The Muse Never Signs

copyright, artificial intelligence and psychoanalysis*

Nearly a hundred lawsuits are currently making their way through American courts against companies such as OpenAI, Meta, and Anthropic. Artists, film studios, and newspapers claim that their work was used without permission to train artificial intelligence models. Yet, in a recent essay for The Atlantic, legal scholars Jacob Noti-Victor and Xiyin Tang argue that this may not be the decisive battle. The question that will reshape the future of creative work points in the opposite direction: to what extent can a work generated by artificial intelligence itself be protected by copyright?

In 2025, in Thaler v. Perlmutter, the U.S. Court of Appeals for the District of Columbia Circuit held that a work generated autonomously by an AI cannot be copyrighted. The law requires a “human author.” The U.S. Supreme Court declined to review the decision in March of this year. The underlying question, however, remains unanswered: what quantum of human participation is enough for a work created with the assistance of AI to count as an original work of authorship? A prompt? Curation? Editing? The U.S. Copyright Office has already indicated that prompts alone are not enough. But what about prompts and editing? Extensive editing? How much? It is an invisible, arbitrary line — and courts and legislatures will spend the coming years trying to draw it.

The question seems technical. It is not. It touches a point in creation that runs through the history of human inventiveness.

“To love is to give what one does not have,” Jacques Lacan wrote. The lover offers the moon, the stars — creating, like the poet, out of a lack. For psychoanalysis, this lack, constitutive of the human, is the source of creation. We have covered it with many names — Muse, Holy Spirit, the unconscious, daimon — but they all point toward the same experience: that, in the moment of composition, something passes through the ego. From Taylor Swift to Rilke, artists have spoken of this. Anyone who has ever created anything, even a dream, knows what this means.

If creation always comes from a lack, from something that passes through the ego — if, at the moment of composition, the artist is more medium than source — does it ultimately matter whether the channel is human or machine? When inspiration descended upon the Delphic Sibyl, no one asked whether the oracle had exercised sufficient “human participation” to qualify as the author of the prophecy. The Muse never held a copyright. Why, then, such anxiety over where the breath comes from?

The way out of this impasse lies in a displacement Lacanian psychoanalysis had already made, for other reasons. The psychoanalyst Jorge Forbes argues that the twentieth century mistook the unconscious for an excuse — “my unconscious made me do it” — when it is, on the contrary, the very source of responsibility. Freud himself, in his 1925 essay “Some Additional Notes on Dream-Interpretation as a Whole” — particularly the section entitled “Moral Responsibility for the Content of Dreams” — was already saying that we are, of course, responsible for the evil impulses that appear in our dreams; what else could we do with them? Lacan went further: by virtue of our position as subjects, we are always responsible. Including, Forbes adds, for chance and surprise.

The end of an analysis is the moment when the subject can no longer appeal to the unconscious as a guarantee. It is not that one stops encountering the strange in oneself. Rather, one no longer invokes it as an excuse to evade responsibility. Forbes gives this point a name: enthusiasm — not as an effect but as a cause. It is the ethical position of someone who assumes authorship of what they invent, without any guarantee.

Applied to artificial intelligence, this reversal can shed light on the problem. The criterion of authorship has never really resided in the origin of inspiration — for that origin has always escaped the ego, always come from somewhere “outside.” The criterion lies, and has always lain, in the response. As Forbes puts it, the artist performs a threefold movement: sees something unique, takes responsibility for what they have seen, and publishes it — puts it into the world. Coleridge’s “Kubla Khan,” Almodóvar’s women: it scarcely matters from which Muse they descended. What matters is that someone signed their name to them, implicated themselves, bore them as a point of honor — and of shame.

The Muse never signs. Neither does the oracle. The one who signs is the one who carries the prophecy out of the temple and into the city that disputes its meaning.

From this, a workable legal criterion can be drawn for this urgent question. The “human participation” required for a work to qualify as authorship cannot be measured in keystrokes, hours of editing, or percentages of retouched pixels. It is measured, rather, by the possibility of response. Is there a body that implicates itself in this work to the point of answering for it before the critic, the judge, and time? Is there someone for whom this work is a point of honor, and whose refusal to stand behind it would be a matter of shame? Is there a subject who, when asked, “Is this yours?”, can answer, “Yes” — not as a claim of ownership, but as an ethical assumption of responsibility for what they have invented?

A machine can produce what looks like a sonnet. It can even produce what looks like a surprise. What it cannot do is feel shame for what it has invented, defend its creation, or sustain it over time. The machine has no body. It is not, in Lacan’s sense, a parlêtre — a speaking being capable of being touched by the letter it has written. For psychoanalysis, authorship is not a question of input. It is a question of who is willing to carry what came into the world in the process of creation.

This, then, is what psychoanalysis has to offer the law as it confronts artificial intelligence: do not ask where the work came from. Ask who answers for it.

Felipe K. Massaro, Ph.D., is a clinical psychologist and psychoanalyst based in Brazil. He is a member of the Lacanian Psychoanalysis Institute (IPLA), affiliated with the World Association of Psychoanalysis (WAP).

*First published in Portuguese at Mundo Visto pela Psicanálise (Instituto da Psicanálise Lacaniana): https://ipla.com.br/conteudos/artigos/a-musa-nao-assina/