Major Publishers Quietly Adopt AI Tools While Publicly Resisting Generative Tech

Three of the Big Five U.S. publishers—HarperCollins, Simon & Schuster, and Hachette—have been integrating large language models into daily workflows without public disclosure or author consent. Staff report that tools such as Claude, ChatGPT, and the agentic platform Jasper are now drafting publicity emails, back-cover copy, marketing videos, and even cover art concepts.

At HarperCollins, senior leadership purchased enterprise licenses for multiple LLMs and then "voluntold" junior and senior employees to become "AI Champions," tasking them with finding use cases in monthly brainstorming sessions. When staff raised legal, ethical, and environmental concerns, a meeting lead dismissed them as "the cost of doing business," and an internal portal later downplayed AI's carbon footprint.

The push comes despite entry-level publishing salaries in New York averaging just $47,583 in 2023. Employees describe the AI spend as a slap in the face after years of unmet requests for fair compensation. Meanwhile, the same houses have publicly sued Google for copyright infringement and canceled books by authors suspected of using generative AI—a contradiction that has fueled internal revolt.

Why It Matters for GPU / AI Infrastructure

Publishing's quiet adoption of LLMs signals a growing, steady demand for inference compute across creative industries. Each publicity blurb, cover iteration, and marketing video generated by models like Claude or Jasper consumes GPU cycles at scale. For infrastructure providers, this translates into predictable, recurring workloads that favor low-latency, high-throughput GPU clouds—precisely the capacity that specialized providers such as AiGpu are built to deliver.