The Ghost in the Manuscript: The Growing Crisis Surrounding Thélyson Orélien

The literary world is currently gripped by a high-stakes investigation into the nature of authorship in the age of artificial intelligence. Thélyson Orélien, a Haitian-Canadian author, finds his latest novel, C’était ça ou Mourir (It Was Either That or Die), at the center of a firestorm that threatens not only his reputation but the very legal foundation of his work’s existence.

After weeks of strategic silence, Orélien and his publisher, Éditions du Boréal, have broken their cover to address mounting allegations of AI-assisted writing and historical plagiarism. However, rather than providing the clarity needed to extinguish the controversy, these disclosures have arguably deepened the mystery, leaving the publishing industry to grapple with a profound question: how do you prove the humanity of a creative work when the evidence of its gestation has vanished?

A Chronology of Controversy

The scandal first erupted three weeks ago when C’était ça ou Mourir was abruptly pulled from consideration for the prestigious 2026 Prix du Roman Fnac. The catalyst was a viral claim made by a pseudonymous X (formerly Twitter) account, which provided screenshots of a report from the AI-detection software Pangram. The report alleged that the novel was 100% AI-generated.

The situation grew more complex when the pseudonymous account was identified as belonging to Samuel Fitoussi, a columnist for the conservative French publication Le Figaro. This revelation introduced a layer of political skepticism; critics of the accusation suggested the move might be a performative strike against contemporary literary trends. Fitoussi has vehemently denied any political motivation, insisting his concerns were purely rooted in the integrity of the literary process.

Following the initial AI allegations, the scrutiny intensified. Skeptics began digging into Orélien’s back catalog, uncovering instances of plagiarism in a short story published in 2012 and various articles written for Radio Canada. As the walls closed in, Orélien retreated from public view, canceling his book tour and opting for a period of absolute silence. That silence was finally broken this week, but the resulting statements have provided little comfort to those seeking definitive answers.

The Response: Admissions and Ambiguities

In a detailed letter submitted to Reuters, Orélien addressed the two-pronged attack on his career. Regarding the AI allegations, he remained steadfast in his denial, insisting that C’était ça ou Mourir is entirely a product of human labor. However, he pivoted to address the older allegations of plagiarism, characterizing them as "unacceptable mistakes" committed when he was "just over 20" and lacked the maturity to understand the gravity of his actions.

"I deeply regret my past mistakes," Orélien wrote, offering "heartfelt apologies without reservation."

Simultaneously, his publisher, Éditions du Boréal, issued a statement of support. They acknowledged the gravity of the situation but admitted a critical failure in their editorial process: they could not verify the authenticity of the novel before the manuscript was submitted on September 12, 2025. This admission is particularly damaging given Orélien’s own defense. The author claims he began the novel in 2017—long before the rise of public-facing generative AI—and completed his first draft in 2019.

The gap between these dates creates a "Schrödinger’s Novel" scenario. If the book was written between 2017 and 2019, it is an impossible candidate for AI assistance. Yet, the publisher admits they have no access to drafts, cloud-synced files, or email exchanges that would prove the manuscript’s existence prior to the late-2022 explosion of tools like ChatGPT.

The Evidence Gap and the Credibility Crisis

The core of the issue lies in the lack of digital provenance. In the modern era, an author’s "paper trail" is digital. A simple time-stamped email, a version history in a word processor, or a cloud-storage upload would act as an unassailable alibi. The fact that such evidence is purportedly non-existent—or at least unavailable—places Orélien in a precarious position.

His credibility, already frayed by the admission of past plagiarism, is now the only currency he has left to spend. This creates a binary choice for the public:

1. The Serial Offender Theory: Orélien is a repeat offender who transitioned from traditional plagiarism in his youth to the more sophisticated, harder-to-detect "plagiarism by algorithm" in his current work.
2. The Unfair Target Theory: Orélien is a writer who made youthful errors, learned from them, and is now being victimized by an overzealous, politically-charged crusade that uses unreliable AI detection tools to ruin a career.

The reliance on AI detection software like Pangram is itself a point of contention. While these tools have improved, they are not forensic instruments. They are probabilistic engines that look for patterns—patterns that can be triggered by specific stylistic choices or even non-native linguistic structures. To base the cancellation of a literary prize on a single, non-definitive AI report is a dangerous precedent, but to ignore the total lack of historical evidence for the manuscript’s creation is equally negligent.

The Legal and Financial Stakes: Copyright in Jeopardy

Beyond the reputation of the author and the prestige of the publisher, there is a looming legal crisis. If a book is found to have been written by AI, its status under intellectual property law changes instantly.

The United States Copyright Office has been consistent in its stance: copyright protection is reserved for works of human authorship. The precedent was famously established in the "monkey selfie" case and has been reaffirmed in multiple rulings regarding generative AI. If it is proven that C’était ça ou Mourir was generated by an AI, the work could effectively fall into the public domain.

For Éditions du Boréal, this is a nightmare scenario. They have invested in the editing, printing, and marketing of a work that they might not actually "own" in the eyes of the law. If the book lacks copyright protection, the publisher loses the exclusive right to sell or distribute the novel, as any entity could theoretically print and sell their own copies without penalty.

This financial instability explains why the publisher is standing by the author despite the lack of evidence. They are locked in a "sunk cost" scenario where they must either double down on the author’s legitimacy or risk admitting that their flagship product has no legal protection.

Conclusion: A Lingering Limbo

The attempt by Orélien and his publisher to clear the air has, in practice, only added fuel to the fire. By admitting to past plagiarism, Orélien has invited the public to view his current work through the lens of dishonesty. By admitting that they cannot verify the timeline of the manuscript, the publisher has essentially signaled that they have no evidence to refute the claims of AI usage.

We are left with a situation where the truth is obscured. There is no smoking gun, yet the absence of evidence is itself a form of testimony. For the literary community, this case serves as a warning. As generative AI becomes more sophisticated, the burden of proof for authors—and the responsibility of verification for publishers—must become more rigorous.

For now, C’était ça ou Mourir remains in limbo. It is a book that occupies a strange space between legitimate creative labor and potentially hollow digital fabrication. Whether it will be remembered as a misunderstood masterpiece or the first major casualty of the AI era remains to be seen. What is certain, however, is that the industry’s current "trust-based" model of manuscript submission is no longer sufficient. In a world where machines can mimic the human voice with chilling precision, the only currency that matters is the ability to prove one’s own work. Orélien has failed to provide that proof, and in the high-stakes world of literature, that may be a debt he can never fully repay.