In the midst of an insurrection, the indomitable French have found the time to take a step towards reconciling the global intellectual property regime in culture with the new challenges presented by AI. The world’s 2026 publishing calendar is pockmarked by the withdrawal of books from literary prizes, and even from publication, due to moral anxieties about the use of AI in literature. Now, weeks after France’s Prix Goncourt eliminated Thélyson Orélien’s C’était ça ou mourir, a new prize for books written with AI has been announced. French literature is forking―like on Github―between machine-assisted and purely human authorship.
Prix Pascal will consider French manuscripts of 10,000-50,000 words accompanied by a seed prompt of up to 100 words, and details of LLMs, agents and prompts used. Submissions close end-November and the winner will be announced in early December. A fast verdict suggests that these AI-generated manuscripts will be evaluated with AI assistance, too. It is perhaps fitting when CVs are being tuned using AIs to apply for jobs which, in turn, screen applications using AI.
An AI-inclusive prize is an intelligent response to a year of controversies about AI-assisted texts, which started with Hachette dropping Mia Ballard’s debut novel Shy Girl in March. Jerry Falade’s Call Me, I’ll Hide the Body was another noteworthy casualty. There were many others, and the phenomenon has shaken writers, some of whom are tweaking their styles―self-censoring to visibly demonstrate that they are not using AI.
They protest too much, pointlessly. Because, what does an AI do? It tries to pass the Turing test. It mimics humans it knows. It ploughs through terabytes of human culture to absorb vocabularies, speech patterns and behaviours which it can use to masquerade most convincingly as a human. Theoretically, this suggests that AI-detecting tools are detecting only AIs which can’t pretend too well. Human-observable signs attributed to AI input, like a lot of em-dashes, are not reliable, either. Thousands of journalists, academics and writers use them for emphasis, from mere editorialists to Nobel laureates, but this worthless notion has preoccupied the industry for months. But why does the literary world care so much about AI assistance? Because it cares about rights.
The literary establishment has always believed that it is best to write in your own authorial voice―which is as distinctive as your audible voice. And thereby, the author came to be regarded as the authentic repository of all rights relating to their work. Classical copyright notices used to include this statement: “Jane Doe asserts the right to be identified as the author of this work”. Now, it’s become clipped: “Copyright © 2026 by Aruni Kashyap” (from the new novel How to Date a Fanatic). In Arundhati Roy’s Mother Mary Comes to Me, it is terse but speaks volumes: “The moral right of the author has been asserted.” The statement implicitly divides rights into moral and commercial. The latter can be freely sold, given away, lent or inherited. The former are unalienable.
This neat picture is muddied in fresh editions of texts too old to be copyrighted. A new edition of Odyssey may assign translators and researchers the title of author. Homer’s name may not even appear in the copyright notice, though there would be no Odyssey without him. Now, imagine a futuristic, Star Trek-like fanfic of the Odyssey, in which Homer is still excluded. His place is taken by a worldbuilding AI.
It’s obviously a co-author, but its contribution draws on the myriad writers whose words the machine has swallowed, from the contemporary Kilgore Trout (a work of fiction himself, created by fiction writer Kurt Vonnegut), all the way up the river of time to Homer. The nonhuman co-author contains human multitudes who have not consented to be corralled in there, which breaks the assumptions of the copyright regime about originality. How can a publisher copyright a new work whose AI-supplied building blocks are already under multiple copyrights? Therein lies the uneasiness about AIs in creative work.
The machines are here to stay. Despite industry protestations that they are not replacing people, they will. So, it would be prudent to acknowledge AI-assisted creativity through novel forms of licensing. Creative Commons and similar methods have served many industries well―publishing (Cory Doctorow), music (Nine Inch Nails), software (Blender). And foremost, the children of UNIX operating systems―macOS, Linux, Solaris, BSD. Raspberry Pi and Android are children of Linux. These UNIX-inspired systems incorporate the work of thousands of programmers and maintainers across the world, across generations―with their licence.
Similarly, licensing could be amended to acknowledge machine collaborators in creative work. While AI gets bad press for stealing intellectual property and jobs, it can also give voice to those rendered voiceless by illiteracy, impaired health, or just the inability to give mainstream articulation to their imagination. It’s like Uncle Otto Schlemmelmayer’s flute in Isaac Asimov’s short story ‘Button, Button’ (collected in Buy Jupiter and Other Stories), which anyone could play: “No skill necessary―only thought.” Amid the multiple dystopias which AI threatens, there could also be this liberating utopia where only the human imagination matters.
Pratik Kanjilal | SPEAKEASY | For years, the author has been speaking easy to a surprisingly tolerant public
(Views are personal)
