docketing@steinip.com

  1-202-216-9505

Where Technology, Innovation and

Protection Come Together


  • HOME
  • > Blog Main Page
  • > The Copyrightability of AI: Thomson Reuters v. Ross Intelligence and Its Implications

The Copyrightability of AI: Thomson Reuters v. Ross Intelligence and Its Implications

By Bhavna Viswanathan

 

 

Generative artificial intelligence is changing how we create, consume, and distribute content. As AI developers significantly rely on existing works to train their systems, questions about ownership and infringement have become central to copyright law. Can AI developers use copyrighted material to train AI without the author's permission, and when does that use qualify as fair use? The Third Circuit's decision in Thomson Reuters Ent. Ctr. GmbH v. Ross Intelligence Inc. offers insight into the application of existing copyright principles to emerging technologies.

 

Under 17 U.S.C. § 102(a), copyright protection extends to “original works of authorship fixed in any tangible medium of expression.” A work must be independently created and contain at least a “modicum of creativity” to qualify for protection, as established by the Supreme Court in Feist. While this threshold is low, copyright law has consistently recognized human authorship as a fundamental requirement.

 

This distinction has become relevant again in January 2025, when the U.S. Copyright Office released Part 2 of its Copyright and Artificial Intelligence report, addressing the copyrightability of AI-generated works. The report clarifies that a human providing prompts alone is generally insufficient to establish authorship. When the expressive elements of a work are generated entirely by AI without meaningful human creative contribution, they are not eligible for copyright protection. Human contributions involving original expression, creative modifications, and sufficiently creative selection and arrangement may still qualify.

 

On September 29, 2026, the U.S. Court of Appeals for the Third Circuit issued its decision in Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc., No. 25-2153, slip op. at 1 (3d Cir. Sept. 29, 2026), addressing the use of copyrighted legal research materials in AI development.

 

The central question is → Does Ross Intelligence's use of Westlaw's copyrighted headnotes to train its AI-powered legal research platform constitute fair use under copyright law?

Ross Intelligence used approximately 25,000 Westlaw headnotes in developing its AI-powered legal research platform. The District Court specifically found that 2,243 of these headnotes had been copied, and those were the headnotes at issue in the Third Circuit appeal. Thomson Reuters, which owns Westlaw, alleged that Ross copied its editorial materials without authorization to develop a competing product.

 

The Third Circuit held that Westlaw's headnotes were sufficiently original to qualify for copyright protection. Though the headnotes summarized existing judicial opinions, the court recognized that Westlaw editors exercised creative judgment in selecting relevant legal principles and expressing them in independent summaries. Applying the standard established in Feist, the court found that the headnotes possessed the “requisite originality” for copyright protection.

 

Ross also argued that its use of the headnotes qualified as fair use under 17 U.S.C. § 107, but the court rejected this defense after examining the four statutory factors, which consider:

1.         The purpose and character of the use

2.         The nature of the copyrighted work

3.         The amount and substantiality of the portion used

4.         The effect on the potential market.

 

The court found Ross's use to be commercial and “insufficiently transformative” because its platform served a similar purpose to Westlaw. It also considered the extent of the copying and the potential harm to Thomson Reuters' market, including the emerging market for licensing copyrighted works as AI training data. The Third Circuit affirmed the lower court's partial summary judgment in favor of Thomson Reuters.

 

The Court also relied on established Supreme Court precedent to support its conclusion that Westlaw's headnotes were copyrightable. In Callaghan v. Myers, 128 U.S. 617, 647–49 (1888), the Supreme Court held that although judicial opinions produced in an official capacity cannot be copyrighted, legal reporters may obtain “a copyright for the volume which will cover the matter which is the result of his intellectual labor” such as “headnotes.”

 

The Court reaffirmed this principle in Georgia v. Public Resource Org., Inc., 590 U.S. 255, 265 (2020), explaining that copyright protection extends to independently created explanatory materials that do not carry the force of law. The Third Circuit applied this reasoning to Westlaw's headnotes, finding that copyright protection would not give Thomson Reuters a monopoly over judicial opinions because the law itself remains freely available and can be expressed in different ways.

 

Although Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc., No. 25-2153, slip op. at 1 (3d Cir. Sept. 29, 2026), concerns the use of copyrighted materials to train AI rather than the copyrightability of AI-generated outputs, the decision has implications for both. It reinforces that existing copyright protections remain applicable to works used in the development of new technologies.

 

The decision in Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc., No. 25-2153, slip op. at 1 (3d Cir. Sept. 29, 2026), shows why AI developers must consider how they use copyrighted material, especially when that use could affect the original creator or compete with existing works. The Copyright Office's guidance also makes clear that AI-assisted work may receive copyright protection when a person makes meaningful creative contributions.

 

Copyright law continues to protect original human creativity while adapting to new technologies, and as AI continues to evolve, courts will need to balance the rights of creators with the growing demand for technological innovation.

 

References

[1] Feist Publ'ns, Inc. v. Rural Tel. Serv. Co., 499 U.S. 340, 345–46 (1991).

[2] U.S. Copyright Office, Copyright and Artificial Intelligence, Part 2: Copyrightability (2025).

[3] Thomson Reuters Enter. Ctr. GmbH v. Ross Intelligence Inc., No. 25-2153, slip op.(3d Cir. Sept. 29, 2026).

[4] 17 U.S.C. § 107.

[5] Callaghan v. Myers, 128 U.S. 617, 647–49 (1888).

[6] Georgia v. Public Resource Org., Inc., 590 U.S. 255, 265 (2020).

[8] 17 U.S.C. § 102(a).