US Publishers Embrace AI Amid Controversy
This year, the primary narrative in the publishing industry has been the pushback from American publishers against artificial intelligence (AI), including legal action against Google for copyright infringement and the cancellation of books by authors suspected of using large language models (LLMs). However, behind the scenes, at least three of the five major publishers in the United States—HarperCollins, Simon & Schuster, and Hachette—have discreetly integrated AI tools into their editorial processes without public acknowledgement or the consent of authors.
According to interviews with more than two dozen employees, who spoke on the condition of anonymity, these publishers have been leveraging LLMs like Claude and ChatGPT for various tasks, such as sending emails to literary agents, drafting promotional text and blurbs, designing covers, and generating marketing materials. At the same time, several editorial team members assert that high-profile cases of authors accused of using AI to write their books, such as Hachette’s horror novel “Shy Girl” or Macmillan’s thriller “Call Me, I’ll Hide the Body,” represent merely the tip of the iceberg. “Every publisher has a story about a very prestigious author whose book was cancelled due to the blatant use of AI, without it making headlines,” remarks one editor.
Internal Resistance and Ethical Concerns
At HarperCollins, executives acquired licenses for Claude from Anthropic months ago, but an employee notes that they soon realised they lacked the knowledge to implement it effectively. Subsequently, dozens of staff members, both junior and senior, were “pressured to volunteer to become AI champions” and tasked with identifying use cases during monthly brainstorming sessions with cross-departmental staff. During one of these sessions this spring, when some HarperCollins employees voiced concerns about the legal, ethical, and environmental implications of AI, the session leader dismissed their worries as “the cost of doing business.” Shortly thereafter, a statement was added to the new AI section of HarperCollins’ employee portal, which noted that while AI does have an environmental impact, it constitutes a small portion of the overall digital carbon footprint for most individuals.
Some HarperCollins employees expressed confusion over the brainstorming sessions. “It felt like a slap in the face. We have been making valid business arguments for years to secure a decent wage, yet they are spending that money on software that most of us really don’t need,” one employee commented. The average starting salary for editorial staff in New York City at the Big Five and Scholastic was $47,583 in 2023. In addition to Claude, employees claim that HarperCollins has also acquired licenses for ChatGPT and Jasper, an autonomous platform that promises to manage marketing workflows.
The Shift Towards AI in Marketing
According to an employee, at least one division within HarperCollins has focused on AI since obtaining software licenses, which includes generative videos. However, workers from HarperCollins, Simon & Schuster, and Hachette indicate that one of the most common uses of LLMs is for generating promotional copy. Readers encounter promotional text everywhere, from Amazon and Goodreads to back covers, aimed at pitching books to booksellers, journalists, content creators, and critics. Nevertheless, using AI for promotional text is seen as a short-sighted strategy. “We, as publicists, have a duty to act in good faith with readers and our media colleagues, who have little tolerance for AI-generated sales pitches,” suggests an employee from one of the Big Five.
Authors and Agents Express Concerns
On the other hand, Meg Reid, editor at independent publisher Hub City Press, sadly acknowledges that AI-generated advertising “seems inevitable.” “We will see more of this shift among corporate publishers seeking efficiency.” Meanwhile, earlier this year, literary agents began to suspect that certain editors at the Big Five were using AI to draft rejection letters. “Among colleagues, we compared emails about different books from the same editors, and they all exhibited the same classic signs of AI writing, such as ‘not this, but that’ and the three-term rule,” recalls one agent. Employees from Simon & Schuster, HarperCollins, and Hachette concur that the agents’ suspicions were likely accurate.
Publisher Responses to AI Integration
A spokesperson for Hachette stated that the company “supports the use of AI tools for operational purposes, when they help our authors’ books reach more readers,” but does not endorse creative uses of AI, including communication with authors and partners. Wibke Grutjen, global marketing and communications director at Simon & Schuster, noted that employees across all areas have access to a limited number of AI tools at the corporate level, but usage is not mandatory. Grutjen emphasised that the company prohibits the use of any AI tools that have not been assessed and approved, with the aim of freeing employees from basic tasks to enable them to engage in deeper work. HarperCollins did not respond to requests for comment.
Workshops and Internal Competitions on AI Use
Several months ago, employees from Simon & Schuster attended a workshop with representatives from OpenAI. “The person leading the workshop had no idea what we did,” recalls one attendee, but one of the AI company’s recommendations for using ChatGPT in the publishing sector was to draft emails to literary agents. According to employees, this practice has also spread among editors at HarperCollins and Hachette. In May 2024, Simon & Schuster reportedly organised a competition among its employees to identify the best uses of AI, offering a $10,000 prize for the winner and $5,000 for each of the two finalists. Since then, the company has “reevaluated its AI projects due to ongoing complaints from employees during all-hands meetings.”
Concerns Over Transparency and Copyright Issues
Numerous staff members have noted that AI resources are also appearing on book covers, with some foreign publishers distributed by Simon & Schuster openly acknowledging their use of AI in shared designs. All agents consulted expressed concern that publishers may be pasting text from their authors’ manuscripts into LLMs to generate rejection letters and promotional descriptions. “We have added clauses about AI to our contracts stipulating that ‘this material cannot be processed by a generative model,'” argues one agent, who requested anonymity to avoid professional repercussions. “This was apparently intended to protect authors from tech companies, but it also applies to the company that buys and publishes the book. If they use a model operating in an open environment, they could be inputting unpublished text from the author into an LLM.” Unlike the AI models that most people use online, such as Gemini or ChatGPT, a model operated in a closed environment functions on a computer or isolated network, thus preventing data sharing with external AI cloud servers.
Future Implications of AI in Publishing
A source from HarperCollins claims that several editors have asked their managers whether they could paste a complete manuscript into an LLM connected to external services, but the company’s legal team advised against it. Employees at Hachette are also preparing for the company’s AI policies to “evolve” under the new information director, Carl Hixson, due to his previous role as technology director at SRMG, a state-backed company that has publicly embraced AI with agents. Some employees from the Big Five report that they are reluctantly experimenting with LLMs only because executives have dismantled teams through layoffs, subsequently tripling their workloads. “There are skeleton crews everywhere you look. Large teams have been reduced to just five or six people, and we are handling double or triple the number of books. It’s no surprise that people are cutting corners. I don’t judge them. What I criticise is the company essentially telling us: ‘Do it, or we’ll fall behind,'” one employee commented.
Emerging Tensions Over AI Tools
In recent weeks, rumours have circulated among Simon & Schuster employees that the company was conducting an experimental trial with Skan AI, supposedly at the behest of KKR, the investment firm that acquired Simon & Schuster in 2023. According to its own website, Skan AI provides executives with the empirical foundation needed to reduce costs, prioritise automation, and fulfil board transformation mandates. In response, Simon & Schuster employees began gathering signatures for an open letter directed at their CEO, Greg Greeley, expressing their “firm opposition” and “requesting a commitment not to incorporate Skan AI or any similar software into S&S workflows now or in the future.” Many employees are concerned that Skan AI’s “automation detection” tools might be used to monitor and replace them.
The news of the open letter was reported on October 6 in the Book Gossip newsletter of New York Magazine. On the same day, Greeley sent a memorandum to Simon & Schuster employees confirming that the company had been exploring Skan AI’s Blueprint product but emphasising that “no decision has been made regarding the use of Blueprint or any other productivity monitoring or workflow analysis tools,” and that he “had not seen the materials related to this exploratory phase” before the concerns were raised the day prior. Pangram, which has become the go-to platform for AI detection, indicates that at least part of Greeley’s memo was drafted by AI.
AI’s Potential Impact on Translation and the Industry
Nearly two years ago, at the UBS Global Media and Communications Conference