An Open Letter to the Académie Goncourt
Singapore, September 2026
To the Académie Goncourt and everyone following what happened to this year's longlist:
On September 21, an anonymous account ran passages of Thélyson Orélien's novel C'était ça ou mourir through an AI detector and declared it mostly machine-generated. Four days later, you removed the book from your first selection. Your statement gave two reasons: plagiarism by the author, recently brought to light, and several analyses concluding the novel was very probably produced largely by AI. Orélien denies using AI. And one French publisher is now launching a “100% human” label.
Let me be clear about one thing first. Plagiarism is taking another writer's work. It is a separate matter, it has nothing to do with AI, and I am not defending it.
What concerns me is the principle in your statement: that you do not endorse or sanction, in any way whatsoever, “the creation of texts generated with the help of AI.” Those words reach far beyond one novel. They reach every writer who uses these tools honestly and says so. Including me.
Your president has said a charter is needed to tell human work from machine work. I agree. But I'm writing because I want you to ask the right question.
I wrote Fifty-Five Minutes at No. 5, which just came out on September 23, 2026, marking the twenty-fifth anniversary of the Brookwood mine disaster. I used AI tools to write it. I stated that clearly in the front matter before anyone had the chance to “catch” me. I am not stepping into this debate to defend another writer. I am stepping in because I have already navigated the exact dilemma you are trying to settle.
Right now, everyone is asking: Was AI involved?
It is the wrong question because you cannot reliably answer it. Detection tools spit out probability scores, not proof. Ruining a writer's reputation over an arbitrary percentage is absurd. But even if you could prove it, the answer tells you almost nothing about the actual book. Typing a single prompt to churn out a novel and spending months agonizing over a manuscript while using an algorithm to organize, test, and draft are both technically “AI-involved.” They are not the same thing. No reader on earth thinks they are.
The question that actually matters is older than any machine: whose work is it?
Let me tell you how my book was made. I came to the Brookwood story by happenstance while working on mine-safety records, and I stayed because thirteen men had been forgotten by the country. The official record ran to about 375 pages: a federal investigation report and years of legal decisions. I fed all of it to several AI systems. They summarized it, indexed it, and pulled out every date and every name. Not one of them told me there was a book in there. That call was mine.
The machines brought the intelligence. They could reason across every page faster than any person alive. What they could not bring was the soul: the reason to care, the conviction that thirteen men deserved better than to be forgotten. That is what separates us from them. And that is the question your charter should ask. Not “Did the author use AI?” but “Did the author provide the soul?”
Everything that followed was mine as well. I decided what the book was for. I built its structure. I set its rules: stay faithful to the record, no invented dialogue, no invented thoughts, and silence wherever the record is silent. Then I used AI tools to organize the facts and draft prose, following my outline, under my editing. Every fact was checked against the record. Every sentence is my responsibility. Without that help, the story would not have been told by the anniversary. It might not have been told at all.
That is what authorship looks like when a writer uses a powerful tool. How do you know whether the author provided the soul? Ask four simple questions of any book. Who conceived it? Who shaped its structure and its meaning? Who set its limits? Who verified it, and who answers for every sentence? If the answer is a human being, it's that person's book, whatever tools they used. If the answer is a prompt, it isn't.
Build a rule around detection instead, and you do damage both ways. You put honest writers under suspicion because a machine guessed, and then you ask them to prove a negative. And you teach every writer who does use these tools that the smart move is to keep quiet. A ban won't stop machine-made books. It will stop disclosure. The only authors it punishes are the ones who tell the truth.
You have made your decision about one book, and the plagiarism alone may have justified it. But the principle you announced will outlast this case. No writer should have to defend a reputation against a detector's score. And a writer who uses these tools openly deserves to be judged on one thing: is the book truly theirs?
You have already taken a position. I am asking you to take the right one. Draw the line where literature has always drawn it: at the author, and the soul the author brings. Ask writers to tell you how they made their work. Then judge the work.
I am publishing this letter openly because this debate needs more voices than a jury room can hold. Agree or disagree, write to me at admin@55at5.com. I will read every reply, and with the writer's permission I will publish the thoughtful ones here, including those that disagree with me.
Respectfully,
Tejesh Srivastav