Publishing staff describe AI’s quiet role at HarperCollins, S&S and Hachette

HarperCollins, Simon & Schuster and Hachette have incorporated generative AI into parts of their publishing operations, including publicity, marketing, correspondence with literary agents and, according to employees, cover-art work. The use has extended beyond back-office experimentation to material that can appear on book covers, Amazon and Goodreads pages, and pitches to booksellers, journalists, critics and online creators. A Wired investigation drew on interviews with more than two dozen industry workers. Employees at all three publishers described publicity copy as one of the most common uses of large language models, including ChatGPT and Anthropic’s Claude. Workers also said some editors had used AI to compose emails to literary agents. Employees at the three publishers said literary agents suspected that some Big Five editors were using AI to write rejection letters. HarperCollins bought licences for Claude, ChatGPT and the marketing tool Jasper. Senior leaders also asked junior and senior employees to identify potential applications in monthly brainstorming sessions. At least one division used AI-generated marketing videos after obtaining software licences, while a current employee said staff had received no guidance on what to tell authors or agents about the company’s AI use. Simon & Schuster held an AI use-case contest in May 2024, offering a $10,000 first prize and two $5,000 runner-up awards. Employees later objected at town halls, and the company rolled back some of its promotional efforts. Staff have also begun organizing an open letter against the possible introduction of Skan AI and similar workplace-analysis software; chief executive Greg Greeley said the company had explored Skan’s Blueprint product but had made no decision to adopt it. Simon & Schuster said its employees can use a limited number of approved enterprise AI tools, that use is voluntary and that unvetted tools are prohibited. It said the purpose was to give staff more time for focused and creative work with and for authors. Hachette said it supported AI for operational purposes that help books reach readers, while opposing creative uses, including communications with authors and partners. Hachette Book Group’s published author guidance similarly draws a line between operational uses and AI that replaces a human author, designer, illustrator or translator. It says creative uses require avoidance or the relevant rights holder’s consent, and that internal AI systems are subject to security and author-content checks, including safeguards against training language models on authors’ work without consent. HarperCollins did not respond to Wired’s requests for comment on its policies. As we reported from the Frankfurt Book Fair this month, publishers have increasingly presented AI as useful for translation, research, marketing and routine work while reserving authorship and editorial judgment for people. The Authors Guild has warned that putting manuscripts or authors’ personal information into consumer-facing AI systems without permission may breach copyright or privacy rights. It has urged editors, agents and other publishing professionals not to upload authors’ work to such chatbots without written authorization, and says any contractually permitted use should rely on safeguarded models that do not use the material for training. The pressure from writers predates the latest investigation. In June 2025, more than 70 authors, including Dennis Lehane, Gregory Maguire and Lauren Groff, asked major US publishers not to publish machine-created books, replace staff with AI, or use copyrighted writing without consent or compensation. Their accompanying petition gained more than 1,100 signatures in under 24 hours. Nearly two-thirds of respondents to Publishers Weekly’s 2025 Salary & Jobs Report said their companies used AI in some form.