“It Was Either That Or Die,” by the Haitian-Canadian author Thélyson Orélien, was the hit of the French literary season. Now it has been pulled from the shortlist for a prestigious prize.
By Catherine Porter, Ségolène Le Stradic and Tiffany Hsu for The New York Times
September 25, 2026
When the Haitian-Canadian author Thélyson Orélien arrived in France this month to promote his novel, it was in many ways his dream tour.
Rave reviews have followed his new book, “It Was Either That Or Die,” depicting a migrant’s harrowing journey out of Haiti and across a dozen countries to safety in Canada. So have television appearances and nods from prestigious literary prizes.
That was until this week, when A.I. detection programs in the hands of social media users assessed that the novel was likely written with the help of artificial intelligence, leading the novel to be pulled from consideration for a prestigious literary award.
Mr. Orélien denied using A.I., and his French publisher said he began writing the novel in 2017 and finished a draft in 2019, years before generative A.I. programs appeared commercially.
But Mr. Orélien was still swept into intense debates that have gripped the global publishing industry and shocked France, where literature is so esteemed that writers are interred in the Panthéon, the mausoleum of national heroes in Paris.
Over just a few days, writers and editors filled TV and radio shows, arguing whether A.I. meant “the beginning of the end of literature.” Publishers grappled with the limits of detection tools, what A.I. use was acceptable, if any, and what they should disclose to readers.
And as critics sparred with defenders of Mr. Orélien, who is Black, over whether he was targeted because of his race and his book’s treatment of migrants, he was accused of plagiarism in earlier writings, inflaming the controversy.
On Friday the Goncourt Academy — the institution overseeing France’s highest literary award — decided to pull the novel from its shortlist, given the “consistent findings of analyses by credible researchers and journalists” suggesting the book was, “in all likelihood, largely the product of artificial intelligence.”
“The point is to send a clear message,” the Academy said on social media, “one that affirms that we neither endorse nor approve, in one way or another, the creation of texts generated by artificial intelligence.”
In a statement provided by his French publisher, Mr. Orélien said, “One can discuss my style, but to treat its techniques as evidence of artificial intelligence has nothing to do with literary criticism.”
At a Paris literary event this week, he told the radio station France Inter that he “had nothing to feel guilty about.”
“I knew that the message of this novel wasn’t going to please everyone,” he added.
His French publishing house, Grasset, said in a statement on Tuesday that it “strongly objected” to the “malicious” accusations, and it reaffirmed its “absolute confidence” in the author.
But Les Éditions du Boréal, the prestigious Quebec publisher that first discovered the book, said on Thursday that “at this stage, it is impossible for us to confirm or deny the allegations the author is facing, and in these circumstances, it is crucial not to draw hasty conclusions.”
The publisher said it had suspended promotional activities out of concern for the author and was helping him compile a dossier of early drafts.
The controversy grew out of accusations from a little-followed, anonymous social media account, which a columnist at Le Figaro, a conservative daily newspaper, later revealed he partly ran.
The newspaper widely publicized the account, which was created in June and appears to have started posting this month. Before focusing on Mr. Orélien, it called a handful of other people “slopper,” meaning they overused A.I., without offering details.
The columnist, Samuel Fitoussi, said in an interview that he had no political agenda in any of the accusations and ran the book through an A.I. detection tool, Pangram, because readers had flagged it. He said he thought an anonymous account would be trusted more than his column in a right-leaning newspaper.
“We wanted to open up the discussion,” he said, “not about this particular man, but to raise awareness about A.I.-generated writing and the fact that A.I. is already capable of winning literary awards.”
Three A.I. detection tests
Experts largely agree that, while A.I. detectors can help authenticate material, they are rarely reliable as the final word on a text’s origins. Many lack the sophistication to distinguish the extent of A.I. involvement, though some offer an estimated percentage.
Three A.I. detection tools and several independent experts assessed that A.I. appeared to have been used in Mr. Orélien’s book to some degree.
Pangram, the gold standard of detectors to many experts, was highly confident that A.I. played a part in writing all 34 chapters of the book. Another detector, GPTZero, also reached Pangram’s conclusion.
Max Spero, Pangram’s chief executive, said that in French the tool’s false positive rate — the times it incorrectly identified a human writing sample as A.I.-generated — is 0.0026 percent, or one out of more than 38,000 texts.
Mr. Spero, who has faced some criticism over Pangram’s accuracy, acknowledged there are no independent studies to confirm that rate. And English is the dominant language used to train most detection services, making many experts less confident in verdicts of non-English text.
“It’s likely that A.I. was used in this case,” said Thierry Poibeau, a French researcher who studies A.I. “That said, it’s still hard to say for sure, and we can’t be absolutely certain of the result.”
An appraisal of the book by Winston, a tool whose co-founder was once a professional English-French translator, was less conclusive than Pangram’s and GPTZero’s. Winston found that sections of the book ranged from 12 percent likely to be human up to 66 percent likely; the acknowledgment page was rated 100 percent human.
Mr. Orélien “has been using A.I. in some way,” said Brian Jabarian, a French native speaker and professor at Carnegie Mellon University who has researched detection tools. He called it “quite shocking” to see many common A.I. tells, like the construction of “this is not x, this is y.”
The controversy also spurred scrutiny over Mr. Orélien’s other work.
On Tuesday, Radio Canada, a national public radio station, said that Mr. Orélien had plagiarized four opinion articles published in 2012 and 2013 on French and Canadian news sites. It also said that chunks of his award-winning 2012 short story, “Où va le monde,” resembled parts of a French book published in 1960, “Le Matin des Magiciens.”
Above one of the opinion columns, the Canadian newspaper La Presse has written, “Since this text was published, we have learned that its author committed plagiarism.”
Mr. Orélien did not respond to requests for comment; his Quebec publisher called plagiarism allegations “a cause for concern.”
A literary furor
Mr. Orélien’s book, recounting a high school teacher’s flight from gangs, through the Darién Gap and across the United States, was the talk of French literary circles even before publication. In France, the book has sold more than 54,000 copies in weeks and earned lavish reviews and prizes.
Before the novel was stripped from consideration for the Goncourt prize, the jury’s president, Philippe Claudel, said he was worried about “the way they’re throwing an author’s name to the wolves,” noting that Mr. Orélien was “a best seller, a French-speaking foreigner, Black and who writes about migration,” all features that could potentially fuel a controversy.
Mr. Orélien, 38, grew up in Gonaïves, Haiti, in a middle-class family, with an accountant father and seamstress mother. This year, he recalled in an interview how he left Haiti after the 2010 earthquake that killed more than 200,000 people and destroyed his home, heading for Canada, where he had relatives and a visa.
Several prominent writers have defended Mr. Orélien, including the acclaimed author Virginie Despentes, who said on a popular evening TV show that writers should not answer to A.I. systems, even those designed to detect the work of A.I.
“Thélyson Orélien doesn’t have to justify himself,” she said.
On a major radio show, an academic who studies A.I. effects on culture, Alexandre Gefen, urged writers to embrace the technology, saying, “Literature is not pure inspiration that simply comes to the writer. That’s not the question we should be asking; the question is, is the book good?”
That idea scandalized Abel Quentin, the author of “Sanctuaries,” a novel calling to resist generative A.I. “Artificial literature is an oxymoron,” he said on the same show.
Publishers, who have been warning against A.I. taking their work to train models and pump out “parasite books” on Amazon, were grappling with how to respond. Some have taken a hard line, such as Mr. Quentin’s publisher, Éditions de l’Observatoire, which had already announced a new label, called “Sanctuary,” to guarantee “100% human-written” works, to be stamped on its books.
Emma Saudin, the company’s fiction editor, said her authors would have to sign a contract saying that they had not used A.I. for brainstorming, structuring or writing. “It seems we need to be on the same page from the start,” she said, adding, “I want to publish books that are written by a human being.”
Renaud Lefebvre, the general director of the publishers union, said it was helping members draft an A.I. code of ethics and potential contract language, such as a clause stipulating that no A.I. would be used to produce a work. But he expressed concern about the idea of a “100 percent human” label.
“It’s not up to humans to say they’re humans,” he said. “It’s up to the artificial to say it’s artificial.”
Guillaume Allary, founder of the publisher Allary Editions, said forbidding authors from using A.I. was like barring them from ordinary search engines.
“We’re at the beginning of a redefinition of what an author is,” said Mr. Allary, adding that writers would start using A.I., much as Renaissance artists worked with entire teams of people. This reimagining, he said, would mean doing away with “the romantic image of the writer alone with his pen.”
For now, Mr. Allary did not see a clear solution. “As long as the contract between the reader and the writer isn’t clear,” he said, “there will be scandals.”
Catherine Porter and Ségolène Le Stradic reported from Paris, and Tiffany Hsu from San Francisco. Giulia Imbert contributed reporting from Paris, Norimitsu Onishi from Montreal, and Juliana Castro Varón and Ashley Shannon Wu provided research from New York.
Catherine Porter is an international reporter for The Times, covering France. She is based in Paris.
Ségolène Le Stradic a reporter and researcher for The Times, based in Paris.
Tiffany Hsu reports on the information ecosystem, including foreign influence, political speech and disinformation
