Considering copyright infringement liability.

A possible purposeful culpable conduct requirement to establish contributory copyright infringement.

This blog post recaps and analyzes the U.S. Supreme Court's December 2025 oral argument in the Cox Communications v. Sony Music Entertainment case, summarizing the parties' arguments, issues, and relevant caselaw. While the post concludes that the Court seems likely to shake up American copyright law by articulating a different--and likely a stricter--legal standard for what constitutes contributory copyright infringement, the exact parameters of that standard won't be known until the Court issues its opinion in 2026.

Copyright owners have sued several developers of large-scale generative AI systems for copyright infringement because of their uses of massive quantities of in-copyright works as training data for building AI models. Fair use will be the main defense against these charges. If fair use defenses succeed, developers will be free to continue to commercially exploit models already built on copyrighted data as well as to use these data to train new models or fine-tune existing ones. If copyright owners prevail, developers may be liable for billions of dollars of damages. Developers could also be enjoined from further model development on in-copyright works and even ordered to destroy models trained on infringing works. Numerous commentators have proposed collective licensing as a compromise solution to the copyright-training-data dilemma. Other commentators have questioned the feasibility of such a compromise. This article discusses several proposals for collective licensing to enable development of generative AI systems while providing some compensation to copyright owners. It assesses the complex normative, economic, and practical problems that must be addressed if such a regime is to become feasible. It discusses the implications of a licensing mandate not only for the large firms whose models are widely used today, but also for start-ups, research centers, and higher education developers of generative AI systems, as well as the general public.

This Article reviews the highs and lows of U.S. copyright case law construing software copyright scope over the nearly 50 years since copyright protection was first extended to computer programs. When the amendment was passed in 1980, initial expectations were that the scope of copyright in computer programs would be quite thin; some early cases, however, interpreted that scope rather broadly. Subsequent decisions resisted this trend, as courts sought to refine the analysis of software copyright scope, applying the doctrines of merger and scenes a faire, as well as § 102(b), to filter out unprotectable elements. Fair use also has played a role in delimiting the scope of software copyright, most recently in the Supreme Court’s Google LLC v. Oracle America, Inc. decision.

Over the past forty-some years, entertainment industry plaintiffs have made several policy-based copyright exceptionalist arguments to the Supreme Court in cases seeking to hold technology developers and service providers liable for contributory infringement because these technologies and services enabled members of the public to engage in widespread copying of popular copyrighted works. In these cases, the plaintiffs relied on a standard derived from the Second Circuit’s 1971 decision, Gershwin Publishing Corp. v. Columbia Artists Management, Inc., which defined contributory infringement as extending to anyone who knowingly and materially contributed to third party infringements. Among the policy-based arguments made in support of the Gershwin-based standard was that the only way to ensure that copyrights could be adequately protected was by holding technology developers and service providers contributorily liable, given the infeasibility of suing individual users who were making the copies. This Article characterizes such arguments as policy-based copyright exceptionalist because they derive neither from the copyright statute nor from traditional common law standards for holding defendants liable for tortious acts they either encouraged or facilitated. In all three cases, the Court was unpersuaded by the industries’ exceptionalist arguments and simply ignored the Gershwin material-contribution-with-knowledge standard. In the first two cases, the Court looked instead to transsubstantive principles of common law and rules from related laws. In Sony v. Universal, the Court found in patent law the closest analogous contributory infringement standard, which allows developers of the staple article of commerce to sell their infringement-enabling products in the market as long as the products have substantial non-infringing uses. In MGM v. Grokster, the Court similarly “borrowed” patent law’s inducing infringement rule to hold liable developers of peer-to-peer file-sharing technology who actively promoted uses of their technology to infringe copyrights. In Cox v. Sony Music, the Court declined to hold this broadband service provider liable for contributory infringement because it had not actively induced third party infringement, as in Grokster, nor tailored its service to facilitate infringement, for as in Sony, its service had substantial non-infringing uses. This Article reviews the history of copyright contributory infringement and the exceptionalist arguments in Sony, Grokster, and Cox, and considers the implications of Cox for future secondary copyright liability claims.

Assessing the evolving perspectives regarding the copyright aspects of generative artificial intelligence outputs.

Sifting through a mixed bag of legal decision disagreements.

Why the recent statewide artificial intelligence regulation legislation was vetoed.

The principal goal of this Article is to explain that the Supreme Court's decision in Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith preserved the flexible and well-balanced standards for assessing fair use defenses that the Court first established in its 1994 decision in Campbell v. Acuff-Rose Music, Inc. Campbell, which recognized that when putative fair users copy parts or all of earlier works for different purposes than the earlier works' authors, such transformative purposes tend to favor fair use defenses because such uses are less likely to harm markets for the originals. Campbell eschewed bright-line rules and directed courts to assess all fair use factors and balance them together in a flexible and holistic manner. Contrary to what some have mistakenly asserted, nothing in Warhol abjures the rich body of fair use case law, both before and after Campbell, that has recognized many types of fair use justifications. The Article begins by tracing the evolution of copyright's fair use doctrine and its broad scope, reflected in Congress's intent in codifying the doctrine in the 1976 Copyright Act and in cases such as Campbell and Warhol. This Article then reviews a wide variety of cases in which courts have found fair uses to be justified, not only based on specific parts of the text of 17 U.S.C 107 but also falling under three other well-established categories of fair use justifications. Such robustness in the case law gives this doctrine the openness to other fair use justifications that Congress intended, in keeping with the constitutional purposes of copyright law.

A stealth issue in many close copyright fair use cases is the potential invalidity of second comers’ copyrights under 17 U.S.C. § 103(a) if the secondary use is ultimately held to infringe the derivative work right. Section 103(a) of the Copyright Act says, in effect, that no copyright exists in any part of a derivative work in which the first work’s expression was “used unlawfully.” Courts have largely ignored § 103(a). But recent cases, such as Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith (2023), have raised the profile of § 103(a). This Article traces the origins, history, and case law on § 103(a), highlighting its ambiguity and significant ramifications in cases in which it was arguably relevant. By exploring the expansive range of possible interpretations and applications of § 103(a), the Article argues that a careful reading of the statutory text, in keeping with its legislative history, reveals the dual purpose of § 103: to prevent infringers from being unjustly enriched when they take protected expression from earlier works and to provide protection to the original expression contributed by derivative work authors. When courts understand this dual purpose, they can and should limit the extent to which § 103(a) should invalidate copyrights, especially in close cases of derivative work infringement and fair use rulings. When invalidations of copyrights would be inequitable as to good faith secondary uses of source works, would result in windfalls to source work authors, or would otherwise be contrary to fundamental principles of copyright law, this Article argues that the best reading of § 103 would apply it narrowly to achieve copyright’s goal of promoting the “progress of science.”

The Ninth Circuit will soon decide if five anonymous programmers can claim as much as $9 billion in statutory damages from OpenAI and GitHub for violations of an obscure law outlawing infringement-enabling removal or alteration of "copyright management information" (CMI). This Article closely analyzes that law, which is codified as §1202 of Title 17 of the U.S. Code. The law is remarkable in several respects. Statutory damages available for §1202 violations far exceed those available for actual copyright infringement. Copyright infringement need not be alleged, let alone found, for liability under §1202 to attach. Indeed, it is not even necessary to own a copyrighted work to claim a violation of §1202. These unusual features make §1202 an attractive complement in complaints alleging copyright infringement or infringement-adjacent conduct. This Article argues that courts should construe §1202 in keeping with Congress' reasons for enacting §1202: to address the risk of massive piracy of dissemination of perfect digital copies of works whose CMI was tampered with via the Internet. The Article traces §1202's origins and explains how the law's text and structure support a limited scope of application. In accordance, courts have attempted, often successfully, to limit §1202 cases; this Article identifies five inherent limitations in §1202's text, such as its double scienter requirement, and five additional, less obvious limiting principles, most significantly the so-called "identicality" requirement. The Article then focuses on the current Doe v. GitHub appeal in the Ninth Circuit and its implications for future §1202 litigations and copyright law.

On both sides of the Atlantic, policymakers are struggling to reign in the power of large online platforms and technology companies. Transparency obligations have emerged as a key policy tool that may support or enable achieving this goal. The core argument of this blog is that the Digital Services Act (DSA) creates, at least in part, a global transparency regime. This has implications for transatlantic dialogues and cooperation on matters concerning platform governance.

The main dispute between Lynn Goldsmith and the Andy Warhol Foundation (AWF) between 2017 and 2022 was whether Warhol’s 1984 creation of sixteen works of visual art based on Goldsmith’s photograph of Prince under an artist reference license was fair use or infringement. Even though the trial court and the Second Circuit disagreed about whether these creations were fair use, the Court did not address that dispute at all because it interpreted Goldsmith’s merits brief as having abandoned that claim. This Article considers the influence of the Office of the Solicitor General (OSG)’s brief on the reframing of the Warhol case as only concerning the transformativeness (or not) of a 2016 license granted by the AWF to Condé Nast to the use of one of Warhol’s Prince Series works on the cover of a commemorative magazine. After reviewing the evolution of this copyright litigation and the lower court decisions, this Article speculates about how and why OSG may have chosen to narrow the question presented in the Warhol case and why Goldsmith might have acquiesced in this. The OSG’s reframing of the issue enabled the Court to avoid addressing a largely invisible issue in the Warhol case: the implications of § 103(a) of the 1976 Copyright Act for AWF’s claims of copyright in Warhol’s Prince Series works. Even though the Court only ruled that AWF’s 2016 license grant was nontransformative, the Article suggests that much of the Warhol decision’s dicta is nonetheless likely to influence future fair use analyses. It also considers the viability of AWF’s claim of copyright in the Prince Series and suggests alternative ways the Court could have resolved the Warhol case.

The Office of the Solicitor General (OSG) has historically been very influential with the Supreme Court, especially as an amicus curiae in private litigant cases with an average win rate overall of 75%, and more than 90% in the Court's patent cases. This Article is the first to consider OSG's record in the Court's copyright cases. It reviews all thirty-one of the copyright cases on which the Court granted cert since 1984. It contrasts the rarity of OSG's participation in the Court's twentieth century copyright cases with its predominant role in the Court's twenty-first century cases. OSG's analyses have been very persuasive in the Court's copyright procedure and constitutional challenge cases, but less persuasive in the Court's copyright remedies and substantive interpretation cases. The Article drills down into these divergences between OSG's and the Court's views in substantive copyright cases, finding some were highly divergent, while others had convergent outcomes but divergent analyses. In only one of the Court's ten substantive interpretation copyright cases, Andy Warhol Foundation for the Visual Arts, Inc. V. Goldsmith, was OSG's merits argument highly persuasive with a majority of the Court. The Article reflects on why OSG and the Court so often differ in their interpretations of U.S. copyright law. It concludes with offering suggestions on how OSG might better improve its record as amicus in the Court's future copyright cases.

A long-range perspective on generative artificial intelligence litigation.

Can states force social media platforms to stop removing lawful-but-awful postings?

"Machine unlearning" is a popular proposed solution for mitigating the existence of content in an AI model that is problematic for legal or moral reasons, including privacy, copyright, safety, and more. For example, unlearning is often invoked as a solution for removing the effects of specific information from a generative-AI model's parameters, e.g., a particular individual's personal data or the inclusion of copyrighted content in the model's training data. Unlearning is also proposed as a way to prevent a model from generating targeted types of information in its outputs, e.g., generations that closely resemble a particular individual's data or reflect the concept of "Spiderman." Both of these goals–the targeted removal of information from a model and the targeted suppression of information from a model's outputs–present various technical and substantive challenges. We provide a framework for ML researchers and policymakers to think rigorously about these challenges, identifying several mismatches between the goals of unlearning and feasible implementations. These mismatches explain why unlearning is not a general-purpose solution for circumscribing generative-AI model behavior in service of broader positive impact.

For clarity and concision, we will address below the following issues: (A) Copyright ability of Generative AI outputs, (B) Claims of infringement for use of works to train models, (C) Claims of infringement for Generative AI outputs (D) Transparency & Record keeping, (E) 1 We offer these comments in response to the NOI in our personal capacities only and our views are not necessarily shared by our institutions or any other affiliated entities. Christopher Sprigman is a member of the law firm Lex Lumina PLLC, which represents clients in relation to copyright and AI issues; however, Sprigman has no involvement in that representation. Many of our comments are adapted from our prior and forthcoming articles addressing copyright and Generative AI.