Human-authored book labels grow amid AI scandals

Schemes to label books as written by humans are gaining ground as publishers and authors respond to scandals over alleged undisclosed use of artificial intelligence.

Several certification initiatives from Britain, France and Germany were being showcased this week at the Frankfurt book fair, where the impact of AI on publishing was widely debated.

French self-publishing company and literary agency Librinova runs a scheme in which authors can pay a small fee to have books assessed and labelled as a “human creation.” Co-founder Charlotte Allibert said about 200 works had received the stamp.

The assessment uses AI detection software, an interview with an auditor and a personal check of the book. Allibert said more authors, especially self-published writers, were worried about being suspected of using AI.

British startup Books By People said its certification mark for human-authored works had been registered by the country’s intellectual property office. Its standards require the writing to originate from the author, while allowing limited AI use for tasks such as spelling or grammar checks.

In the United States, the Authors Guild has introduced a self-certification “human authored” mark. Authors must sign a licence agreement, and about 6,500 titles have received the stamp.

The debate intensified after an anonymous X account alleged last month that Haitian-Canadian author Thelyson Orelien used AI to write his debut novel, “It Was Either That or Die,” citing analysis by AI-detection websites. Orelien has denied the claim and said he believes the criticism is racially motivated.

Some in the book trade remain sceptical of labelling systems. Sebastian Guggolz, head of the German Publishers and Booksellers Association, said there was no simple division between books made with AI and without AI, and said the industry needed to address how its practices should be made transparent.