Introduction
The story of Stephen Thaler reads almost like a legal science-fiction saga—one where the courtroom becomes the battleground for defining creativity itself. Over the past few years, Thaler has transformed from a computer scientist into a controversial figure at the centre of one of the most pressing debates in intellectual property: can machines truly invent, and if they can, should the law recognize them as creators?
Thaler first drew global attention through his AI system DABUS, a neural network designed to generate patentable inventions. Together with his lawyer, Ryan Abbott, he launched an ambitious series of legal challenges across multiple jurisdictions, arguing that DABUS should be formally recognized as the inventor of its own creations. The courts, however, were not persuaded. Time and again, they returned to a fundamental point: patent laws were written with human inventors in mind, and nowhere did they contemplate non-human creators. As a result, even if an AI system might technically generate an invention, (almost) without any human input, it could not—at least under current legal frameworks—be named as its inventor.
Undeterred, Thaler brought the same line of reasoning into the realm of copyright. With another AI system, the so-called “Creativity Machine,” he sought recognition not for inventorship, but for authorship. His claim was bold yet consistent: if a machine independently produces a creative work, then that machine alone should be acknowledged as its author.
This blog focuses on this second chapter of the Thaler saga: the copyright battle. Beyond the legal dispute lies a deeper question: is Thaler a visionary ahead of his time, challenging outdated legal concepts? Or is he focusing on the wrong fight, while more immediate and complex issues loom over the relationship between AI and copyright?
After all, today’s most urgent debates are not only about whether AI can be an author, but whether training AI on copyrighted works constitutes infringement, whether AI-generated outputs unlawfully reproduce existing creations, and how creative industries can seek fairness for artists whose work has been absorbed into these systems without consent. In that context, Thaler’s crusade might seem both premature and oddly symbolic.
Perhaps, then, he is something of an intellectual property antihero—pushing boundaries that the law is not yet ready to cross. Or, borrowing from a familiar phrase, maybe Stephen Thaler is the hero we deserve… but not the one we need right now.
The Copyright Part of the Thaler Saga: to be or not to be an author?
If the patent battles around DABUS opened the door to the question of machine inventorship, the copyright disputes brought that debate into even more philosophically charged territory: authorship itself. Here, Stephen Thaler’s argument was not just about legal recognition, but about redefining creativity in an age of artificial intelligence.
At the heart of the dispute stood the position of the U.S. Copyright Office (USCO) Review Board, which firmly rejected Thaler’s claim. Its reasoning rested on the foundational principle according to which copyright law does not extend to nonhuman expression—meaning that works produced without human authorship fall outside the scope of protection.
Thaler, however, challenged this view on multiple fronts. He argued that the concept of originality—central to copyright law—has always been defined in relatively modest terms. A work need only be independently created and exhibit a minimal degree of creativity. Nowhere, he contended, does this definition explicitly require a human author. From this perspective, an AI-generated work could, in principle, satisfy the originality threshold.
Beyond doctrinal interpretation, Thaler also advocated for a purposive reading of the Copyright Act—one that evolves alongside technological progress. If the law aims to promote creativity and innovation, should it not adapt to new forms of creative agents? This line of reasoning led him to invoke a notorious benchmark in the relationship between machine and human: the Turing Test. For Thaler, the key question was not who—or what—created the work, but whether the output is indistinguishable from that of a human creator. If it is, why should copyright law treat it differently?
Recognizing the practical implications of ownership upon AI, considering that the latter cannot own things, Thaler further proposed the “work for hire” model as a solution. Under this framework, even if the AI were recognized as the author, the rights could still vest in a human or legal entity—much like how employers own works created by employees. This, in his view, offered a way to reconcile legal tradition with technological disruption.
The courts, however, remained unconvinced. In 2023, the District Court for the District of Columbia upheld the USCO’s decision, affirming that human authorship is a “bedrock requirement” of copyright law. Yet, notably, the court acknowledged that a broader policy question remains unresolved: how much human involvement is necessary for a work involving AI to qualify for protection? This unresolved issue hints at future legal challenges that may prove even more complex than Thaler’s own case.
Subsequent appeals did little to shift the legal landscape. In 2025, the Court of Appeals for the District of Columbia reached the same conclusion, reinforcing the centrality of human authorship through a textual analysis of the Copyright Act. Quite recently, March 2026, the U.S. Supreme Court declined to review the case by denying the petition for certiorari. Importantly, this procedural decision should not be read as an endorsement of the lower courts’ reasoning or as a definitive stance on AI authorship. Rather, it simply reflects the Court’s choice not to take up the case, leaving the existing rulings in place without resolving the broader legal question at the highest level.
What emerges from all this is a clear judicial consensus according to which AI cannot be considered author of a creative work, but not necessarily a final answer. While the (lower) courts have drawn a firm line excluding fully autonomous AI from authorship, they have also left open the more nuanced and pressing question: where exactly does human creativity end and machine contribution begin? That, perhaps, is the real legacy of the Thaler saga in copyright law.
Beyond Thaler: Human Involvement in AI-Assisted Works
The U.S. Copyright Office’s 2025 Report on Copyright and Artificial Intelligence (Part 2: Copyrightability) shows that the real issue is something other than the recognition of fully autonomous AI authorship: how much human involvement is enough when AI is used as an assisting tool for creativity.
The Report confirms a principle that echoes the reasoning seen in Thaler’s litigation: human contribution remains the cornerstone of copyright protection. Works generated entirely by AI—without meaningful human contribution—are not eligible for copyright.
Yet, crucially, the Office moves beyond this binary framing and develops a far more nuanced analysis. Rather than asking whether a work is “AI-generated” or “human-made,” the Report focuses on the type and degree of human contribution involved.
To make sense of this, the Office identifies several ways in which humans interact with AI in the creative process:
- AI as an assistive tool: where AI supports, but does not replace, human creativity. In such cases, copyright protection generally remains intact upon the human.
- Prompts: simply instructing an AI system, even though the prompts are detailed and/or iterative, is typically insufficient to establish authorship, given the unpredictable nature of AI outputs.
- Expressive inputs: where humans provide original material that is transformed or incorporated into the output.
- Selection, arrangement, or modification: where humans curate, edit, or meaningfully transform AI-generated content, potentially giving rise to protectable authorship. This is line with previous Copyright Office’s decisions such as the notorious Zarya of the Dawn.
Ultimately and explicitly, the Report frames copyright authorship as a case-by-case assessment, dependent on whether human contributions are sufficiently original and perceptible in the final work.
Interestingly, the Office does not call for immediate legislative reform. Instead, it concludes that existing copyright law is flexible enough to accommodate AI-assisted creation, provided that its bedrock—human authorship—is preserved.
Conclusion
Perhaps this is where the true legacy of the Thaler saga becomes visible. While Thaler’s claims sought to push the law toward recognizing AI as an autonomous author, the broader institutional response has taken a different path. Rather than expanding authorship to machines, policymakers are refining the boundaries of human creativity in an AI-assisted world.
After all, establishing a clear—ideally increasingly refined—distinction between what is human-created and what is generated by AI may serve as a useful starting point in any sector and field where AI has been introduced and discussed. For example, even on the abovementioned question of AI training and potential copyright infringement, the central concern is not simply the use of data, but the status of what remains fundamentally human—images, texts, and creative works that have been used to train AI systems. The real challenge lies in determining how this “humanness” can be preserved, protected, and properly acknowledged in outputs that may increasingly obscure their origins.
