The impact of copyright law on the use and training of generative artificial intelligence (“AI”) continues to pose novel questions to courts. Recently, two cases addressed critical questions in this ever-evolving field of law, with one Court holding that a developer had engaged in copyright infringement through its use of creative and original materials for the training of AI models, and another determining that AI-generated work does not qualify for authorship under the Copyright Act of 1976. These decisions will play a role in shaping the landscape of AI, and its protections and boundaries under copyright law.
I. Delaware District Court Finds AI Developer Engaged in Copyright Infringement.
On February 11, 2025, Judge Stephanos Bibas, sitting on the Delaware District Court by designation from the Third Circuit, granted summary judgment to Thomson Reuters Enterprise Centre GMBH and West Publishing Corp. (“Plaintiffs”) on their copyright infringement claim, rejecting Ross Intelligence, Inc.’s (“Defendant”) fair use defense regarding its use of Plaintiffs’ materials to train its AI search tools.
Plaintiffs are industry leaders in the online legal research industry, offering “subscribers access to a comprehensive collection of legal information that is easily searchable….”[1] Plaintiffs’ services offer allegedly creative and original materials, such as (i) editorial enhancements, (ii) case headnotes, (iii) notes of decisions, and (iv) Plaintiffs’ unique West Key Number System.[2] Defendant, Ross, is a legal research company that “builds AI-driven products to augment lawyers’ cognitive ability.”[3]
On March 6, 2020, Plaintiffs brought suit for copyright infringement against Defendant, claiming that Defendant used critical features of Westlaw without permission.[4] After Plaintiffs refused to grant Defendant, a competitor, access to their copyrighted content, Defendant allegedly induced a third party with a limited Westlaw license to unlawfully distribute and reproduce proprietary material.[5] The misappropriated content from Westlaw was then used to train a machine learning and AI natural language search engine.[6]
The parties moved for summary judgment on copyright and fair use, and on September 25, 2023, Judge Bibas denied the parties’ motions finding that issues of fact remained.[7] However, as explained in his February 11, 2025, ruling, Judge Bibas “invited the parties to renew their summary-judgment briefing” and ultimately conceded that his original opinion was incorrect.[8]
A. Copyright Infringement.
To establish direct copyright infringement, a copyright owner must demonstrate that (1) it owned a valid copyright, and (2) the infringer copied protectable elements of the copyrighted work.[9] A copyright owner must demonstrate (a) actual copying, (b) resulting in a copy substantially similar to the protected work.[10]
The Court found that Plaintiffs had established the validity of a compilation copyright, noting that Thomson Reuters has maintained copyright registrations for nearly 40 years. Judge Bibas also noted that Westlaw headnotes constitute original “factual compilations” that certainly use “a minimal degree of creativity.”[11]
Next, the Court determined that Defendant had copied Plaintiffs’ original headnotes, basing its finding on expert testimony, Defendant’s access to the materials through a third party, probative similarity, and the Court’s own comparative analysis.
Finally, Judge Bibas found that 2,243 headnotes were substantially similar, leading the Court to grant Plaintiffs’ motion for summary judgment on copyright infringement. In granting Plaintiffs summary judgment, the Court also rejected Defendant’s numerous affirmative defenses, including the defense of fair use.
B. Fair Use Doctrine.
The Fair Use Doctrine is an affirmative defense to copyright infringement. If use of a copyrighted work constitutes fair use, permission from the copyright owner is not required. Judge Bibas reversed his earlier finding on fair use after weighing the four factors that are considered by courts: “(1) the use’s purpose and character, including whether it is commercial or nonprofit; (2) the copyrighted work’s nature; (3) how much of the work was used and how substantial a part it was relative to the copyrighted work’s whole; and (4) how [the] use affected the copyrighted work’s value or potential market.”[12] The Court emphasized that the first and fourth factors are the most significant in determining fair use.
Weighing the fair-use factors, the Court found that the first factor—the purpose and character of the use—favored Plaintiffs, concluding that Defendant’s use was commercial in nature and not transformative. Specifically, Defendant used Plaintiffs’ headnotes as AI training data to develop a competing product. The Court then evaluated the second factor—the nature of the copyrighted work—and found it favored the Defendant, observing that while Plaintiffs’ work involved some creativity, the headnotes, unlike a novelist’s or artist’s work, were “not that creative.” As for the third factor—the amount and substantiality of the portion used—the Court again sided with Defendant, noting that the final output provided to end users did not contain Plaintiffs’ headnotes verbatim. However, the fourth factor—the effect of the use on the potential market—weighed in favor of Plaintiffs. The Court emphasized that Defendant aimed to create a market substitute and could not have developed its research engine without infringing on Plaintiffs’ copyright. Accordingly, the Court granted Plaintiffs’ motion for summary judgment and rejected the Defendant’s fair use defense.[13]
II. Court of Appeals for the D.C. Circuit Holds “Non-Human Machine” Cannot be a Copyright Author.
In another case likely to have widespread impact on AI, on March 18, 2025, the D.C. Circuit upheld the denial of a copyright application that listed a generative AI as the author of an artwork.[14]
Dr. Stephen Thaler, a computer scientist, developed a generative AI called “Creativity Machine.” He submitted a copyright registration application for a work titled A Recent Entrance to Paradise. However, Dr. Thaler did not create the image himself—the Creativity Machine did. As a result, he listed the AI as the sole author and himself as the Copyright Claimant on the application.[15]
The United States Copyright Office denied Dr. Thaler’s application because “a human being did not create the work.”[16] Dr. Thaler sought reconsideration of the Copyright Office’s decision, but the Registration Program again denied Dr. Thaler’s application.[17]
Dr. Thaler sought review in the United States District Court for the District of Columbia. For the first time, Dr. Thaler argued that “the work is copyrightable because a human—Dr. Thaler— provided instructions and directed his AI.” Dr. Thaler also argued that he owns the copyright under the work-made-for-hire provision. The District Court affirmed the denial of the application.[18] Dr. Thaler then appealed to the D.C. Circuit.
A. Human Authorship Requirement.
The D.C. Circuit acknowledged that while the Copyright Act does not explicitly define "author," traditional principles of statutory interpretation indicate that copyright protections apply only to humans. The Court emphasized that a copyright is a form of property, stating, “[a]n entity that cannot own property cannot be an author under the statute.”[19] The Court cited multiple provisions of the Copyright Act to support this interpretation, noting that Congress structured copyright protection around a traditional human life. It concluded, “[m]achines do not have property, traditional human lifespans, family members, domiciles, nationalities, mentes reae, or signatures.”[20]
Alternatively, the Copyright Act consistently refers to machines, devices, computer, and programs as tools that assist human authors rather than as authors themselves.[21] As early as the 1960s, with the rise of computer technology, the U.S. Copyright Office recognized that determining authorship would become increasingly complex as technology advanced. In response, the Copyright Office formally adopted the human-authorship requirement in 1973, reinforcing the principle that copyright protections apply only to works created by humans.[22]
The Court rejected Dr. Thaler’s claim that the artwork generated by his AI system was eligible for copyright protection under the work-for-hire doctrine. It determined that the doctrine was inapplicable, as there was no copyrightable interest to assign to Dr. Thaler, given that the work was not created by a human author.[23]
B. Impact on Works Made with Artificial Intelligence.
The Court stated that the human-authorship requirement “does not impede the protection of works made with artificial intelligence.”[24] The Copyright Office has both accepted and rejected attempts to register works made by human authors using AI largely “based on how the AI tool operates and how it was used to create the final work.”[25]
The Court emphasized that it is up to Congress to amend the Copyright Act in response to the rapid and ongoing advancements in AI technology. Until then, courts are obligated to “apply the statute as it is written,” meaning that copyright protection remains limited to human authors—at least for the time being.[26]
Cullen and Dykman’s Intellectual Property team continues to monitor important developments in trademark and copyright law. Should you have any questions about this legal alert, please feel free to contact Karen Levin (klevin@cullenllp.com) at (516) 296-9110, Ariel Ronneburger (aronneburger@cullenllp.com) at (516) 296-9182, or Jordan Milite (jmilite@cullenllp.com) at (516) 296-9128.
This advisory provides a brief overview of the most significant changes in the law and does not constitute legal advice. Nothing herein creates an attorney-client relationship between the sender and recipient.
Footnotes
[1] Compl. ¶ 11, Thomson Reuters Enter. Ctr. GmbH v. Ross Intel. Inc., No. 1:20-cv-613-SB (D. Del. May 6, 2020).
[2] Id.
[3] Our Company, Ross Intelligence, https://www.rossintelligence.com/about-us.
[4] Compl. ¶ 1, Thomson Reuters Enter. Ctr. GmbH, No. 1:20-cv-613-SB (D. Del. May 6, 2020).
[5] Id. at ¶ 11.
[6] Id. at ¶ 33; Thomson Reuters Enter. Ctr. GmbH v. Ross Intel. Inc., No. 1:20-cv-613-SB (D. Del. Sept. 25, 2020).
[7] Thomson Reuters Enter. Ctr. GmbH, No. 1:20-cv-613-SB (D. Del. Sept. 25, 2020).
[8] Thomson Reuters Enter. Ctr. GmbH v. Ross Intel. Inc., 2025 U.S. Dist. LEXIS 24296, *1 (D. Del. 2025) (“A smart man knows when he is right; a wise man knows when he is wrong. Wisdom does not always find me, so I try to embrace it when it does—even if it comes late, as it did here. I thus revise my 2023 summary judgment opinion and order in this case.”)
[9] Id. (citing Feist Publications, Inc. v. Rural Tel. Serv. Co., 490 U.S. 340, 361 (1991)).
[10] Id. at *5 (citing Damn Things from Den. V. Russ Berrie & Co. 290 F.3d 548, 561-62 (3d Cir. 2002)).
[11] Id. at *7 (citing Feist, 499 U.S. at 348).
[12] Id. at *16 (citing 17 U.S.C. § 107(1)-(4); Authors Guild v. Google, Inc. 804 F. 3d 202,220 (2d Cir. 2015)).
[13] Id. at *16-23.
[14] Stephen Thaler v. Shira Perlmutter, et. al., No. 23-5233 (D.C. Cir. Mar. 18, 2025).
[15] Id. at 2, 6-9.
[16] Id. at 6.
[17] Id. at 3.
[18] Id. at 8.
[19] Id. at 10.
[20] Id. at 12.
[21] Id.
[22] Id. at 13-14.
[23] See id. at 23.
[24] Id. at 19.
[25] Id. (citing Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence, 88 Fed. Reg. 16,190, 16,192 (Mar. 16, 2023)).
[26] Thaler, No. 23-5233 at 21.