Macmillan staff say workplace AI assistant NOA was unhelpful before it was shut down

Macmillan gave its staff an AI chatbot called NOA, short for No Ordinary Assistant. According to WIRED, employees say the publisher held a party for the tool. They also say almost nobody used it. In October 2025 a company-wide email told staff that NOA would be shut down.

WIRED's report is based on interviews with current and former employees at several Big Five publishers and with bestselling authors, all unnamed because they fear professional consequences. Their accounts describe a gap. Executives promote AI tools, while the people who edit, market and write the books mostly avoid them, distrust them or find them in the way.

The rollout of NOA

An email to staff introduced NOA as a chatbot built by Chaptr, a German software company "backed" by Holtzbrinck, Macmillan's parent company. The email called it "a safe environment where you can input proprietary information without the risk of sharing that data with any third party."

In training sessions, staff were told that NOA could hallucinate and that a human always had to check its output. They were still encouraged to use it to write emails, draft copy, do marketing research and come up with ideas. Several staffers say they were told to paste "only" a "percentage" of an author's manuscript into it. Users could set its tone to options such as "lighthearted," "spartan," "sarcastic" and "regretful." Many of Macmillan's human assistants found the name "offensive," employees told WIRED.

One employee recalls a Chaptr representative telling staff that the company understood publishers dealt with something very important: content. "And here I thought we were publishing literature," the employee says. Later in 2024, staff were encouraged to have NOA write their goals for performance reviews. "We built a specialized prompt to have NOA write goals for you!" one internal email read.

Several employees describe NOA as unreliable. "It was terrible, and no one used it," one says. Workers say that since Chaptr retired NOA, leadership has been trying to bring Google Gemini into publishing work. One editor called the Gemini evaluations that some staff produced of literary agents' submission letters "useless."

Kelly Collins, Macmillan's vice president of communications and community, told WIRED that the company uses Gemini for its "business" but that leadership has issued no directive to use it "for any part of the publishing process, including to evaluate submission letters from agents." She said the accounts "do not accurately reflect how NOA was rolled out and communicated." WIRED asked which parts were inaccurate, and she did not say.

The same pattern at Penguin Random House

Staff at Penguin Random House, the largest of the Big Five, say AI learning sessions encouraged them to use ChatGPT Enterprise. This happened while PRH's German division is suing OpenAI for copyright infringement. Hachette and Macmillan are also suing leading AI companies.

In February, representatives from Hay House, a self-help publisher PRH acquired in 2023, showed some PRH employees a slide deck. It described ChatGPT Enterprise as a "core productivity, creativity, and collaboration tool." One of the four return-on-investment pillars in the deck was "Creativity," including "storytelling innovations."

Every PRH worker WIRED spoke with said they avoid ChatGPT whenever possible. "It makes it harder for us to do our jobs," one says. Others say their managers have told them not to use AI for creative tasks such as publicity copy and cover design, and that the mixed messages have caused confusion. PRH spokesperson Claire von Schilling said the company explores AI "guided by clear principles and safeguards" and that its imprints decide for themselves whether AI fits particular workflows.

Authors say trust is eroding

The Big Five authors WIRED interviewed have together sold more than 30 million copies. One says their publisher sent marketing copy with quotes from their book that AI had invented, then denied AI was involved. Another says no Big Five publisher would promise during contract talks that AI would not be used on their books in any way.

Authors connect this to layoffs. They say editors with less time are turning to AI to cope. "We're seeing editors not reading the material at all and using Claude to create marketing language that barely has any relation to the book," one says. Others say editors have asked them to write their own revision plans.

Every author WIRED spoke with wants publishers to disclose when and how they use AI. One wants the Big Five to agree on a shared set of principles. Some are thinking about self-publishing. Brandon Sanderson, who publishes some books with Macmillan's Tor imprint, raised more than $15 million on Kickstarter in 2022 to publish four books himself. John Scalzi, who signed a 10-book deal with Tor in 2024, wrote publicly that AI-generated publicity is "immediately dismissible."

What it means for work

The tasks publishers assigned to AI are the routine edges of editorial work: emails, marketing copy, research, triaging submissions, even writing performance goals. Staff say the tools did these tasks badly enough that checking the output took more effort than doing the work. That is why many stopped using them. The people affected first are editorial assistants, publicists and editors, whose jobs depend on knowing the book. When that knowledge is skipped, authors notice. Still unresolved: whether publishers will disclose their AI use, how they square it with their own copyright lawsuits, and whether authors with large readerships decide they can do the same work without a publisher.