Publicly, the elite tier of American book publishing—collectively known as the Big Five—has postured as a fierce bulwark against the encroachment of generative artificial intelligence. Over the past year, major industry players have initiated high-profile legal battles, such as Hachette Book Group joining class-action lawsuits against tech giants like Google for alleged copyright infringement in training large language models (LLMs). Simultaneously, publishers have swiftly canceled book contracts for authors suspected of utilizing artificial intelligence to draft their manuscripts, as seen with controversial titles like the Hachette horror novel Shy Girl and Macmillan’s crime thriller Call Me, I’ll Hide the Body.
However, an investigation into the internal operations of at least three of the Big Five publishing houses—HarperCollins, Simon & Schuster, and Hachette—reveals a stark dichotomy. Behind closed doors, these corporate entities have systematically integrated generative AI tools into their everyday workflows. According to more than two dozen industry workers who spoke on the condition of anonymity to protect their employment, LLMs such as Anthropic’s Claude, OpenAI’s ChatGPT, and enterprise marketing platforms like Jasper are now routinely deployed to draft emails to literary agents, formulate back-cover and publicity copy, design promotional materials, and even compose corporate memos, often under direct mandates from upper management.
The Quiet Onboarding of Enterprise AI
The integration of generative AI into legacy publishing operations has largely bypassed public scrutiny and author consent. At HarperCollins, senior leadership procured enterprise licenses for Anthropic’s Claude months ago. According to current employees, executives initially struggled to identify practical applications for the software. In response, leadership directed dozens of junior and senior staff members to act as “AI Champions,” tasked with discovering corporate use cases during mandatory monthly brainstorming sessions.
These internal rollouts have frequently met internal resistance. During a spring brainstorming session, when employees raised valid legal, ethical, and environmental concerns regarding the carbon footprint and data privacy implications of LLMs, a meeting facilitator allegedly dismissed the qualms as “the cost of doing business.” Shortly thereafter, corporate updates to the HarperCollins online employee portal minimized these concerns, inserting a statement asserting that while AI carries an environmental toll, it accounts for a minimal fraction of an individual’s total digital carbon footprint.
Staff members have expressed deep frustration over these corporate priorities, contrasting the rapid capital expenditure on enterprise software with modest compensation structures. Data from Publishers Weekly notes that the average entry-level salary for New York City publishing employees at major houses hovered around $47,583 in 2023. Workers argue that funds spent on unwanted software licenses could have been better allocated toward workforce compensation.
Beyond HarperCollins, Simon & Schuster has similarly pursued internal adoption. In May 2024, Simon & Schuster hosted an internal employee contest offering financial incentives—including a $10,000 grand prize and $5,000 for runners-up—to staff members who devised the most innovative applications for AI within the publishing pipeline. OpenAI representatives were also brought in to conduct workshops for Simon & Schuster employees, explicitly recommending the use of ChatGPT for composing correspondence to literary agents—a practice that agents later identified due to recurring stylistic markers characteristic of LLM generation.
A Timeline of Integration and Resistance
The timeline of AI integration across the publishing sector reflects an accelerated push by corporate leadership that has continuously clashed with the traditional human-centric ethos of the literary trade:
- Late 2022 to 2023: Major industry executives, including HarperCollins CEO Brian Murray, publicly signal optimism regarding AI, framing it as an opportunity for operational expansion rather than a threat, particularly in foreign rights translations and audiobooks. Private corporate exploration of enterprise AI tools begins.
- May 2024: Simon & Schuster hosts an internal AI innovation contest with cash prizes to incentivize workflow automation among employees, while parallel enterprise software purchases occur across HarperCollins and Hachette.
- Late 2024: Literary agents begin noticing uniform patterns in rejection letters and pitches from major publishing houses, suspecting the unchecked use of LLMs by editorial staff. Contractual clauses restricting the input of unpublished manuscripts into open-loop AI models begin proliferating.
- Mid-2025: Workforce reductions leave skeleton crews managing increased publication volumes, driving exhausted employees to reluctantly adopt AI tools as a necessity to manage workloads.
- Late 2025: Plans regarding workplace monitoring software, such as Skan AI, leak within Simon & Schuster, prompting an employee-led open letter and immediate public pushback, resulting in executive denials and revised communications.
The Skan AI Controversy and Surveillance Fears
The tension between corporate cost-cutting mandates and editorial staff reached a boiling point in late 2025 when rumors spread regarding an experimental trial of Skan AI at Simon & Schuster. The software, allegedly introduced at the urging of private equity firm KKR—which acquired Simon & Schuster in 2023—is marketed as a tool to provide leadership with the metrics required to prioritize automation and reduce operational costs.
In response to the trial, employees organized and circulated an open letter to CEO Greg Greeley expressing “strong opposition” to workflow surveillance and potential workforce displacement. The protest gained external attention, leading Greeley to issue a corporate memo confirming that the company had explored Skan AI’s Blueprint product but asserting that no definitive decisions had been made regarding productivity-monitoring tools. Notably, forensic checks via AI-detection platforms such as Pangram indicated that portions of Greeley’s explanatory memo were themselves generated by artificial intelligence.
The root of this employee friction lies in widespread staffing reductions. Workers across the Big Five report that long-standing editorial and publicity teams have been systematically hollowed out, leaving small teams of five to six employees handling double or triple their historical book volume. For many workers, turning to AI is not a preferred creative choice, but a survival mechanism forced upon them by executive mandates to maintain impossible production schedules.
The Threat to Intellectual Property and Agency Relations
The underground adoption of generative AI has severely strained relationships between publishers and the literary agent community. Agents have grown increasingly alarmed over the possibility that editorial staff might input sensitive, unpublished author manuscripts into open-loop LLM platforms to generate rejection letters, marketing copy, or editorial feedback.
To combat this vulnerability, literary agencies have rapidly updated standard representation contracts to explicitly forbid the processing of client material through unvetted language models. Unlike closed-loop, air-gapped systems operating on secure internal networks, consumer-facing models risk ingesting and retaining proprietary intellectual property.
Publishing executives maintain that their internal guidelines prohibit unvetted applications. Wibke Grutjen, global chief marketing and communications officer at Simon & Schuster, stated that employees have access to a limited selection of vetted enterprise-level tools on a voluntary basis, emphasizing that the primary objective is to free up time for deeper creative focus alongside authors. Similarly, a Hachette spokesperson affirmed support for operational efficiencies that expand book reach while explicitly distancing the company from creative uses or AI-generated communication with external partners.
Broader Industry Implications and the Road Ahead
As conglomeration continues to reshape the publishing landscape—a dynamic analyzed by scholars such as Dan Sinykin in Big Fiction: How Conglomeration Changed the Publishing Industry and American Literature—the cultural divide between corporate C-suites and creative workers widens. Corporate leadership remains focused on efficiency, margin expansion, and automated translation rights, as evidenced by Europa Editions executive publisher Michael Reynolds praising AI translation tools at industry conferences.
Conversely, independent publishers and literary agents argue that the human element is foundational to the cultural value of literature. Without transparent guidelines, clear disclosures, and collaborative communication from major publishers, the systemic integration of artificial intelligence risks eroding the trust upon which the modern literary ecosystem is built. As industry professionals navigate this transition, the overarching demand remains simple: accountability, transparency, and a clear demarcation between technological utility and the irreplaceable work of human creators.
