Papers & Research

The chapter from which this site takes its name. Argues that the "progress" IP law is meant to promote must mean human progress — human self-fulfillment through creative and inventive activity — and proposes a causation-based test for distinguishing works and inventions with a genuine human cause from those without. Engages copyright, patent law, and the philosophical foundations of IP across common law and civil law traditions, drawing on Plato, Aristotle, and Spinoza's conatus to ground the normative claim that human self-realization through art and science is progress, while mere technological change is not.

Argues that the advent of generative AI requires IP law to confront, for the first time, the question of "artificial" intellectual property — works and inventions produced by machines without a human proximate cause. Using the causation framework developed in tort law, the article proposes distinguishing first-degree IP (human proximate cause) from second-degree IP (machine output), and argues against extending protection to the latter absent a demonstrated showing of societal benefit. Machines are not authors, period; and the "if value, then right" rationale is firmly rejected as normatively empty.

Introduces the concept of "second-degree" intellectual property: works and inventions produced by AI machines that are themselves built, programmed, and trained by humans, but where there is no human proximate cause for the specific output. Drawing on the causation framework from tort law, the article distinguishes simple "but for" causation — insufficient for IP protection — from proximate causation, which requires a direct causal link between a human creative or inventive act and the output. Argues that any extension of IP protection to second-degree outputs should require a systematic showing that it would advance human progress.

Examines what kind of human engagement with AI-generated text rises to the level of authorship for copyright purposes. Argues that editing must constitute genuine creative collaboration — not mere correction or selection — to satisfy the human cause requirement. Develops a three-tier framework comparing US and EU approaches.

Eighteen national reports submitted to ALAI 2026 reveal a striking convergence: across common law and civil law traditions, every jurisdiction surveyed holds that purely machine-generated output is not protected by copyright, and that the human-authorship requirement is a qualitative, case-by-case inquiry not admitting of numerical thresholds. The paper synthesises that convergence, analyses two recent German decisions (the Amtsgericht München's February 2026 ruling on AI-generated logos and the Landgericht Frankfurt's December 2025 judgment on burden of proof), and addresses the tension between Painer and the Court of Justice's December 2025 ruling in Mio and Others. From that comparative foundation, the paper proposes a "collaborative writing standard" for LLM-edited texts, built on three diagnostic criteria: structural displacement (whether the human has reorganised the work's architecture rather than merely improved its surface), voice (whether the final work's distinctive character is attributable to the human's choices), and expressive specificity (whether the work contains formulations distinctively the human's). The standard's corollary, that much skilled but non-transformative LLM editing falls into an unprotected middle, is identified as a legislative problem, not a judicial one.

Examines the copyright issues at the heart of large language model training, including the scope of text and data mining exceptions, the treatment of training data as reproductions, and the conditions under which LLM outputs may infringe existing works. Develops a framework for analyzing liability across the training and generation phases and considers how human authorship doctrine applies to outputs of models trained on copyrighted material.

A detailed examination of the authorship requirement in copyright law as applied to AI-generated works. Traces the historical foundations of authorship doctrine, surveys current approaches across major jurisdictions, and argues that granting copyright to machine outputs would undermine the human incentive structure that copyright is designed to support.

Examines whether and how intellectual property law — and the press publishers' right introduced by the EU's CDSM Directive in particular — can serve as a tool for preserving the economic conditions that support quality journalism. Considers the structural challenges posed by AI-generated news content to the viability of human journalism, and proposes approaches that would keep IP law oriented toward the human cause in the media context.

A concise overview of the legal issues at the intersection of artificial intelligence and journalism, including automated journalism and robot reporters, the application of copyright and the press publishers' right to AI-generated news content, and the implications for the role of human journalists in a media environment increasingly shaped by generative AI.

Examines how the EU's press publishers' right, introduced in the CDSM Directive, intersects with the use of AI in news production and aggregation. Considers whether automated journalism outputs can qualify for protection and what the press publishers' right can and cannot accomplish in the age of generative AI.

Argues that a legal singularity is underway: for the first time, nonhuman entities not directed by humans may enter the legal system as a new "species" of legal subjects, making the legal system interspecific rather than exclusively human. Using the zero-member LLC as a concrete example, contends that rather than attempting to ban powerful AI, embedding it in legal form offers the better path: it defines targets for legal action, guides AI research toward law-following behavior, and creates a framework for AI governance. Proposes that training AI systems to internalize the spirit of the law, combined with adversarial automated compliance monitoring, offers a route to governing AI agents that may soon outperform humans across a wide range of cognitive tasks.

A public-facing companion to the Science article on interspecific law. Argues that the emergence of AI-operated corporate entities — zero-member LLCs managed autonomously by AI — is not a distant hypothetical but an imminent legal reality, and that the right response is not prohibition but proactive legal incorporation: wrapping AI agents in legal form so that their actions can be tracked, governed, and held accountable under human law.

Using a transdisciplinary methodology drawing on neuroscience, philosophy of mind, moral philosophy, and linguistics, this essay addresses a question courts will increasingly face: what is it that, as a matter of law, differentiates human beings from intelligent machines? Argues that "sapience" — the unique way in which reason and emotion interact in the human brain and body, rooted in biological embodiment and moral agency — provides a more useful and legally operationalizable criterion than sentience. Proposes a test that asks whether an entity's thinking has both biological and moral aspects characteristic of human sapience.

Reviews the application of patent incentives to inventions made by AI machines, considering two policy goals: accelerating innovation with the help of AI while continuing to promote human progress. Surveys decisions by courts across multiple jurisdictions — most of which have correctly held that machines cannot be "inventors" as the law now stands — and argues that patent law should continue to promote human ingenuity rather than redirect financial incentives to the owners and operators of AI systems.

Argues that the misalignment between the global reach of AI systems and the territorial structure of national legal systems requires a coordinated transnational regulatory approach. Proposes institutional and normative frameworks drawn from experience with international IP law and other multilateral regulatory regimes.

A landmark comparative study of the originality requirement in copyright law across major jurisdictions. Traces the divergence between the US "modicum of creativity" standard established in Feist and the "author's own intellectual creation" standard developing in EU law, with implications for how courts distinguish protectable expression from unprotectable information.

Examines the uneasy relationship between IP law and human rights frameworks, arguing that human rights can provide a principled normative foundation for IP protection — one that privileges the interests of human creators — in contrast to the trade-agreement-driven expansion that has characterized international IP law in recent decades.

Examine si les productions des systèmes d'intelligence artificielle peuvent bénéficier de la protection par le droit d'auteur, en développant le cadre de la « cause humaine » dans le contexte du droit d'auteur canadien et comparé. Soutient que la protection ne devrait être accordée qu'aux œuvres dont la cause proximale est humaine, et propose un test de causalité pour distinguer les œuvres protégeables de celles qui ne le sont pas.

Examine les fondements doctrinaux de l'exigence d'auteur humain en droit d'auteur et analyse les conséquences de l'émergence des systèmes d'IA capables de produire des œuvres littéraires et artistiques. Anticipe plusieurs des questions qui feront l'objet de litiges importants dans les années suivantes. Version française de l'argumentation développée dans The Machine as Author (Iowa Law Review, 2020).

A monograph developing the philosophical and legal foundations of the argument that law, including intellectual property law, should be oriented toward human progress in the full sense — human flourishing, self-realization, and the expansion of human capacities — rather than toward the maximization of output or economic efficiency alone.