AI in Publishing: The Quiet LLM Use That Has Staff Revolting

AI in publishing is no longer a quiet experiment: staff at three major publishing houses tell WIRED that large language models (LLMs) are being used daily for publicity, cover art, back cover copy, and internal emails. What started as a discreet productivity tool has become a friction point between management and a workforce worried about jobs and craft.

From curiosity to conflict

The adoption of AI in publishing didn't happen overnight. For years, publishers have automated logistics and marketing tasks with conventional software. The arrival of generative models like ChatGPT and its rivals accelerated that trend: suddenly, an LLM could draft a synopsis, propose a tagline, or generate cover variations in seconds. What was missing was an open debate about limits, transparency, and compensation.

The result is a culture clash: creative and communications teams see AI as a threat to their professional judgment, while some executives frame it as an essential skill for the future.

What's actually happening

According to WIRED's sources, across the three publishers AI is used for:

• Publicity and marketing: generating press releases, media pitches, and social copy.
• Cover art: drafts and visual variations created with generative tools before a human designer takes over.
• Back cover copy: synopses and author bios written or polished with LLMs.
• Internal emails: automated replies and draft communications between departments.

The most uncomfortable detail: several executives have reportedly asked junior staff to act as AI ambassadors, defending its use to skeptical colleagues. In other words, the political burden of a top-down decision is being pushed onto those with the least power to question it.

Who wins and who loses

The immediate winners are publishers cutting costs and time on repetitive tasks. The losers, for now, are creative workers watching their expertise get replaced by automated drafts, and readers who may get more generic copy if no one edits with judgment. My take: the problem isn't AI itself, but the lack of transparency and the absence of clear agreements on how it's used, who oversees it, and how the savings are shared.

What comes next

Expect unions and works councils to push for AI clauses in collective bargaining agreements. We'll also see more leaks from employees exposing undeclared uses, as has happened in other industries. Public pressure could force publishers to publish ethical guidelines, but without real enforcement mechanisms, the risk is that they remain PR exercises.

The underlying question remains open: can an industry built on the human voice delegate precisely that to a model that has no voice of its own?

Source: Wired