Lately, the United States District Court for the Southern District of New York ("S.D.N.Y.") rendered a decision allowing the copyright lawsuit filed by The New York Times and other media institutions against OpenAI and Microsoft to proceed with a lawsuit, entering the next stage of substantive trial proceedings[1].

The New York Times had accused Microsoft and OpenAI of using its copyrighted works without authorization to train AI models, asserting that these companies reproduced New York Times articles and derived significant commercial benefit from them[2]. OpenAI responded that training AI models using publicly available content constitutes “fair use,” which is a right protected by the law and crucial to U.S. competitiveness[3].

Whether the use of copyrighted works to develop AI models qualifies as fair use has become a hot topic in information technology industry and intellectual property law in recent years. While this issue has been discussed in Taiwan, no clear judicial ruling has yet been made. In contrast, U.S. case law provides a relevant reference: Thomson Reuters Enter. Ctr. GMBH v. Ross Intelligence Inc.

In this case, Thomson Reuters accused Ross Intelligence of using its Westlaw database’s headnotes to train an AI legal research tool without authorization, thus infringing its copyright. After analyzing the four elements of fair use, the court rejected Ross’s fair use defense: (1) Purpose and nature of the use: Weighed against the defendant, as the use was commercial in nature; (2) Nature of the copyrighted work: Weighed in favor of the defendant, since case summaries have a lower degree of creativity; (3) Amount used: Weighed in favor of the defendant, since the final product did not display Westlaw content directly to users; (4) Effect on the market: Weighed against the defendant, as the use could directly affect the plaintiff’s market position[4].

Regarding the issue of fair use in AI development, both copyright holders and AI developers should pay close attention to the comprehensive evaluation of the above four factors. It is advisable to conduct a thorough legal risk assessment at the early stages of AI project planning to carefully weigh these considerations.

‍

[1] New York Times Co. v. Microsoft Corp., S.D.N.Y., Mar. 26, 2025, Order on Motions to Dismiss

[2] The Times’s complaint: https://nytco-assets.nytimes.com/2023/12/NYT_Complaint_Dec2023.pdf

[3] Open AI’s response: https://openai.com/blog/openai-and-journalism

[4] Thomson Reuters Enter. Ctr. GMBH v. Ross Intel. Inc., No. 1:20-CV-613-SB, 2025 WL 458520 (D. Del. Feb. 11, 2025)

‍