobs.
This is a report built on anonymous sources plus limited company statements, so this post separates what Wired attributes to staff from what the companies actually said. If you write, edit, or sell books, or you simply follow how AI is quietly entering creative industries, here is the picture.
Quick answers
| Question | Answer |
|---|---|
| Who is named? | HarperCollins, Simon & Schuster, Hachette |
| What AI is used? | Claude, ChatGPT, Jasper (HarperCollins licenses, per staff) |
| For what? | Agent emails, publicity and back-cover copy, cover art assets, marketing videos |
| Did publishers confirm? | Partly: Hachette and Simon & Schuster gave policy statements; HarperCollins did not respond |
| Were authors told? | Wired says no public disclosure or consent |
| Biggest open worry? | Whether author manuscripts are pasted into outside LLMs |
What Wired actually reported
According to Wired's interviews, publishers are using LLMs like Claude and ChatGPT to "email literary agents, write publicity and back-cover copy, design cover art, and generate marketing materials, often at the direction of their executives." Staff also say the highly publicized cases of authors accused of using LLMs, such as Hachette's horror novel Shy Girl and Macmillan's crime thriller Call Me, I'll Hide the Body, are the tip of the iceberg. One editor told Wired that "every single imprint has a story of someone very esteemed who had a book canceled because of blatant AI use" that never made the news.
The most common use, according to workers at all three houses, is publicity copy. That is the text readers meet on Amazon and Goodreads and on back covers, and the pitch material sent to booksellers, journalists and critics. A Big Five publicist called AI-written publicity copy "short-sighted", arguing that publicists owe good-faith engagement to readers and media. Meg Reid, publisher of the independent Hub City Press, told Wired it "feels pretty inevitable" at the largest houses, with smaller presses insisting that the humans involved are the point.
Treat these as staff accounts. Wired says the sources spoke anonymously, and the only on-the-record company positions are the two statements covered below.
HarperCollins: "AI Champions" and monthly brainstorms
Wired's account of HarperCollins is the most detailed. A current employee said senior leaders bought Claude licenses from Anthropic months ago "but very quickly realized they didn't know how to implement them." Dozens of staff were then "voluntold" to become "AI Champions" and asked to find use cases in monthly brainstorming sessions. Staff say the company also bought ChatGPT and Jasper licenses, and that at least one division is "all in", including AI-generated marketing videos.
When employees raised legal, ethical and environmental concerns at a spring session, a meeting leader reportedly called them "the cost of doing business." A statement later added to the AI section of the employee portal said that while AI has an environmental impact, it is "a small part of most individuals' total digital carbon footprint." HarperCollins did not respond to Wired's requests for comment.
Context matters here. HarperCollins was the publisher that drew attention for licensing backlist titles to an AI company in 2024, with authors given an opt-in choice. The new reporting concerns something different: internal workflow use, where authors never see an opt-in screen.
Simon & Schuster: contests, a workshop, and Skan AI
At Simon & Schuster, staff described a workshop with OpenAI representatives. One attendee said the presenter "genuinely had no idea what we did," yet one suggested use for ChatGPT in the book trade was composing emails to literary agents. Wired also reports an employee contest in May 2024 with a $10,000 first prize and $5,000 for each of two runners-up. Since then, an employee says, the company "rolled back some of the AI-pushing because employees kept complaining about it in town halls."
The sharpest conflict is over Skan AI. Staff heard the company was running an experimental trial "at the apparent urging of KKR," the investment firm that bought Simon & Schuster in 2023. Skan describes its product as giving leaders evidence "to reduce costs, prioritize automation." Employees began collecting signatures for an open letter to CEO Greg Greeley opposing the software, worried about monitoring and about "automation discovery" tools being used to shadow and replace them.
New York magazine's Book Gossip newsletter reported the letter on October 6. The same day, per Wired, Greeley told staff the company had been exploring Skan AI's Blueprint but that "no decision has been made to use Blueprint or any other productivity-monitoring or workflow-analysis tool." Simon & Schuster's chief marketing and communications officer, Wibke Grutjen, told Wired employees have a limited number of enterprise-level AI tools, that use is not mandatory, and that unvetted tools are prohibited.
Wired also notes that the AI detector Pangram flagged at least part of Greeley's memo as AI-written. Detectors are imperfect, as we covered in Pangram's Gmail labeler false-positive problem, so that is a data point, not proof.
Hachette: operational yes, creative no
A Hachette spokesperson told Wired the company "support[s] the use of AI tools for operational purposes, when they help our authors' books reach more readers," but does "not support creative uses of AI, including communicating with our authors and partners." Staff say practice has drifted past that line: employees at all three houses told Wired editors increasingly use LLMs for agent emails, and that agents believe some rejection letters were machine-written because they shared telltale patterns like "not this, but that" and the rule of three.
Hachette staff also expect policy to "evolve" under new chief information officer Carl Hixson, who previously served as chief technology officer at SRMG, a Saudi state-backed company that has embraced agentic AI, Wired reports.
Why workers say they are doing it
The article's thread on labor is as important as the one on authors. Staff describe layoffs and swollen workloads: one employee said large teams were reduced to five or six people "doing double or triple the number of books." Others told Wired they experiment with LLMs reluctantly, because executives gutted teams and then said "do this or we fall behind." The average entry-level salary for New York City publishing employees at the Big Five and Scholastic was $47,583 in 2023, per the figure Wired cites.
This mirrors patterns we have covered elsewhere: UK employers cutting entry-level roles, Indian tech workers' layoff fears, and Flock Safety's staff cuts amid surveillance backlash. In each, management pushes AI adoption while headcount shrinks, and workers are the ones asked to square that with their own professional ethics.
Cover art and the disclosure gap
Staff told Wired AI assets are appearing in cover art, and some foreign presses distributed by Simon & Schuster openly acknowledge AI use when sharing designs. Editors also report LLM-translated manuscripts with AI-use disclosures at the Frankfurt and London book fairs, where US publishers buy translation rights.
Cover art is where law and ethics collide. Whether purely machine-made art can be protected is unsettled, and our explainer on why EU rules deny protection to AI-generated content shows the same logic abroad: if a cover is mostly machine-made, the publisher may not be able to enforce rights in it. Image licensing has its own history, which the Getty and OpenAI deal illustrates. Creative industries are also dealing with AI at the label and platform level, as in Spotify's AI artist labels and the Sony and Warner suit against Anthropic.
The manuscript question
The fear agents voiced to Wired is specific: that publishers might paste text from authors' manuscripts into LLMs to write rejection letters or publicity descriptions. One agent said they have added contract clauses stating the material "can't be run through an LLM," a protection originally aimed at tech companies that "also applies to the company buying and publishing the book."
Wired explains the distinction between open-loop tools (cloud chatbots) and closed-loop or air-gapped models. A HarperCollins source said multiple editors asked managers if they could paste a full manuscript into an open-loop LLM and legal said no. That suggests at least one guardrail exists, but no source claims manuscripts were uploaded. The honest status: a plausible risk, a contractual response, and no confirmed incident.
What authors can do
None of this is legal advice, but the reporting suggests concrete steps:
- Ask your agent to add AI clauses that cover the publisher's own use of your text, cover art and likeness, not only third-party training.
- Ask your editor in writing which AI tools touch your book, from copy to translation to cover.
- Ask whether jacket copy and blurbs are human-written, and whether you get approval rights.
- Keep drafts of your own descriptions so you can compare them with what ships.
- Record disclosures at the foreign-rights level if you sell translations.
Emma Dries, an agent at Triangle House Literary, told Wired most people just want clear usage guidelines and disclosures, adding that "everybody should be communicating more." A current HarperCollins staffer said there has been no guidance on what to tell authors or agents about any of this.
Why it matters beyond books
Publishing is a bellwether. Its customers (readers) and suppliers (authors) both care deeply about human authorship, and the industry has publicly taken a hard line. If AI quietly slips into workflows there, the same gap between public posture and internal practice is likely elsewhere. The problem is rarely whether a tool is used; it is whether the people affected were told.
For AI builders the lesson is practical: enterprise rollouts that push "AI Champions" without clear data policies, disclosure rules or opt-outs create exactly this backlash. For readers, the lesson is that a human-sounding back cover is no longer proof of a human behind it. For background on how detection and provenance are meant to help, see our guide to AI writing tells in Claude Opus 5 and the explainer on what C2PA content credentials are.
What to watch next
- Whether HarperCollins responds publicly, and whether any house publishes an AI-use policy for authors.
- Whether Simon & Schuster rules out Blueprint or similar monitoring software.
- Whether agents' AI clauses become standard in Big Five contracts.
- Any author who finds AI-written jacket copy or cover art on a book and goes public.
Based on Wired's reporting dated October 9, 2026; details come largely from anonymous sources and may change as publishers respond.