Authorship dispute over AI use in novel

Haitian-Canadian author Thélyson Orélien's novel C'était ça ou mourir has sparked debate over the role of artificial intelligence in literature, after it was accused of being largely generated by AI. The novel, which focuses on an Haitian refugee's journey to Canada, won the Fnac novel prize in September 2026 before the controversy began.

Accusations and publisher's response

An anonymous account called "Balance ton Claude" claimed the novel was nearly entirely produced by AI, citing the Pangram detector. The Goncourt Academy later removed the book from its selection, citing AI use and plagiarism. Orélien denied using AI but admitted to past plagiarism. His publisher, Boréal, later found AI tags in the metadata of some files and ended its relationship with Orélien.

Pangram test results

The novel was tested using the Pangram tool by La Presse and received a score of 94%, while Quebec novelist Bryan Perro’s 2006 book scored 73%. These results suggest that a book written before AI could still score highly, but the test's reliability remains questionable.

Key questions raised by the affair

The controversy has sparked three major questions: whether the sentences were written by a machine, whether AI use should have been declared, and whether a writer's own criteria for judging their work are truly their own. The last question is considered the most complex, as it involves how a writer's standards are formed and whether they are genuinely self-determined.

AI's influence on writers' independence

The article discusses how AI might affect a writer's intellectual independence by influencing their judgment. If a writer receives similar advice from multiple sources, including AI, it may seem credible, even if the advice comes from a single common source. This could undermine the writer's ability to evaluate their work independently.

Predictions and racial bias

Essayist Samuel Fitoussi, who later admitted to starting the accusatory account, predicted that established authors would benefit from a presumption of authenticity, while unknowns would face suspicion. This prediction was quickly seen in the case of an unknown French first-time novelist. However, it is crucial to distinguish between valid concerns about the manuscript and the racial harassment reported by Orélien.

Limitations of AI detection tools

The article concludes that while AI detection tools can provide clues about a text's origin, they cannot determine whether the writer's criteria for judgment were genuinely formed through their own process. This applies equally to those who accuse others of using AI, as their own standards for originality and authenticity also have a history.