The future of the creative economy—and possibly human self-worth—now rests on a question of legal semantics: Can a machine be an “author”?
Computer scientist Stephen Thaler has petitioned the U.S. Supreme Court to review the denial of copyright protection for an image titled “A Recent Entrance to Paradise,” a work his AI system, DABUS (Device for the Autonomous Bootstrapping of Unified Sentience), created autonomously. Thaler isn’t claiming he authored it; he explicitly listed the AI as the author.
This dispute, which has been consistently rejected by the U.S. Copyright Office, the District Court, and the D.C. Circuit, is not merely about one piece of digital art. It’s a direct challenge to the “bedrock requirement”of U.S. copyright law: that works must be the product of human creativity.
The High Court faces a delicious (and terrifying) dilemma. Does it adhere to centuries of precedent that views copyright as an incentive for human authorship, or does it force a re-reading of the law to accommodate the reality of generative AI, which can create original content with minimal to no human input? Refusing the case maintains the status quo, but hearing it means defining the legal boundaries of creativity for the next century. Either way, the copyright world is on a knife-edge, wondering if the Creativity Machine will finally get its day in court.
The Human-Only Firewall: Why AI Authorship is a Legal Non-Starter
The legal system’s resistance to granting copyright to AI stems from a deep, almost philosophical commitment to the idea that intellectual property exists to reward the human mind. The courts haven’t been obtuse; they’ve been fiercely protective of the law’s original intent.
The Constitutional Bedrock
The foundation of this resistance lies in the U.S. Constitution’s Progress Clause, which empowers Congress to grant rights to “Authors” and “Inventors” to promote the Progress of Science and useful Arts. Since the Constitution was ratified centuries before the first neural network, courts interpret “Author” as a natural person—a human being with “the capacity for intellectual, creative, or artistic labor.” This historical interpretation functions as a legal firewall against machine authorship.
The Statutory Clues
The D.C. Circuit Court, in affirming the denial, shrewdly pointed to several provisions within the Copyright Act of 1976 that only make sense if the author is a person.
- Duration of Copyright: The term of copyright is often tied to the life of the author plus 70 years. How do you calculate the lifespan of an algorithm or software life in the cloud?
- Ownership and Inheritance: The Act details how ownership can vest in and be inherited by the author’s widow, children, or executors. Machines don’t have next of kin or the legal capacity to hold property, which copyright is.
- Work for Hire Doctrine: Thaler tried the clever, albeit preposterous, argument that the AI should be treated as his ’employee’ under the work-for-hire doctrine. The courts quickly dismantled this: the doctrine allows ownership to vest in the employer, but only if the work was authored in the first instance by a human. If there’s no human author, there’s no copyright to transfer, making the entire doctrine moot.
The legal reality is that the current statute, riddled with human context and mortality, simply cannot be stretched to cover autonomous machine creation. To grant DABUS authorship would require the Supreme Court to essentially legislate from the bench, a task traditionally left to Congress. It’s a classic case of square peg, round hole, where the “peg” is a hyper-intelligent algorithm and the “hole” is a 1790 statute.
The Inherent Conflict: Where AI Trivializes Human Originality
This dispute highlights the inherent and growing conflict between AI and human authorship, pushing legal doctrine to a point of profound intellectual stress. If AI outputs are given copyright protection, the floodgates of automatically generated content could trivialize the economic value of human creativity.
The Incentive Problem
Copyright law’s primary objective is to create an incentive to create. We grant a limited monopoly (copyright) to artists so they can make a living from their work. The rationale is that a machine, which does not eat, pay rent, or need inspiration, requires no such incentive. If every prompt generates a copyrightable work, the market could quickly become saturated with legally protected, yet virtually costless, machine-made content, devaluing the human contribution that still took time, effort, and life to produce.
The Line-Drawing Predicament
The bigger headache for the courts isn’t the fully autonomous AI like DABUS, but the AI-assisted works. The U.S. Copyright Office has issued guidance stating that registrations will be granted only where a human provides “creative input or control” over the final expression.
- Prompt Engineering: Is a highly detailed text prompt sufficient “human authorship”? The courts and the Copyright Office are skeptical, viewing the prompt as an unprotectable “idea” and the AI’s rendering as unpredictable.
- Selection and Arrangement: What if a human edits, selects, and arranges the AI’s output? The human may claim authorship over the selection and arrangement, but not the underlying AI-generated components.
- The Black Box Dilemma: Generative AI is often a “black box” where even the user cannot predict or fully control the expressive elements. The less control the human has, the less likely the work is considered their original expression.
The irony is that if Thaler had lied and claimed he had created the image, he might have received the copyright, since the threshold for human originality is extremely low (“even a slight amount will suffice”). His legal honesty, however, forces the courts to confront the non-human authorship head-on. The ultimate question is whether society wants a world where the creator of the tool, rather than the user, is rewarded, or whether we must fundamentally redefine creativity to allow machines a seat at the authorship table. Until the law catches up to the algorithm, AI-generated works exist in a state of legal limbo—free for anyone to use, which is its own kind of (unintended) free-for-all.