Property Rights in the Age of AI

In the AI Story

The destabilization begins with the most fundamental question: who is the creator? When a human writes a novel, the property right is clear. When a human writes with an AI that generates prose, the property right is no longer clear. Edo Segal confronted this question directly in The Orange Pill and answered with transparency: the ideas were his, the collaboration was genuine, the authorship was a new form of creation. The answer is honest. It is not a legal framework. It is one individual's resolution of a question the existing copyright regime has not addressed.

The U.S. Copyright Office has been explicit that the requirement of human authorship excludes works generated entirely by AI. But the requirement provides no clear guidance for the vast middle ground of human-AI collaboration, where human contribution ranges from minimal (a brief prompt) to substantial (judgment, taste, editorial control shaping output into something neither contributor alone could produce). The ambiguity creates transaction costs. A publisher does not know whether the copyright is valid. An investor funding an AI-code startup does not know whether the code is protectable. Each uncertainty is a cost inhibiting the exchange that would otherwise occur.

The problem extends to training data. Large language models are trained on vast corpora produced by human authors who did not consent to use as training data and who receive no compensation. Whether this constitutes infringement or fair use is being adjudicated in multiple jurisdictions with no clear resolution. The uncertainty inhibits both AI development (facing potential liability) and creator compensation (no established mechanism for capturing value from training use).

The stakes extend beyond economics. Property rights in knowledge define who participates in the knowledge economy. The framework knitter whose skill was his property watched that property become worthless when the power loom replicated it. The software developer whose expertise is her property faces an analogous threat. The property right in expertise — the informal but economically real right to command a premium for scarce knowledge — is being eroded by a technology that makes knowledge production cheap and knowledge itself abundant.

Origin

North developed the centrality of property rights through his historical work on European economic development, particularly The Rise of the Western World (1973) and the 1989 paper with Barry Weingast on the Glorious Revolution. The framework drew on Coase, Alchian, and Demsetz but integrated property rights into a comprehensive theory of institutional economics.

The application to intellectual property in the AI era extends North's framework into a domain he did not live to analyze directly. The 2023 Thomson Reuters v. Ross Intelligence case, the ongoing New York Times v. OpenAI litigation, and the U.S. Copyright Office's 2023 guidance on AI-generated works represent the first generation of formal rules emerging from the institutional void.

Key Ideas

Clear rights minimize costs. When property is secure and transferable, transaction costs fall and exchange flourishes.

Authorship is radically destabilized. The copyright framework's requirement of human authorship provides no guidance for the vast middle ground of AI collaboration.

Training data is contested territory. Whether ingesting copyrighted material to train AI constitutes infringement or fair use is unresolved — and the resolution will determine value capture.

Expertise as property is eroding. The informal right to command premiums for scarce knowledge is being dissolved by technology that makes knowledge cheap and abundant.

Ambiguity is itself a cost. The uncertainty inhibits investment across the AI economy, producing the stagnation North identified in his studies of developing economies with unclear land rights.

Debates & Critiques

Major debates include: whether copyright should extend to AI-generated or AI-assisted works; whether training on copyrighted material should require licensing; whether new categories of property (such as a training-data royalty) should be created; and whether the entire framework of individual authorship should be reconceived for collaborative human-machine creation. The stakes are enormous: resolution will determine whether AI-era value flows to creators, to AI developers, or to some new institutional arrangement that does not yet exist.