Amicus Briefs in ‘Thomson Reuters v. Ross’ Case Urge Application of Established Standards of Copyrightability, Fair Use
In late November, a diverse group of copyright owners, legal and economic scholars, and AI researchers submitted amicus briefs in support of Thomson Reuters (TR) in its ongoing infringement case against ROSS Intelligence. The case is currently before the Third Circuit Court of Appeals, where ROSS has challenged a lower court decision confirming that TR’s Westlaw headnotes are copyrightable and that ROSS’s unauthorized use of the headnotes for the development of a competing legal research tool does not qualify as fair use.
While the case involves technology that is different than the technologies powering large language models (LLMs) and other generative models at the center of dozens of other lawsuits, it addresses significant fair use and copyrightability issues that could have a wide-ranging impact on copyright owners and AI developers. As amici explain, it’s critical that the lower court’s decision be affirmed so that long-established standards of copyright protection are upheld and that fair use not be applied in a way that excuses unauthorized, competitive uses in the name of supposed “transformative” AI.
Briefs Underscore Copyrightability of Headnotes, Warn of Stripping Copyright Protection
RELX, a multinational company that offers the LexisNexis legal research platform, filed a brief explaining that protecting copyright in headnotes benefits the entire legal research industry, the public, and the legal system. The brief traces copyright protection for legal annotations like headnotes back to the 1800s, making clear that courts have consistently held that while judicial opinions themselves are public domain, privately authored explanatory materials like headnotes are copyrightable. The brief goes on to detail the editorial process behind annotations, explaining that attorney-editors engage in substantial creative work: reading cases, identifying key issues, synthesizing holdings, drafting summaries, selecting taxonomy classifications, and linking cross-references. As the brief explains, different companies produce significantly different headnotes, which demonstrates genuine creative choices and refutes any “merger doctrine” argument that there’s only one way to express legal principles.
The brief warns that, if competitors like ROSS are allowed to freely copy headnotes without authorization, publishers would stop creating them because the business model would collapse. This would be particularly harmful to courts, small practitioners, legal aid organizations, and self-represented litigants who depend on these research tools. Of particular note, the brief stresses that LexisNexis and Thomson Reuters are “ardent competitors,” and the fact that a competitor is supporting Thomson Reuters’ copyright claims demonstrates that fair competition in creating headnotes is both possible and desirable. This clearly undermines ROSS’s argument that upholding copyright would create a “monopoly in the law.”
The Copyright Alliance submitted a brief that also largely focuses on the copyrightability of works that involve original selection, coordination, and arrangement—a group of works that clearly includes Westlaw headnotes. As the brief explains, ROSS’s attempt to cast aside longstanding precedent to misrepresent the Westlaw headnotes as non-copyrightable would have critically damaging consequences for the Copyright Alliance’s members across many creative industries, including publishing and visual arts, that rely on protections afforded to selection, coordination, and arrangement. The brief also explains that past fair use cases—specifically Google v. Oracle, Kelly v. Arriba Soft, and the Google Books case—that ROSS and its amici argue support a finding of fair use are in no way analogous to the unauthorized use of the Westlaw headnotes and that ROSS’s infringement harms AI training markets.
The American National Standards Institute (ANSI), along with six other standards development organizations (SDOs), filed a brief offering the unique perspective of organizations that produce technical standards that, while not what people usually think of when considering traditional expressive works, are no less worthy of copyright protection. The brief explains that, much like technical standards, Westlaw headnotes involve human judgment in selecting, synthesizing, and articulating information and principles. Like Westlaw headnotes, SDOs’ technical standards are privately authored works that exist alongside government materials and provide immense public value. The brief warns that weakening copyright protection through the misapplication of the law would reward bad actors like ROSS while undermining the ability of SDOs to fund the creation of valuable public resources.
Briefs Explain ROSS’s Use is Not Transformative, Harms AI Licensing Markets
A group of film studios, including Disney, Paramount, Universal, and Warner Bros., filed a brief focusing on fair use and warning that courts should not create new AI-specific rules in place of well-established fair use principles. The brief explains that, despite what ROSS claims, the unauthorized use of the headnotes was not transformative because they were used for a virtually identical purpose as the original works: to connect legal questions with judicial opinions for the purpose of providing a legal research tool. Importantly, the brief makes clear that the Supreme Court’s Warhol v. Goldsmith decision confirmed that when considering whether a use is transformative, the ultimate purpose of the use must be considered. Thus, courts should not focus on an immediate act of copying, when, as in Warhol and this case, the ultimate purpose of the use was the same as the original work.
The Recording Industry Association of America (RIAA), joined by the National Music Publishers Association (NMPA), filed a brief that also refutes the notion that ROSS’s use of the Westlaw headnotes was transformative, explaining that there is nothing transformative about copying something when the ultimate aim is to achieve the same purpose as the work that is copied. The brief explains that the Supreme Court’s Warhol decision “resets the table” on transformative fair use after years of inconsistent application by tying the first fair use factor’s focus on purpose to the fourth factor’s notion of substitution. Considering the clearly substitutional nature of ROSS’s use, the brief rightly concludes that is neither transformative nor qualifies as fair use.
The RIAA brief also makes a critical point about AI training markets that should be recognized in all of the ongoing AI infringement cases: relevant markets for purposes of factor four include existing and potential markets for training AI models on the works at issue. The brief explains that the district court got it right in this case, finding that ROSS’s use negatively affected the potential market for AI training data, whether or not Thomson Reuters had exploited that market. Speaking to licensing markets, the brief details the emergence of numerous licensing deals in the music space and beyond that demonstrate the vital AI licensing ecosystem that’s threatened by the defendant’s actions.
Also weighing in on fair use, the Center for Art Law, a nonprofit dedicated to artists’ rights and intellectual property issues at the intersection of art and law, filed a brief that urges the court to reject any categorical AI-related exemptions. Like other amici, the Center for Art Law stresses that there is nothing inherently “transformative” about AI training, which simply “involves indexing, cataloging, or extracting aesthetic elements” and “does not itself create new authorship or transform the underlying work in a meaningful way.” The brief goes on to detail the existential threat that artists face from AI companies that copy their works en mass, citing the (i) the loss of control over creative output, (ii) substitution of original works,(iii) disruption of licensing markets and commission-based income, and (iv) threat to the livelihoods of both emerging and established artists.
Jonathan Irwy, a Fellow at the Wharton Accountable AI Lab, filed a brief that tackles transformativeness from the perspective of a technical expert and explains that what ROSS claims is “transformative” use related to its headnote copying is nothing more than translating the headnotes into a different language. According to Irwy, ROSS’s claim that it’s use is transformative because it is converting the headnotes into machine-readable language (also known as numerical vectors), is simply wrong. The brief explains that ROSS merely encodes the content with numerical representation, comparable to a change in medium that would never constitute a transformative use. The brief concludes:
“ROSS is still using the headnotes’ expressive judgments to enrich its own product. If mere re-encoding were enough to make a use transformative, any mass digital processing of text would automatically qualify, simply because computers work with numbers.”
The AI Coalition for Data Integrity (AICDI), supported by organizations including music groups, writers’ guilds, SAG-AFTRA, and others, also filed a brief focusing on transformativeness and market harm, which notably explains that recent decisions in other AI cases do not support ROSS’s position. The brief argues that district court decisions in Kadrey v. Meta and Bartz v. Anthropic incorrectly treated AI training as automatically “transformative” and failed to properly consider market harm. Discussing the fourth fair use factor, the brief makes clear that ROSS’s conduct is clearly harmful in three distinct ways: (i) direct substitution in Thomson Reuters’s core market, (ii) usurpation of derivative and licensing markets, including AI-training licenses, and (iii) broader market dilution arising from AI- powered substitutes. The brief warns that ignoring such cognizable harms would be disastrous for copyright owners and rapidly developing licensing markets. It concludes that upholding the District Court’s decision will “reinforce copyright’s incentive structure, preserve the growing market for AI-training licenses, and ensure that innovation proceeds without eroding the legal frameworks that sustain the creation of high-value editorial content.”
Publishers’ Briefs Detail Impacts on Emerging AI Licensing Markets
The News/Media Alliance (N/MA) filed a brief highlighting news publishers’ reliance on copyright to protect their investment in quality journalism and explaining that uses like Ross’ undermine their ability to create and disseminate original content. The brief details the extensive and rapidly growing licensing market where publishers are striking deals with AI companies, citing to over 100 licensing agreements with AI companies like OpenAI, Meta, Google, Amazon, Microsoft, and others. N/MA warns that enabling AI companies to divert users away from legitimate subscription-based or ad-supported publisher sites to competing services under the guise of fair use would destroy this rapidly growing market.
The N/MA brief also notes that concerns about Ross’s fair use defense are magnified by the wide deployment of retrieval-augmented generation (“RAG”) technology, through which AI companies scrape external content sources—often third-party websites—to supplement and assist in generating the output of AI products that rely on large language models. Explaining that publishers are adapting their licensing strategies to accommodate RAG uses, the brief makes clear that such licensing arrangements would be threatened if ROSS’s infringing activity was excused as fair use.
The Association of American Publishers (AAP) also filed a brief presenting extensive evidence of an emerging marketplace where publishers license their content for AI training. The brief details numerous deals between publishers and AI companies, explaining that the market is worth $6 billion today and projected to reach $52.4 billion by 2035, benefits both publishers and AI developers, and must be recognized under a fourth factor analysis. The AAP brief also makes an important point about copyright law’s Constitutional foundations, emphasizing that incentives for creation made possible by the Constitution’s grant of exclusive rights must not be undermined by notions of fair use that allow commercial entities to freely copy copyrighted works for AI development.
Copyright Scholars’ Brief Debunks Non-Expressive and Intermediate Copying Arguments
A group of 15 prominent copyright law professors filed a brief that clearly distinguishes ROSS’s activity to that deemed fair use in past cases upon which ROSS and its amici rely (as well as likening ROSS’s use to activity in cases that found against fair use). The professors make the crucial point that ROSS cannot escape liability by claiming it copied factual or non-expressive material, explaining that “the Copyright Act does not permit copying of protectable material simply because the copier asserts that the reason is to obtain the ‘factual’ material therein.” Citing to the seminal American Geophysical Union v. Texaco case, the brief explains that while Texaco claimed that its copying of journal articles was done internally for the purpose of scientists referencing the facts underlying expressive content, that did not render its reproduction fair use. The brief also cites to Sega v. Accolade to make the important clarification that there is no pro se fair use exception for “intermediate” copying even if there is no infringing output.
This is a critical point that other courts considering AI infringement cases should recognize, given that AI developers and their supporters continue to misleadingly argue otherwise. As the professors warn, adopting overbroad exceptions for non-expressive or intermediate copying “would swallow the fundamental rule of copyright protection, namely, that a copyright owner may prevent others from reproducing its work without authorization.”
Conclusion
Amici supporting Thompson Reuters represent a variety of interests from a diverse group of copyright owners, scholars, and researchers and groups working towards accountability in AI, but they all want the same thing: a proper application of established copyright laws to ensure long-protected works maintain their copyrightability and fair use is not misapplied in way that distorts notions of transformativeness and destroys emerging markets. While the facts of the TR v. Ross case must be considered independently and any decisions should not overly influence other AI infringement cases, it’s essential that the court recognize the foundational principles of our copyright system that ensure that both copyright owners and AI developers can thrive.
If you aren’t already a member of the Copyright Alliance, you can join today by completing our Individual Creator Members membership form! Members gain access to monthly newsletters, educational webinars, and so much more — all for free!