Created by No One: Generative AI and the Unravelling of Authorship

The rise of generative artificial intelligence is profoundly disrupting one of the oldest pillars of French intellectual property law: the figure of the author. Long conceived as inseparable from the human person, the creative work now finds itself challenged by productions generated by algorithms capable of composing, writing, and producing music, images, text, and video with no apparent human involvement. A question, at once legal and philosophical, inevitably arises: are we entering an era of authorless works? Should we be concerned? And above all, how does copyright law approach these creations?

Roald Dahl’s Eerily Prescient Ethical Question

A short story published by Roald Dahl in 1953, The Great Automatic Grammatizator, offers a reading of the present moment that feels remarkably prescient. It tells the story of a machine capable of producing bestselling texts, pushing human writers to the margins and ultimately compelling them to sign away their rights, until the machine’s owner becomes the near-exclusive publisher of successful literature.

The story ends on the tragic image of an impoverished author, tempted to sign a contract assigning his rights to the operator of the great machine, who steels himself to resist: “This very moment, as I sit here listening to the howling of my nine starving children in the other room, I can feel my own hand creeping closer and closer to that golden contract that lies over on the other side of the desk. Give us strength, oh Lord, to let our children starve.”

This story anticipates several contemporary developments: the standardisation of creative works, the economic concentration driven by technology, and the gradual erasure of the author as a creative subject.

The stakes, then, are not merely legal: they are cultural, economic, and ethical. Can we conceive of a world in which creation is predominantly generated by automated systems? Or will human imperfection, with its idiosyncrasies, ruptures, and risk-taking, remain at the heart of what it means to create?

As a Matter of Law, There Is No Work Without an Author

Under French law, the answer is, at present, relatively clear. Article L. 111-1 of the Intellectual Property Code enshrines a fundamental principle: only a natural person may be recognised as an author.

This requirement is closely tied to the concept of originality, understood as the imprint of the author’s personality. An artificial intelligence, however, possesses neither legal personality nor subjectivity. It cannot, therefore, under current positive law, be recognised as an author.

The consequence is stark: a production obtained without meaningful human creative input is, in principle, excluded from the scope of copyright protection.

This position is widely shared internationally. In the United States, courts have recently confirmed that the absence of a human author precludes protection under copyright law. And in March 2026, the US Supreme Court declined to disturb this approach, consolidating the requirement of human authorship.

In the current state of the law, a work without a human author is therefore not legally protected — even if, as discussed below, it may still be monetised.

The question then shifts to works whose creation has been assisted by AI. In this context, AI may be treated as a creative tool — much like the camera for a photographer, notation software for a composer, or, in a less automated example, the brush and palette for a painter.

From this perspective, a work produced with the assistance of generative AI could be eligible for protection, provided that an identifiable human creative contribution can be demonstrated, whether through the reuse or integration of AI-generated output into a broader work, or through meaningful modification of that output by creative means or through further prompting.

Recent US case law illustrates this logic. A work produced partly through generative AI was found to be protectable where a substantial human contribution was present — in that instance, significant retouching and compositional work.

Conversely, prompts, however elaborate, are generally regarded as mere instructions, insufficient to constitute a creative contribution.

In other words, the law protects a mastered expression arising from the creative work of a human being, not a mere intention.

AI-Generated Works in the Style of Classic Authors Are Doubly Unprotected

The absence of protection for AI-generated works is even more apparent where the creative project involves imitating the style of authors whose works have entered the public domain. Such works also raise distinct questions about the place of the authors they seek to emulate.

The collective Molière Ex Machina staged, in April 2026, a play by Molière that Molière never wrote, produced in collaboration with Mistral AI, text, costumes, and music included. The collective’s guiding question was: what might Molière have written had he lived a little longer? The project was rigorously supervised, with human beings intervening at various stages of the creative process: historical analysis informing the prompts, selection of works on which the generative AI tool was trained, human requests to modify outputs, adjustments made during rehearsals and in response to the actors’ interpretations, and so forth.

The rapid advances in AI tools could make it even easier and faster, in the future, to generate works in the style of classical authors, on demand, and without limit. It is not inconceivable that such works could eventually reach an aesthetic quality on a par with the writers, composers, photographers, or filmmakers whose styles they replicate.

Who has not dreamed, as the members of Molière Ex Machina did, that their favourite author or composer had lived long enough to produce more? Which music lover has never longed to hear a tenth symphony by Beethoven, an undiscovered opera by Monteverdi, or a piano sonata by Brahms? Which reader has never wished for dozens of new novels by a favourite author who died too young to write them? Which cinephile has never imagined films that Chaplin, Kubrick, or Hitchcock never had the time or the idea to make?

Copyright law is indifferent to the quality or public success of a work. An AI-generated creation might rival human works, move audiences, and reach the level of past or present human genius, and yet remain unprotected, neither to the benefit of the public-domain authors who inspired it, nor to that of its contemporary designers who lack the status of authors.

Are We About to Be Overwhelmed by Authorless Works?

Over 40 % of new music tracks uploaded to Spotify or Deezer in early 2026 (around 75,000 tracks per day, a figure that continues to rise) are produced by generative AI tools such as SUNO. They account for just 1 to 3 % of streams so far, though that still represents several million monetised plays, to the detriment of real artists and producers.

One candle maker with no background in music production used AI to create a song that garnered 1.6 million streams.

Given that these works are not protectable by copyright, one may ask whether the developers and distributors who supply them to platforms are genuinely entitled to receive remuneration for them. The royalties paid by platforms include a share directed to distributors, but also to phonographic producers and performers in exchange for the authorisation by those rights holders to exploit their economic rights.

Should the monetisation of AI-produced albums not be set at a lower level than that of other albums, or reduced to zero? It should be noted that these authorless albums (which invariably draw on well-defined musical styles, reproducing their codes: R&B, country, Japanese jazz, children’s music, etc.) are produced by training AI tools on pre-existing works created by the very rights holders who are being short-changed.

Should the rights holders whose albums were used to train SUNO and similar AI tools not be entitled to choose between: (1) an opt-out mechanism allowing them to prohibit access to their works for AI training purposes; and (2) additional remuneration drawn from the royalties paid to distributors of AI-generated albums produced through such training? That would, of course, presuppose the ability to reliably identify any album produced entirely by AI, which is not yet possible.

Musicians, for their part, appear relatively unconcerned about AI’s limitations. Many believe it will push artists to surpass themselves, while benefiting from the technical assistance it can provide. This is perhaps because AI-generated music tracks represent the ultimate form of “commercial” music: a standardised output, whose only creative input is borrowed from pre-existing works. Far from being “original,” these are better described as “new” creations, their lyrics and melodies did not previously exist, but they bear no authentic creative imprint.

Without authors engaged in genuinely human creative activity, AI itself could not renew itself, lacking the new original works on which it needs to be trained.

Do We Still Need Authors?

At this stage, the answer remains yes — on several levels.

Technically, first: AI has not yet demonstrated the capacity to produce works that would become defining creations in the history of literature, music, or cinema. Its mode of operation is one of training and machine learning. AI does not appear — not yet — capable of producing original works that are truly unheard-of, in the etymological sense. Cynical as it may sound, one might go so far as to say that generative AI still needs authors — new authors — to continue attempting to replace them, in an endless race towards infinity.

Economically, generative AI is already beginning to supplant certain standardised forms of creation: background music, generic illustrations, low-value editorial content for which, arguably, authors are dispensable. But it also competes — rather unfairly — with mainstream works, as seen above in the case of streaming platform albums. The need for authors is undoubtedly less pressing for works of modest ambition than for those expected to define their era and endure.

It is on the artistic level that the question becomes most delicate. While entirely AI-generated influencers already exist and are knowingly followed by their audiences, the prospect of autonomous artistic corpora might resolve the dilemma of separating the author from the work. One could guiltlessly enjoy authorless novels and set aside the ethical debates surrounding the blacklisting of Wagner the antisemite, Céline the collaborator, or Gauguin the colonialist who slept with thirteen-year-old Tahitian girls. The author is sometimes a liability for admirers of their work.

But this vision of an art without authors is deliberately provocative and extreme. It seems unlikely that humanity’s appetite for creativity, innovation, and authenticity will ever be fully satisfied.

The position of the performer, finally, does not appear to be genuinely threatened. The “two-beat arts,” as Henri Gouhier called them — theatre, dance, and music — emerge so far unscathed from the expansion of automated production. Hearing a tenth Beethoven symphony produced by generative AI for the first time on Spotify is, admittedly, an experience. Hearing it performed by an orchestra in a concert hall is quite another.

Jérémie Leroy-Ringuet, April 2026