‘Human-made’ book labels gain ground as publisher drops novelist at centre of French AI row

Books could soon carry a 'Human-made' stamp, much as food carries an organic label, as publishers try to reassure readers amid growing suspicion that artificial intelligence is being used to write novels.

Several certification schemes, from France, the UK and Germany among others, were showcased in the start of October at the Frankfurt Book Fair, the largest in the world, where the impact of AI on creative writing was one of the most hotly debated topics.

Interest in the idea has grown rapidly following a series of controversies this year. 

The most high-profile concerns Thélyson Orélien, whose novel was dropped from the Prix Goncourt selection, France’s most prestigious literary prize, amid claims it was largely AI-generated. 

On Wednesday (October 21), his publisher in Quebec announced it was cutting ties with him (see below).

The Orélien affair is not the only case to have fuelled the debate this year. Earlier in 2026, publisher Hachette cancelled the US release of Shy Girl, a horror novel by Mia Ballard, and withdrew it from sale in the UK after it was accused of being AI-generated.

Ms Ballard told the New York Times she had not used AI herself. However, she said an acquaintance she had hired to edit the book, which she originally self-published, had done so.

A 'création humaine' label in France

In France, the self-publishing platform and literary agency Librinova earlier launched a création humaine (human creation) label.

Authors who want it must first put their manuscript through AI-detection software. They then sit an interview with an auditor, who questions them about how the book was actually written. Around 200 titles have been certified so far.

Librinova co-founder Charlotte Allibert said the label was a response to mounting anxiety among writers, particularly those who publish their own work, who are increasingly afraid of being accused of using AI.

'Organic literature' in the UK

Across the Channel, the start-up Books By People has created its own certification mark, which was formally registered with the UK Intellectual Property Office earlier this month.

To qualify, a book must meet the company's standard for what it calls 'organic literature'. The writing must come from the author, although limited use of AI is allowed for tasks such as checking spelling and grammar.

Co-founder Esme Dennys said the aim was to build a framework that was transparent and trustworthy enough for both publishers and authors to rely on.

Publishers pay a fee to join and can then apply for their authors' books to carry the Books By People mark. Five independent publishers have signed up so far, with more expected to be announced in the coming months.

In the US, the writers' body the Authors Guild already offers a 'Human Authored' label. It works on a self-certification basis, and roughly 6,500 books now display it.

Not everyone is convinced

Despite the string of scandals, some in the book trade doubt that labels are the answer.

Most of the current schemes permit a small amount of AI use, for research or proofreading for example, and critics fear this leaves too much room for interpretation. Others worry that the checks used by these fledgling initiatives could be easy to get round, allowing machine-written books to pass as human work.

Sebastian Guggolz, head of the German Publishers and Booksellers Association, said the industry was still struggling to work out its approach.

Speaking at the Frankfurt fair, he argued that the issue could not be reduced to a simple choice between 'with AI' and 'without AI', and that systems merely grading how much AI went into a book were unlikely to be enough.

Instead, he said, every writer and the profession as a whole needed to ask a deeper question: how do we want to make our working practices transparent?

Publisher drops Thélyson Orélien

Mr Orélien's debut novel C'était ça ou mourir (It Was That or Die), published in France by Grasset, was one of the critical and commercial hits of the autumn. In September, however, an X account (formerly Twitter) called Balance ton Claude claimed it had largely been written by AI, and earlier acts of plagiarism by the author resurfaced.

At the end of September the Académie Goncourt removed the book from its selection for France's top literary prize, citing the plagiarism, which he has since admitted, and analyses suggesting the text was very probably AI-generated. 

The Prix Renaudot, another major French literary award, took a different view and kept the book among the nine titles on its second selection, announced on 7 October. Sales have continued to climb despite the controversy. Mr Orélien denied using AI to write it and said the attacks against him were racial harassment.

On October 21, his Quebec publisher Boréal said an external analysis of his working drafts had found traces of AI in the files' metadata, despite his repeated assurances. It is ending its collaboration with him and will not reprint the novel.

Mr Orélien told Libération the book was written by him, but acknowledged using AI for historical research.

What is AI-assisted writing worth?

An opinion piece published in Le Monde argued that the affair should prompt a more practical debate than the question of a work's 'authenticity', which has dominated discussion so far. The real issue, it said, is what still justifies rewarding intellectual work done with AI, whether with money, prestige or status.

The author suggested that Mr Orélien may have been sincere in saying he wrote the book if, like many people, he used AI to research, summarise and draft to the point where he could no longer tell his own ideas from recycled phrasing. Because AI builds on what has already been widely written, its output can feel familiar and self-evident to the user, and only close attention and broad knowledge allow writers to keep control.

The piece also pointed out that the alarm was not raised by the literary professionals who praised the novel and picked it for prizes, but by anonymous social media users relying on an AI tool to detect AI.

As AI increasingly industrialises intellectual output, it concluded, the value of human work will lie less in what sounds good than in what rings true, and in the hard-won judgment, culture and skill that make up a craft.