Bestselling novel by Thélyson Orélien withdrawn from the Prix Goncourt after accusations of AI use
The Pangram detector flagged the bestselling novel by Thélyson Orélien as likely written by AI, so the Goncourt Academy withdrew it from the shortlist of the Prix Goncourt. The author denies this and says he had been writing the text since 2017, before generative AI was commercially available.
Last Friday, the Goncourt Academy withdrew the novel by Thélyson Orélien from the shortlist of the Prix Goncourt, the most prestigious French literary prize, after an anonymous account on X claimed the work had been written "almost exclusively by artificial intelligence." Behind the claim was, among others, Samuel Fitoussi, a columnist for the conservative newspaper Le Figaro, who admitted involvement in the anonymous group. The Academy justified the withdrawal by citing "the matching conclusions of analyses by trustworthy researchers and journalists," according to which the work is "very likely largely the product of artificial intelligence." The accusations were based on the Pangram detector, described as the "gold standard" among AI detectors.
Orélien, a Haitian-Canadian author living in Canada since 2010, denies the accusations. According to his French publisher, he began working on the novel It Was Either That or Die (C'était ça ou mourir) in 2017 and finished the draft in 2019 — that is, before generative AI became commercially available. The author stated that he has been the target of "numerous racist attacks," and together with publishers in France and Quebec is preparing documentation to defend his authorship.
At the same time, the Canadian papers La Presse and Radio-Canada published findings about plagiarism in Orélien's earlier texts from 2012–2013: according to a Radio-Canada investigation, at least five of his columns and commentaries were partly copied from other authors, and La Presse, using Pangram, found at least four sentences in his award-winning 2012 short story that matched a 1960 book by Louis Pauwels. Orélien has not responded to these findings.
The case has sparked a debate about the reliability of AI detectors — studies do report an accuracy of over 95% for Pangram with a minimum of false alarms, but errors do occur, as also demonstrated by the previously disputed and subsequently cleared winning work of the Commonwealth Writers Prize. Opinions among French authors diverge: writer Virginie Despentes rejected the idea that authors should have to account for their use of AI, while other commentators point to stylistic features considered typical of AI-generated text.
Why it matters
The case shows that even AI detectors with a high claimed accuracy can lead to serious reputational consequences based on inconclusive results, which is relevant to anyone using such tools to assess other people's texts, or who publishes their own work and could be accused in this way.
Two audiences, two different impacts
What this means
For individuals
Authors and other creators of texts need to be aware that AI detectors can mistakenly flag even human-written text (especially text with an unconventional style) as AI-generated, which can damage their professional reputation without clear proof.
For a business
Publishers, academic institutions, or companies using AI detectors to screen texts face the risk that even a tool with a high claimed accuracy (according to studies, over 95%) can produce a false accusation with real reputational and legal consequences, as shown by the withdrawal of a bestseller from a prestigious competition.
Risks and complianceCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.