Wednesday, November 18, 2026

INFORMOPEDIA

World news and deep-dive explainers — climate, economy, technology, science, culture, health

Breaking
Culture

Who Wrote This? ‘Human-Authored’ Book Labels Gain Ground Amid AI Scandals

France’s most prestigious literary prize took action on a Friday in late September. The Académie Goncourt removed a smash-hit novel by Haitian-Canadian author Thélyson Orélien from its longlist after an anonymous account on X alleged he had used artificial intelligence to write his debut, It Was Either That or Die. Orélien denies it. He believes the attacks are racially motivated. But the jury said it wanted to preserve the integrity of the prize and stand against the “creation of texts generated with the assistance of AI.”

The scandal, built on AI-detection software’s reading of book excerpts, has become the emblem of publishing’s new anxiety. And it has given fresh momentum to an idea that would have sounded absurd a few years ago: certifying that a book was written by a person.

The most ambitious scheme comes from a British startup called Books by People. Earlier this month it won approval from the UK’s Intellectual Property Office to distribute a certification mark — a thumbprint-style stamp that will appear on book covers — signaling that the contents were not generated by AI. Five independent publishers have signed up, and the first certified book is expected to reach shelves before the end of the year.

Co-founder Esme Dennys told AFP the goal is “a comprehensive, transparent, reliable framework that publishers and authors can fall back on.” The verification goes beyond a signature: each manuscript is run through proprietary software and early drafts and research materials supplied by the writer and editor are manually examined for what Dennys calls “indicators of humanity” — the traces of a natural process of drafting and redrafting.

The Americans got there first. In January 2025, the Authors Guild unveiled its “Human Authored” certification, a self-certification system in which writers submit book details through a portal, sign a license agreement, and receive a unique numbered mark for covers, spines, and marketing. About 6,500 titles have received the stamp so far, according to AFP. The rules tolerate a de minimis amount of AI-generated text to cover grammar and spell-check tools, and using AI for research or brainstorming does not disqualify a book — as long as the prose itself is human-written. “Readers deserve to know whether they’re experiencing authentic human creativity,” said Guild CEO Mary Rasenberger.

South Korea is going further, and more systematically. On September 29, the Korean Publishers Association, the Writers Association of Korea, and the Publishing City Culture Foundation jointly announced a three-level labeling standard: Level 1 for “Human Authored Work” where AI played only a limited role such as proofreading; Level 2 for “AI-Assisted Work”; Level 3 for “AI-Generated Publication” lacking human verification. Levels 2 and 3 must be flagged on the imprint page, and authors must disclose the level of AI use when submitting manuscripts. It is voluntary — “a promise from the field,” as the association puts it — but it is the industry’s most honest attempt to admit that AI involvement is a spectrum, not a binary.

The momentum is real, and so are the scandals feeding it. Earlier in 2026, Hachette cancelled the US release of Mia Ballard’s horror novel Shy Girl and withdrew it from sale in Britain over allegations of AI use. Ballard told The New York Times she had not personally used AI, but that an acquaintance she hired to edit the book had. Each case has made one thing obvious: in a market flooded with machine-generated books, a human name on a cover no longer settles the question of who actually wrote it.

Not everyone in the trade is convinced labels are the answer. Sebastian Guggolz, head of the German Publishers and Booksellers Association, has argued there is “no simple ‘with AI’ or ‘without AI’.” He is right, and the schemes know it: most allow some AI use for research and grammar, which leaves a gray zone open to interpretation. There is also the uncomfortable fact that verification depends on the same detection technology whose reliability is now deciding literary careers — the Orélien case itself rested on an anonymous account’s software analysis.

Here is what the labels really are: an admission that trust has collapsed, packaged as a solution. For decades, a publisher’s colophon was an implicit guarantee that a book was the work of the person whose name was on the spine. AI has broken that guarantee so thoroughly that the industry now has to rebuild it, line by line, with stamps and portals and manuscript audits.

That rebuild is worth attempting. Readers genuinely want to know what they are holding, and authors who do the slow work of writing deserve a way to say so. But a label can only tell you what a publisher claims happened. The Goncourt episode suggests the harder question is the one nobody has answered: in a world where a prize jury can be moved by a stranger’s detector results, who gets to be the judge of a human mind — and what happens when the judge is wrong?

Avatar photo
Informopedia News Desk

Join the discussion

Your email address will not be published. Required fields are marked *