How are major publishers implementing AI tools?
Major publishing houses are increasingly utilizing Large Language Models (LLMs) like ChatGPT and Claude for operational and marketing tasks, often without public disclosure. While the industry has made headlines for suing tech giants over copyright, internal practices suggest a different reality for the 'Big Five' US publishers.
According to interviews with over two dozen anonymous workers, publishers such as HarperCollins, Simon & Schuster, and Hachette are employing AI to handle a variety of administrative and creative-adjacent tasks. These include drafting emails to literary agents, generating back-cover copy, designing marketing materials, and creating publicity pitches for booksellers and journalists.
Specific software and use cases
At HarperCollins, employees report the acquisition of licenses for Claude, ChatGPT, and Jasper, the latter of which is an agentic platform designed to manage end-to-end marketing workflows. Some divisions have reportedly gone 'all in,' even utilizing AI-generated marketing videos. Beyond marketing, AI is also appearing in cover art, with some foreign presses distributed by Simon & Schuster openly acknowledging its use in design processes.
Why are publishing staff revolting against AI integration?
Staff resistance stems from a combination of ethical concerns, perceived devaluing of human expertise, and the pressure to act as 'AI Champions' for tools they did not request. Many workers feel that the investment in software comes at the expense of fair wages and adequate staffing levels.
At HarperCollins, junior and senior staff were reportedly 'voluntold' to become AI Champions, tasked with finding new ways to implement the technology during monthly brainstorming sessions. When employees raised concerns regarding the environmental and legal impacts of these tools, some leaders dismissed these objections as merely 'the cost of doing business.' This sentiment has caused significant friction, particularly as entry-level salaries in the New York publishing sector remained relatively low, with an average of $47,583 in 2023.
At Simon & Schuster, the tension escalated when workers learned the company was exploring Skan AI. This software, which provides evidence for automation and cost reduction, prompted employees to sign an open letter to CEO Greg Greeley. The workers expressed strong opposition to productivity-monitoring tools, fearing they would be used to shadow and eventually replace human roles.
What are the risks to author privacy and intellectual property?
The use of open-loop LLMs poses a significant threat to the confidentiality of unpublished manuscripts. Literary agents have expressed deep concerns that editors might paste sensitive author text into AI models to generate rejection letters or publicity descriptions, potentially feeding proprietary data into public training sets.
To mitigate this, some agents have begun adding specific clauses to contracts prohibiting the processing of manuscript material through LLMs. The distinction between 'open-loop' and 'closed-loop' models is critical here; while a closed-loop model operates on an air-gapped network to protect data, many standard AI tools share information with cloud servers, creating a massive privacy loophole for unreleased creative work.
The tension between corporate and creative interests
Industry analysts suggest a growing divide between corporate leadership and creative professionals. Dan Sinykin, author of *Big Fiction*, notes that C-suite executives at major publishing houses often prioritize profit maximization and efficiency in ways that clash with the values of editors, publicists, and translators. This tension is exacerbated by the fact that many publishing teams have been reduced to 'skeleton crews' due to layoffs, forcing remaining staff to use AI shortcuts just to manage tripled workloads.
How is AI affecting the relationship between publishers and agents?
The adoption of AI in routine communications has damaged the trust between literary agents and publishing houses. Agents have reported noticing 'classic signs' of AI writing in rejection letters, such as repetitive structures and specific linguistic patterns, which suggests that editors are using LLMs to automate sensitive correspondence.
Hachette has addressed these concerns by stating that while they support AI for 'operational purposes' to help books reach more readers, they do not support its use for creative tasks or for communicating with authors and partners. However, the lack of clear, industry-wide guidelines remains a primary source of chaos. Agents like Emma Dries have emphasized the urgent need for publishers to provide transparent disclosures regarding exactly when and how AI is being utilized in the publishing lifecycle.
What is the future of AI in book translation and production?
The industry is seeing an increase in AI-assisted translation, particularly at international book fairs where LLM-translated manuscripts are increasingly common. While some independent publishers, such as Europa Editions, have expressed excitement about the possibilities for AI in translation, the practice remains highly controversial within literary circles.
The shift toward AI-driven efficiency appears to be creating a bifurcated market. As noted by Meg Reid of Hub City Press, a divide is emerging between large corporate publishers seeking maximum efficiency through automation and smaller independent presses that insist human involvement is the fundamental value of the publishing process.
Frequently asked questions
Is AI being used to write books in the Big Five?
While there have been high-profile cases of books being canceled due to suspected AI authorship, such as Hachette’s *Shy Girl*, many editors suggest these are just the tip of the iceberg. There is ongoing tension regarding how much AI is used in the creative process versus administrative tasks.
Does using AI for publicity copy harm book sales?
Some publicists argue that using AI for pitches is short-sighted, as journalists and media professionals often have a low tolerance for AI-generated content. This can damage the 'good faith' engagement required to secure meaningful media coverage for authors.
Can AI be used to monitor employee productivity?
Yes, some companies are exploring tools like Skan AI, which can analyze workflows and identify opportunities for automation. This has led to significant pushback from employees who fear these tools are intended to justify further layoffs or surveillance.
How can authors protect their manuscripts from AI?
Authors and agents are increasingly using contract clauses that explicitly forbid running manuscript material through LLMs. This is designed to prevent unpublished text from being fed into open-loop models that could use the data for training.
Is AI translation becoming standard in publishing?
AI-assisted translation is becoming more prevalent, especially in the acquisition of international rights. While some industry leaders see this as an opportunity to expand markets, others view it as a threat to the quality and craft of human translation.
Key takeaways
- Major publishers are quietly using LLMs for marketing, publicity, and administrative tasks.
- Staff at top publishing houses are protesting AI use due to ethical and job security concerns.
- Agents fear that manuscript privacy is at risk if editors use open-loop AI models.
- A divide is forming between corporate efficiency-seekers and human-centric independent presses.
Conclusion
The integration of AI within the publishing industry is characterized by a profound disconnect between executive strategy and staff reality. While leadership views generative AI as a tool for operational efficiency and profit maximization, employees see it as a threat to creative integrity, privacy, and job stability. As the 'Big Five' continue to experiment with automation and workflow analysis, the industry faces a critical turning point. The future of publishing will likely be defined by whether companies can establish transparent, ethical guidelines that protect both the intellectual property of authors and the professional dignity of the workers who support them.
