Goncourt Prize Drops Novel After AI-Writing Allegations: Here’s What Happened

Published: September 26, 2026
Byline: Ahaqir

The Académie Goncourt has removed Thélyson Orélien’s C’était ça ou mourir from its first list for the 2026 Prix Goncourt. The September 25 decision followed allegations that artificial intelligence helped produce the novel and separate revelations concerning plagiarism in the author’s earlier work. Orélien has denied using AI to write the book. The academy’s decision is real; the extent, if any, of AI involvement in the novel remains disputed. In its statement reported by AFP, the academy said its withdrawal was meant to protect the integrity of the prize.

Which book did the Goncourt remove?

C’était ça ou mourir, roughly, It Was That or Die, is Orélien’s debut novel. Its protagonist, Jonas Dorléon, is a Haitian history teacher who flees violence in Port-au-Prince and travels through the Americas in hopes of reaching Canada. French publisher Grasset’s book page lists an August 19, 2026, French publication date and the 2026 Prix du roman Fnac, which the novel has already won.

The book had been named to the Goncourt’s initial selection. It had also attracted wide attention during France’s fall publishing season. The removal concerns the Goncourt selection, not a withdrawal of its Fnac award or a ban on sales.

How did the AI allegations begin?

On September 21, an account called Balance ton Claude posted results it said came from running passages of the novel through Pangram, an AI-text detector. The account claimed the passages were flagged as overwhelmingly AI-generated. Those percentages are the account’s reported detector results, not an independently established measure of how much of the book was written by a machine. Euronews’ account of the initial posts and the author’s response traces how the allegation spread.

Orélien rejected the accusation. In a response circulated through Grasset, he argued that repetition and rhythm in his writing draw on Haitian and Caribbean literary traditions and should not be treated as proof of AI use. Grasset initially expressed confidence in its author. Orélien has also described racist attacks amid the controversy.

The distinction matters. A detector analyzes a finished text; it does not observe an author’s drafting process. Other detection tools have produced different assessments of this novel. A score on its own cannot settle the authorship question, even when it becomes a powerful claim on social media. The Guardian’s reporting documents both the author’s denial and the disagreement among detectors.

Why did the academy remove the novel?

The academy cited two issues: allegations concerning AI in C’était ça ou mourir and plagiarism revelations involving Orélien’s past writing. In its statement, it said analyses by researchers and journalists led it to believe the novel was, in its assessment, very largely the product of AI. It also said it wanted to uphold its mission of honoring literature written by people. Those are the academy’s conclusions and stated reasons, rather than independently verified findings about the novel’s creation.

Plagiarism and AI use should not be collapsed into one claim. The plagiarism reporting concerns earlier work, while the AI dispute concerns this novel. Nor does the novel’s removal by the Goncourt mean every other prize has taken the same action. The Canadian Press reported on September 25 that the book remained in contention for other French literary prizes and that Orélien declined further comment while focusing on his health.

What happens next?

The Goncourt is expected to continue with its next selection on October 6. Readers should watch for any documented response from Orélien or his publishers and any further explanation from the academy about the analyses it relied on. As of this article’s September 26 update, the confirmed development is the novel’s removal from the Goncourt list; the public dispute over its authorship remains unresolved.

For readers, this case raises a difficult question beyond one prize: what kind of evidence should a literary institution require before treating an AI-writing allegation as settled? The book had already won a readers’ prize and found an audience. Neither its popularity nor a detector’s numerical output can, by itself, answer how it was written.

Follow us on Instagram and Facebook 

Follow us on Instagram and Facebook