Authored By: Hagar Ahmed Rizk
Tanta University , Faculty of Law
Legal Issue Identification:
In an era where artificial intelligence serves as a driving force across various fields, the legal research sector has not remained immune to this transformation. What initially began as an attempt to revolutionize access to legal information through more advanced and intelligent research tools has culminated in a complex legal dispute that may set a precedent redefining the boundaries of intellectual property in the age of data and artificial intelligence.
This case revolves around an unprecedented legal issue concerning the use of copyrighted, edited legal content for training artificial intelligence systems without prior authorization. Thomson Reuters Enterprise Centre GmbH and West Publishing Corp. (“Thomson Reuters”) filed a lawsuit against ROSS Intelligence Inc. (“ROSS”), accusing it of unlawfully acquiring and utilizing protected editorial content as the foundation for developing an AI-powered legal research platform without obtaining legal consent or a proper license.
The case raises a fundamental legal question: Does the use of AI in processing and utilizing edited legal data fall within the scope of “fair use,” or does it constitute a clear infringement of intellectual property rights?
The resolution of this issue will shape the future relationship between technology and intellectual property and directly influence how AI is developed and utilized within the legal field.
Procedural history
This case has undergone several pivotal legal stages, marked by a significant procedural history.
- In May 2020, Thomson Reuters initiated legal action against ROSS Intelligence, alleging that ROSS unlawfully copied content from Thomson Reuters’ legal research platform, Westlaw, to train its AI-powered legal research tool.( [1])
- In 2023, the court denied Thomson Reuters’ motions for summary judgment, allowing the case to proceed to trial.
- However, on February 11, 2025([2]), following a more in-depth review of the evidence, the court reversed its stance, granting a partial summary judgment in favor of Thomson Reuters and referred the unresolved factual issues, such as determining whether some copyrights had expired and assessing the amount of damages owed, to the jury for determination in the next phase of the trial.
Statement of Facts
- ROSS sought to develop an advanced AI-powered legal research tool that utilized natural language processing (NLP) ([3]) to allow users to retrieve relevant case law by posing questions in simple, conversational language. To enable this system to function effectively, ROSS needed to train its AI using a supervised learning system. Consequently, ROSS attempted to obtain a formal license for the content of Westlaw, a vast legal database owned by Thomson Reuters, which includes editorial analyses, legal headnotes, and a precise key number classification system. However, Thomson Reuters refused to grant the license, considering ROSS a potential competitor that posed a threat to its business model. This left ROSS with a fundamental dilemma: How could it train its AI to analyze legal principles without access to well-structured legal data?
- In response, ROSS turned to a third-party company, LegalEase, which prepared training legal memoranda containing questions and answers derived from judicial opinions. While the instructions explicitly prohibited copying Westlaw’s content verbatim, subsequent investigations revealed that these memoranda were heavily derived from Thomson Reuters’ legal headnotes. A forensic analysis uncovered 2,243 headnotes that were identical to those in Westlaw, providing clear evidence of unauthorized copying.
- ROSS then utilized this data to train its AI-powered legal research tool, which had significant economic and commercial implications for Thomson Reuters, undermining its position in the legal research market. As a result, Thomson Reuters filed a lawsuit against ROSS in May 2020, alleging direct and indirect copyright infringement.
Arguments .
- Plaintiffs’ Arguments (Thomson Reuters)
Direct Copyright Infringement
Thomson Reuters asserts that the content used by ROSS constitutes original, editorially curated material that qualifies for protection under the U.S. Copyright Act (17 U.S.C. § 102(a)). The protected works include:
Headnotes: Analytical summaries of judicial decisions written by legal editors, making them a creative intellectual product subject to copyright protection.
Key Number System: A classification structure designed to facilitate case law research, constituting an original intellectual arrangement of legal information, thereby falling under copyright protection. ([4] )
Unauthorized Use of Protected Content
Thomson Reuters argues that ROSS did not obtain a license to use its content but instead resorted to indirect methods of acquiring data through LegalEase, which prepared “training memoranda” that were ultimately used to train ROSS’s AI system.
Even if ROSS did not copy the headnotes verbatim, it used derivative data, which qualifies as “unauthorized reproduction” under Supreme Court precedent.
Manipulation of the Fair Use Doctrine: Attempting to obscure the copying process through intermediaries does not change the fact that the final output is fundamentally based on protected content.
C ) Unlawful Commercial Exploitation
Beyond copyright infringement, Thomson Reuters contends that ROSS’s use of its content was not for personal or academic research but rather for pure commercial gain, thereby significantly weakening ROSS’s fair use defense.
- ROSS Used Protected Content to Develop a Commercially Competitive Product
ROSS’s use of protected content was not for research or academic advancement but to develop a commercial AI-powered legal research platform marketed as a direct alternative to Westlaw.
ROSS derived financial benefits from its business model while leveraging Thomson Reuters’ content without compensating for the data used to train its system.
Under Supreme Court precedent, any use that results in commercial profit at the expense of the original copyright holder constitutes clear infringement, even in the absence of verbatim copying.
- Profiting at Westlaw’s Expense Without Bearing Content Production Costs
ROSS did not invest in creating its own legal database but instead exploited pre-existing content that Thomson Reuters had expended substantial resources to analyze and classify.
This use undermines Thomson Reuters’ competitive advantage, as ROSS offers a similar legal research service without bearing the costs of content production.
Allowing such practices would encourage other companies to exploit copyrighted content without compensation, posing a potential threat to the entire legal research industry.
D ) ROSS’s Attempt to Circumvent Intellectual Property Laws
ROSS positioned itself as an AI-driven company that independently analyzes legal information, whereas, in reality, its platform relied heavily on protected Westlaw content.
Even in the absence of verbatim copying, training an AI model on data derived from a protected source constitutes indirect reproduction, stripping ROSS of any legal protection under the fair use doctrine.
- Defendant’s (ROSS Intelligence) Arguments
Legal Information as “Facts” That Cannot Be Monopolized
ROSS argues that the content used in Westlaw is derived from statutes and judicial decisions, which are public facts that no party can claim ownership over.
The Law Belongs to Everyone: Access to legal information cannot be monopolized, as doing so would contradict the principles of legal transparency and the rule of law.
Even if some editorial modifications exist, they do not grant Thomson Reuters the right to prevent others from using the same information.
Fair Use
ROSS asserts that its use of Westlaw content falls under the “fair use” doctrine pursuant to 17 U.S.C. § 107 for the following reasons:
- The Use Was “Transformative” and Not Mere Reproduction
Training an AI model on legal headnotes cannot be considered “copying” in the traditional sense, as AI does not store or reproduce texts verbatim. Instead, it analyzes the material, extracts patterns, and generates new results.
ROSS did not copy the content for resale but instead used it to train an AI model that reanalyzes case law in ways that were previously impossible.
The AI-generated outputs do not replicate the original Westlaw headnotes but rather produce legal analyses based on an entirely new methodology.
- No impact on the Westlaw market
The target customers of ROSS are not the same users who rely on Westlaw, meaning its product does not directly impact Thomson Reuters’ market.
Westlaw’s clients—lawyers and judges—still require the full editorial content, meaning ROSS did not create a direct substitute but rather a distinct legal research tool.
Westlaw sells legal research subscriptions, whereas ROSS provides an AI-powered analytical tool, making the two markets distinct in terms of usage and commercial value.( [5])
- Strengths of Thomson Reuters’ Position
- Registered Copyrights: Provides solid legal protection against unauthorized use.
- Clear Evidence: Demonstrates direct overlaps with ROSS’ training data.
- Market Protection: Prevents threats to its subscription-based business model.
- Fair Competition: Stops ROSS from gaining an unfair advantage.
- Safeguarding Innovation: Protects the legal research industry from exploitation.
- Strengths of ROSS Intelligence Inc.’s Position
- Public Domain Argument: Laws and rulings are not subject to copyright.
- Legal Ambiguity: Exploits gaps in AI-related copyright regulation.
- Transformative Use: AI generates new insights, not reproductions.
- Different Markets: Targets a distinct audience from Westlaw.
- Promoting Access: Expands legal information access and fosters innovation.
Court Decision
- Judgment of the court
On February 11, 2025, the United States District Court for the District of Delaware issued a partial summary judgment in favor of the plaintiffs, Thomson Reuters Enterprise Centre GmbH and West Publishing Corp., against the defendant, Ross Intelligence Inc., in case number 1:20-cv-613-SB. The court found the defendant liable for direct copyright infringement concerning the Westlaw platform’s headnotes and its proprietary key number classification system. Furthermore, the court rejected all defenses raised by Ross Intelligence.
- Reasons for the judgment
The Court based its decision on the plaintiffs’ valid copyright registrations, which were filed between 1981 and 2019. It determined that Ross had directly copied Westlaw’s copyrighted material to train its AI system, constituting a clear copyright violation.
Additionally, the Court rejected Ross’s fair use defense, ruling that the use was commercial, non-transformative, and harmful to Thomson Reuters’ competitive market position and the other defences due to the originality of the headnotes.
Conclusion :
This judgment represents a historic turning point in the evolution of intellectual property law, redefining the boundaries between technological innovation and the protection of intellectual rights in the age of artificial intelligence. Through its decision, the Court firmly established a critical legal principle: no matter how advanced technology becomes, it cannot circumvent the legal frameworks designed to safeguard creative effort and original editorial contributions.
The ruling stands as a decisive affirmation that unauthorized commercial use of protected content constitutes a clear infringement, even when pursued under the guise of technological advancement. By rejecting Ross Intelligence’s defenses, including fair use and the merger doctrine, the Court set a rigorous standard that shields legal creativity from exploitation under the pretext of innovation.
This decision is not merely a legal victory but an academic milestone that will serve as a foundation for future debates on copyright law in the digital era. It reflects a sophisticated understanding of the delicate balance between fostering technological progress and upholding intellectual property rights. Ultimately, this ruling establishes a new benchmark for justice at a time when technology and law are increasingly intertwined, marking a pivotal moment that reshapes the landscape of intellectual property protection.
Reference(s):
[1] ) BakerHostetler. (n.d.). Thomson Reuters v. ROSS: A case shaping the future of AI and copyright in legal research. Retrieved February 19, 2025, from https://www.bakerlaw.com/thomson-reuters-v-ross/
[2] ) United States District Court for the District of Delaware. 2025. Thomson Reuters Enterprise Centre GmbH & West Publishing Corp. v. ROSS Intelligence Inc. Case No. 1:20-cv-613-SB, February 11.
[3] ) Davis Wright Tremaine LLP. (2025, February). Reuters v. ROSS: Court ruling on AI, copyright, and fair use. Retrieved February 19, 2025, from https://www.dwt.com/blogs/artificial-intelligence-law-advisor/2025/02/reuters-ross-court-ruling-ai-copyright-fair-use
[4] ) Thomson Reuters Enterprise Centre GmbH & West Publishing Corporation v. Ross Intelligence Inc., No. 20-613, p. 6 (D. Del., 2024).
[5] ) Ross Intelligence Inc. (2024). Memorandum of points and authorities in opposition to Thomson Reuters’ renewed motion for partial summary judgment on fair use (No. 20-613-SB, p. 6). United States District Court for the District of Delaware.
