Bernard Minier writes the kind of page-turning crime novels that sell by the stack at airport bookstores. And right now, he sounds like a guy watching someone wheel a printing press into his living room.
The French bestselling author, known for his Martin Servaz thriller series, says he’s in “stupefaction” at what generative AI can do. Not mildly concerned. Not “keeping an eye on it.” Stunned. Because he sees two big risks barreling toward the book business: machines cranking out passable novels at scale, and machines eating the jobs around books, especially translation and editing.
Minier has sold 2 million books. He thinks the machines can write his genre now.
Minier isn’t some obscure literary purist clutching pearls. He’s a commercial powerhouse in France: more than2 million copiessold, translated into roughly15 languages. That kind of success buys you a megaphone, and a front-row seat to what could undercut your whole lane.
His fear is simple: crime fiction runs on structure. Motive, misdirection, pacing, reveals. The same “rules” that make a good procedural addictive also make it easier to imitate. And today’s models, ChatGPT, Anthropic’s Claude, Google’s Gemini, can already spit out coherent plots, recurring characters, and chapter-by-chapter momentum that holds together for hundreds of pages.
Will it win the Edgar Award? Probably not. Will it flood the market with “good enough” thrillers that look like real books on a Kindle screen? That’s the nightmare.
Translation is the soft underbelly, and AI is already biting
Minier’s books travel because translators make them travel. That’s the unglamorous truth of international publishing: you don’t get “translated into 15 languages” without a small army of skilled humans.
But tools likeDeepLandGoogle Translatehave gotten frighteningly competent for certain kinds of prose, especially straightforward narrative. Publishers, always hunting for margin, are already testing AI-assisted translation workflows. The pitch is “efficiency.” The result, for a lot of working translators and editors, looks like fewer paid hours and more low-rate cleanup work.
And once publishers get used to paying less for language, they rarely go back to paying more out of kindness.
A $3 billion French industry is staring at the same AI mess we are
France’s book industry pulls in about€2.7 billiona year, roughly$3.0 billionat current exchange rates, according to the country’s publishing trade group, theSyndicat national de l’édition. It directly employs around80,000 people, before you even count bookstores, distributors, freelancers, and the rest of the ecosystem.
Now drop generative AI into that machine and watch the incentives kick in. Platforms that can generate full “books” on command are already here. Amazon has already had to rein in AI-generated titles on its marketplace because the volume got ridiculous. That’s not a sci-fi warning; it’s a content-moderation problem happening in real time.
Then there’s the legal brawl underneath it all: copyright. These models learn by ingesting oceans of existing writing. Authors in the U.S. are already suing AI developers, arguing their books were used without permission. France’s authors’ society, theSociété des gens de lettres, a major writers’ advocacy group, is watching closely, because the question isn’t academic. It’s about whether creators get paid or get mined.
ChatGPT hit 100 million users in two months. Publishing doesn’t move that fast.
The adoption numbers are the part that should make any publisher sweat.ChatGPT reached 100 million active users in two monthsafter launch, one of the fastest consumer tech takeoffs ever recorded. That kind of speed doesn’t give slow-moving industries time to “work it out.” It just runs them over.
Plenty of writers are already using AI in smaller, quieter ways: brainstorming, beating back writer’s block, punching up outlines, generating alt scenes. Some publishers are experimenting too.Penguin Random HouseandHarperCollinshave talked about controlled uses, summaries, copyediting assistance, internal workflows. That’s the careful version.
Minier is talking about the not-careful version: a market where the most formula-driven genres, thrillers, romance, certain sci-fi, get swamped by machine-made books that are cheap, fast, and “fine.” And once readers get trained to expect 99-cent infinite content, the people who actually write for a living get squeezed first.
He’s not predicting the death of literature. He’s warning about the death of the middle class in publishing. Different tragedy. Same funeral home.
