The case of *C’était ça ou mourir* by Thélyson Orélien is fascinating in this regard. The novel, published in France by Grasset, won the 2026 Fnac Novel Prize and was a major success in bookshops. Then an anonymous account on X accused it of having been ‘almost entirely’ written by artificial intelligence, citing the Pangram detector in particular. The author denies having used AI, and Grasset firmly rejects the accusation. Above all, the reliability of AI detectors alone is not sufficient to turn an algorithmic suspicion into irrefutable proof.
We must therefore be precise: there is currently no evidence to establish as fact that Thélyson Orélien had his novel written by AI. But the controversy raises a far more interesting question: what if he had?
What exactly would that change about the book that readers have read, loved and recommended, and that booksellers have championed? If a text moves its reader, if its characters feel real, if its structure holds together, if a sentence lingers in the memory after the book has been closed, will we now have to ask for the technical specifications of its creation before deciding whether the emotion it evoked was legitimate?
Perhaps we are confusing the work with the tool.
A writer already works with a considerable array of technologies: computers, spell-checkers, digital dictionaries, search engines, word-processing software and document databases. No one asks a novelist to fell the tree themselves that will be used to make the paper for their book, in order to authenticate their status as an author. Generative AI obviously introduces a far more profound shift, as it intervenes directly in the production of language. But this shift does not necessarily eliminate the author. It shifts part of their work towards choice, direction, selection, rewriting and intention.
This is precisely where the debate becomes fascinating. Indeed, a writer, Sébastien Bailly, has just publicly defended the right to work with a machine, lamenting that the current debate reduces authors to two categories: the ‘innocent’ who has never touched AI, and the ‘cheat’ who has used it in secret.
Artificial intelligence is neither a divine muse nor a magical brain. Current systems are based on statistical models trained on vast quantities of data and capable of generating language based on the patterns they have learnt. This raises real questions: copyright, remuneration for creators whose works were used for training, transparency, the potential standardisation of writing styles, and the responsibility of the person who signs the work. But none of these issues will be resolved by shouting ‘cheating!’ every time a writer opens a dialogue box with a machine.
Cultural history is full of tools that were initially regarded as threats. Photography was supposed to kill painting. Cinema was supposed to kill theatre. Television was supposed to kill cinema. Digital technology was supposed to kill the book. Nothing has died exactly as predicted. Practices have shifted, merged and transformed.
AI will probably do the same, but at a far more rapid pace.
And literature may well be nothing more than a miniature laboratory for a far more sweeping upheaval. For if we gradually come to accept that a machine can contribute to the writing of a novel, we will inevitably end up asking it far more unsettling questions: why shouldn’t it play a greater role in the management of our societies?
Today, this idea remains in the realm of speculation rather than a desirable or inevitable political reality. But it is an intellectual experiment worth undertaking. Let us imagine systems capable of instantly analysing millions of pieces of economic, health, climate or demographic data and simulating the consequences of a decision before it is implemented. Faced with such systems, what would become of our endless election campaigns, our talking points and certain decisions taken more on the basis of the next opinion poll than the next twenty years?
This would not, however, mean that AI had objectively discovered ‘the happiness of the people’. A machine can optimise an objective; it cannot, on its own, determine which objective a society should choose. Should we prioritise freedom or security? Equality or the reward for effort? The present or future generations? No equation can make these value choices disappear. And whoever controls that famous slider already wields a considerable amount of power.
Perhaps this is where the true revolution of the future lies: not necessarily AI taking power, but AI making it increasingly difficult to justify inefficiency, lies or certain decisions taken contrary to the available facts.
It will not automatically put an end to wars, famines, corruption or propaganda. Technology can also be used to monitor, manipulate and concentrate power. It all depends on those who possess it, the rules imposed upon it and the objectives set for it.
But it could render obsolete a host of tasks we thought were the preserve of human intelligence. And many professions will then have to face a rather unpleasant question: what exactly is their added value when a machine can carry out in a matter of seconds a part of the work that previously took several hours or several days?
This is why the case of *C’était ça ou mourir* goes far beyond the case of Thélyson Orélien.
We may be living through one of the last moments when the use of artificial intelligence can still be presented as a shameful anomaly that needs to be unmasked in a work of art. In a few years’ time, the question will probably no longer be: ‘Have you used AI?’
It will be: “What did you do with it?”
And that is infinitely more interesting.
