Mark A. Lemley is an American legal scholar and the William H. Neukom Professor of Law at Stanford Law School, where he directs the Stanford Program in Law, Science and Technology. He is also of counsel at the litigation boutique Lex Lumina PLLC. Lemley is one of the most-cited intellectual-property scholars in the United States, and his work on copyright, patents, and the economics of information has become a frequent reference point in disputes over how copyright law applies to the data used to train generative AI systems.
Work on AI and copyright
Lemley's analysis is among the academic foundations of the "fair use" position in the AI training-data debate — the argument that training a model on copyrighted works can be a transformative, non-infringing use. With Bryan Casey he co-authored "Fair Learning," which argues that using copyrighted works to train machine-learning systems should generally qualify as fair use because the model learns unprotectable facts and patterns rather than copying expression; this work is summarized on its own page at Fair Learning — Mark A. Lemley & Bryan Casey (Texas Law Review, 2021).
He has continued to publish on generative AI specifically. In "How Generative AI Turns Copyright Upside Down" (2024), Lemley argues that generative-AI systems unsettle copyright's traditional assumptions, including the premise that copying is the central harm and that humans are the relevant authors (Source: https://law.stanford.edu/wp-content/uploads/2024/09/2024-09-30_How-Gerative-AI-Turns-Copyright-Upside-Down.pdf). He has written and spoken on related questions of AI authorship, fair use, and the limits of style protection (Source: https://www.theregreview.org/2025/06/07/seminar-copyright-and-generative-ai/). The last point bears directly on proposals such as the CREATOR Act and the NO FAKES Act (federal, proposed): Lemley and other intellectual-property specialists have questioned the enforceability of bills that would create rights against AI imitation of an artist's "style," citing the difficulty of legally defining a style and the role of fair use.
Lemley's positions place him on the comparatively permissive side of the AI-copyright debate, in contrast to scholars and creator advocates who argue that unlicensed training is infringement and that licensing markets should govern the use of creative works for AI. His work is cited on both sides of active training-data litigation.
Relationships
- related: Fair Learning — Mark A. Lemley & Bryan Casey (Texas Law Review, 2021) (his co-authored "Fair Learning" argument), AI Content Licensing, AI and Tort Liability (if present), CREATOR Act, NO FAKES Act (federal, proposed).
- related: Midjourney and other generative-image firms whose training-data practices his analysis addresses.