Staff arriving at Macmillan Publishers' Lower Manhattan offices on a morning in January 2024 found the building decorated to welcome a new teammate called NOA, according to WIRED. Purple balloons and banners filled the cafeteria, NOA's name appeared in illuminated marquee letters, and doughnuts on a plate spelled out "NOA IS HERE," the outlet reports, citing several employees who were there.
Why it matters
The episode offers a detailed look at how one of the largest US book publishers introduced generative AI into editorial work — and how quickly the experiment wound down. Macmillan is generally described as the smallest of the Big Five American trade publishers by annual sales, and its imprints include Farrar, Straus and Giroux and Tor Books.
According to WIRED, the NOA story also sits alongside a broader pattern: staff across the Big Five say executives are urging them toward AI tools even as several of the same companies, including Macmillan, pursue copyright lawsuits against leading AI developers. Meanwhile, the outlet reports, a number of bestselling authors have clashed with their publishers over AI and are weighing whether to leave traditional publishing.
Promises of privacy, warnings about errors
The introductory email told employees that NOA offered a secure space where they could enter confidential material without that data being passed to outside parties, WIRED reports — in effect, a closed system that would not keep or reuse information for training.
During training sessions, staff were cautioned that NOA could produce false information and would always require human verification, according to the outlet. Even so, employees say they were urged to use it for drafting emails and copy, conducting marketing research, and brainstorming. Several staffers told WIRED they were instructed to paste only a portion of an author's manuscript into the tool at a time. Users could also set NOA's tone, choosing among options that included lighthearted, spartan, pessimistic, sarcastic, and regretful.
One employee recalled a Chaptr representative telling an early in-person session that the company recognized publishers were handling something of great importance — content. "And here I thought we were publishing literature," the employee said, per WIRED.
By later in 2024, Macmillan was encouraging employees to lean on NOA when writing their performance-review goals. "We built a specialized prompt to have NOA write goals for you!" one email said, according to the outlet.
The tool did not win many converts. Several employees described it to WIRED as unreliable and of little use. "It was terrible, and no one used it," one said. WIRED interviewed multiple publishing employees for its report, some of whom have since left their jobs; the outlet withheld their names, and those of the authors it spoke with, because they feared professional consequences.
A quiet retirement, then Gemini
In October 2025, staff learned through a company-wide email that Chaptr would be discontinuing NOA, WIRED reports. After that, employees say, Macmillan leadership sought to bring Google Gemini into the publishing workflow. Some staffers tried using Gemini to assess submission letters from literary agents, but one editor told the outlet the resulting evaluations were useless.
Kelly Collins, Macmillan's vice president of communications and community, told WIRED that the publisher uses Gemini for business purposes but that leadership has issued no instruction to deploy it anywhere in the publishing process, including the review of agent submissions. She said the accounts shared with WIRED do not accurately reflect how NOA was introduced and communicated, but declined to specify what was inaccurate when asked, according to the outlet. Collins added that Macmillan takes a considered approach to AI, emphasizing employee awareness and education while keeping people at the center.
WIRED also reported on conditions at Penguin Random House, the largest of the Big Five. Several workers there said they had been encouraged to use ChatGPT Enterprise during AI learning sessions, even though PRH's German-language division is suing OpenAI for copyright infringement. In February, representatives from Hay House — a self-help publisher PRH acquired in 2023 — gave some PRH employees a presentation describing ChatGPT Enterprise as central to productivity, creativity, and collaboration since its adoption a year earlier, with creativity listed among four return-on-investment pillars, per the outlet.
All of the PRH employees WIRED spoke with said they steer clear of ChatGPT when they can. "It makes it harder for us to do our jobs," one said. Others said their managers have told them not to use AI for creative work such as publicity copy and cover design, which has left some staff confused. Claire von Schilling, PRH's executive vice president and director of corporate communications and social responsibility, told the outlet the company treats AI like any major new technology, exploring appropriate uses under clear principles and safeguards that protect authors' and the company's intellectual property and preserve the human role in creativity.
What to watch: whether Macmillan and its peers formalize AI guidance for editorial staff, and how the pending copyright cases against AI developers shape those internal policies.
