Image used under license from Shutterstock.com.
On January 14, 2025, the U.S. District Court for the Southern District of New York held oral arguments on the motions to dismiss The New York Times’ (NYT) complaint by OpenAI and its financial backer, Microsoft.[1] The complaint alleges that OpenAI’s use of NYT’s copyrighted works to train its AI models constitutes copyright infringement.[2] NYT further alleges that Microsoft was contributorily liable by assisting, controlling, and profiting from OpenAI’s infringement.[3] However, OpenAI and Microsoft claim that their actions fall within the bounds of “fair use,” thus precluding them from legal liability.[4]
OpenAI is an American artificial intelligence (AI) organization that is most known for releasing ChatGPT, a popular chatbot and generative AI tool.[5] Trained AI models, like ChatGPT, are able to make judgments and predictions when responding to a prompt by identifying patterns and relationships.[6] OpenAI’s ChatGPT runs on a “transformer” model; a type of AI neural network architecture that operates on the principle of next-word prediction.[7] Based on the user’s text prompt, the model determines what the most probable next word would be following the user’s input.[8] So, when ChatGPT is given a string of text as input, its response is generated one word at a time, with each new word depending on the context of the previous words.[9]
NYT and other publishers, including The New York Daily News and the Center for Investigative Reporting,[10] claim that using the publishers’ content to train generative AI chatbots removes the need for users to visit their websites and purchase subscriptions to read their articles. It is alleged that users can use the chatbot to access copies of NYT’s works. In its complaint, NYT included evidence of user queries prompting OpenAI’s chatbot to respond with “verbatim excerpts from the original [NYT] article” without including a prominent affiliate link.[11] Several figures compared the large language model’s output with the actual text from NYT’s articles and highlighted the copied verbatim portions in red text.[12] NYT’s claims extend beyond just “stripping away the news organizations’ author and copyright information” – it also includes concerns of using large language model products to evade paywalls.[13]
OpenAI’s motion to dismiss is based, in part, on its fair use defense. Under this doctrine, copyrighted materials may be used without permission under certain conditions.[14] Fair use claims are fact-specific inquiries and therefore rely on a balancing test of the factors.[15] If the use is noncommercial, transformative, and is not used in a way to harm the original copyright owner, then the use is likely fair.[16]
OpenAI argues that their use of the text is transformative and does not substitute for the original use of the work. Joseph Gratz, an OpenAI lawyer, states, “This isn’t a document retrieval system. It is a large language mode[l].”[17] Attorneys for OpenAI further explain that when “OpenAI’s artificial intelligence models are fed data, it is then sorted into a series of ‘tokens,’ units that make analyzing the data more manageable.”[18] From the weights of data derived from the training process, the model can recognize patterns and relationships. OpenAI further argues that there was no proof of harm to NYT by removing the copyright management information (CMI) such as “mastheads, author bylines, and other identifiable information.”[19] According to Annette Hurst, an attorney representing Microsoft, “The plaintiffs . . . have alleged that this technology is capable of being commercialized to the tune of billions of dollars without regard to any capability for how.”[20]
The major legal question raised by this case, whether AI companies’ use of copyrighted writing to train their models constitutes fair use, was not directly addressed during the hearing on January 14.[21] Instead, Judge Stein considered the dismissal motions to trim the claims in NYT’s complaint.[22]
Although this lawsuit is still very much in its beginnings, its outcome will have major implications for the growing use of generative AI tools in today’s digital media age. Such concerns range from the potential effects on authorship in journalism to the reliability of information produced by the chatbot itself. Whatever it may be, it is important to remain diligent and up to date as this case, and other similar cases, proceed.
[1] Rachel Scharf, OpenAI Products Not Designed To Evade Paywalls, Judge Told, https://www.law360.com/articles/2279493/openai-products-not-designed-to-evade-paywalls-judge-told (Jan 14, 2025, 4:42PM).
[2] See generally, Complaint, The New York Times Company v. Microsoft Corporation, et al., Case No. 1:23-cv-11195, United States District Court, Southern District of New York, https://nytco-assets.nytimes.com/2023/12/NYT_Complaint_Dec2023.pdf (Filed on Dec. 27, 2023).
[4] Sara Guaglione & Marty Swant, OpenAI, The New York Times debate copyright infringement of AI tech companies in trial arguments, https://digiday.com/media/openai-the-new-york-times-debate-copyright-infringement-of-ai-tech-companies-in-first-trial-arguments/ (Jan. 15, 2025).
[5] Coursera, What is OpenAI? Everything You Need to Know, https://www.coursera.org/articles/what-is-openai (Oct. 24, 2024).
[6] Calin Cretu, How Does ChatGPT Actually Work? An ML Engineer Explains, https://www.scalablepath.com/machine-learning/chatgpt-architecture-explained (last updated Mar. 1, 2024).
[7] Aeree Cho, et al., Transformer Explainer, https://poloclub.github.io/transformer-explainer/ (last visited Feb. 16, 2025).
[8] Ashish Vaswani, et al., Attention Is All You Need, https://proceedings.neurips.cc/paper_files/paper/2017/file/3f5ee243547dee91fbd053c1c4a845aa-Paper.pdf (2017).
[10] Guaglione & Swant, supra note 4.
[11] The New York Times Co., supra note 2, ¶ 122-123.
[12] Id. at ¶ 99-100.
[13] Scharf, supra note 1.
[14] Guaglione & Swant, supra note 4.
[15] U.S. Copyright Office, U.S. Copyright Office Fair Use Index, https://www.copyright.gov/fair-use/
[16] https://www.copyright.gov/fair-use/ (last updated Nov. 2023).
[17] Bobby Allyn, ‘The New York Times’ takes OpenAI to court. ChatGPT’s future could be on the line, https://www.npr.org/2025/01/14/nx-s1-5258952/new-york-times-openai-microsoft (last updated Jan. 14, 2025, 4:27 PM).
[18] Id.
[19] Guaglione & Swant, supra note 4.
[20] Id.
[21] Scharf, supra note 1.
[22] Id.