Author’s Note:
At the intersection of technology and law, we often seek quick fixes for complex problems. Yet, when a new technology dismantles the fundamental notions upon which our laws are built, it demands more than simple adaptation. This article delves into such a disruption, arguing that the current mismatch between generative artificial intelligence and our copyright laws isn’t a minor adjustment issue, but a profound conceptual crisis.
To truly grasp the scale of this challenge, it’s essential to examine the very pillars of our legal language and axiomatic systems. History offers a crucial lesson: the most transformative legal shifts — from the abolition of human bondage (once a legally accepted “tradable good”) to the assertion of inherent liberties during America’s founding — were not mere legislative tweaks. They were the culmination of intense philosophical debates that, in their time, might have been dismissed as “navel-gazing.” Yet, it was precisely this deep, principled inquiry — fueled by the insights of great thinkers like Voltaire, Rousseau, and others who reshaped our understanding of liberty and justice — that paved the way for legal frameworks we now consider self-evident truths.
A pervasive sense of resignation often colors these debates: the implicit notion that “what’s good for Caligula (the powerful, the corporations) is good for the people,” leaving the “little man” with no real influence. This self-fulfilling prophecy — of non-intervention leading to continued voicelessness — perpetuates a dangerous apathy. Our own history, however, demonstrates that profound legal reform only emerges when we dare to question fundamental assumptions, precisely the kind of “navel-gazing” that led to the abolition of what was once considered a “tradable good”: human freedom.
For this reason, our analysis draws upon the ideas of foundational thinkers like Ludwig Wittgenstein and Kurt Gödel. Far from being academic adornments, their conceptual frameworks are, in our view, key to diagnosing the incompleteness and semantic collapse that generative AI has exposed in current legislation. Understanding the root of the problem is the indispensable first step towards forging lasting solutions. We aim not just to inform, but to invite a deeper reflection, recognizing that the most significant legal changes in history have stemmed from philosophical debates that dared to question the obvious. Those seeking a direct discussion of legal applications without this foundational philosophical context may find specific sections pertinent, but the central thesis of this work lies in the undeniable necessity of this conceptual analysis for the creators of our laws.
The dizzying pace of generative artificial intelligence has profoundly reshaped the creative landscape and our very concept of authorship. This disruption isn’t a simple evolution that existing legal frameworks can contain. On the contrary, it exposes a deep structural incompatibility between AI’s digital reality and traditional legal architecture. In this context, attempts to force the application of pre-existing laws, like copyright, are futile exercises in “stretching the rubber band” — tactics that, far from offering solutions, reveal the urgent need for entirely new legislation. The core issue isn’t just whether AI can infringe, but whether our current legal language can even describe how it operates, let alone regulate it effectively. This challenge is further compounded by the fact that this technology is now widely available to individual users and can run on consumer-grade hardware, posing unprecedented difficulties for traditional enforcement mechanisms.
The Outdated Legal Language: When “Isomorphism” Breaks Down (Wittgenstein)
Ludwig Wittgenstein’s philosophy teaches us that language is meaningful when there’s an isomorphism — a clear correspondence between words and the reality they describe. When this link shatters, language becomes meaningless. This is precisely the dilemma we face with generative AI and current intellectual property law, particularly its core concepts of “copy,” “originality,” and “authorship.”
Traditional copyright laws, born in an era of tangible reproduction, define “copies” as near-identical replications, “originality” by human creative choice in fixed forms, and “authorship” by this human act. However, AI doesn’t “copy” in the same way. It learns patterns and characteristics from vast datasets to then generate entirely new, synthetic content. For instance, a LoRA trained on Marilyn Monroe might evoke her essence, yet consistently miss her famous mole. This highlights the AI’s interpretive and transformative process, where features are synthesized from learned data, not directly reproduced.
Applying concepts like “copyright infringement” or “derivative works” — which rely on a direct, human creative lineage or a clear act of copying — to AI’s algorithmic generation leads to semantic collapse. Legal language simply can’t accurately describe the underlying technological reality. The fundamental question isn’t about whether an AI’s output looks similar to an existing work, but how it was produced. How do we apply authorship laws to an algorithmic process that isn’t a direct replication, but a creative synthesis? The answer is clear: we cannot do so coherently with our current legal vocabulary without distorting its meaning.
The Incompleteness of the System: The Need to Look From the Outside (Gödel)
This is where Kurt Gödel’s perspective becomes illuminating. His incompleteness theorems demonstrate that within a formal axiomatic system, there are truths that cannot be proven or disproven using only the axioms of that same system. We need an “outside” view to recognize its inherent limitations.
Applied to the legal sphere, the current system of intellectual property laws is “incomplete” for addressing generative AI. Judges, operating within this axiomatic system, find themselves at a dead end. Their role is to apply existing laws, not to reinvent them for technologies their drafters couldn’t imagine. The “dizziness” judges and juries feel in cases like Disney v. Midjourney exemplifies this incompleteness. While some lawsuits pinpoint “textbook” infringements under existing law (e.g., unauthorized IP use for advertising or direct reproduction of characters), understanding the fundamental challenge requires looking beyond these immediate applications. The deeper issue is how the legal system will definitively address the generative process itself — the creation of countless new works inspired by, but not direct copies of, training data. They are being asked to solve a problem the current legal system is fundamentally unequipped to handle. Dismissing these philosophical underpinnings as mere “statements” or irrelevant to concrete cases ignores how these very cases expose the underlying conceptual crisis of our legal language and systems.
The Solution is Political, Not Judicial: A Call for Legislative Action and the Lesson of Comparative Law
Legal history offers clear precedents: when a new technology breaks the isomorphism with the legal framework, the solution comes not from forced judicial interpretations, but from legislative action. We’ve seen in other legal systems how similar problems, such as “signal theft” or unforeseen uses of technologies, were resolved not through fruitless lawsuits based on old copyright laws, but by creating new, specific laws in parliament. These examples prove that the way forward is a law aligned with the new physical and technological reality, establishing clear legal boundaries where none existed before.
Generative AI represents a new category of creation and information use that currently lacks adequate legal representation in the existing code. This isn’t an issue judges can or should resolve. Jurists, with good judgment, remind us: “Don’t send political problems to the courts; first, enact a law.”
The solution to the challenges posed by generative AI regarding copyright and image rights is, therefore, fundamentally political. It is up to parliaments and legislators to acknowledge this incompleteness of the current legal system and assume their responsibility. They must design a new legal framework that understands algorithmic generation, establishes clear limits, fosters innovation, and protects rights fairly in this new digital era. While the path to such legislation will undoubtedly be contentious, often influenced by powerful lobbying interests (as seen with historical copyright extensions like the “Mickey Mouse Copyright Act”), the intellectual imperative for a coherent legal framework remains. Only then can we move from “senseless babble” to meaningful and effective regulation, ensuring the “small man” is not unjustly impacted by laws designed for a bygone era.