The Future of AI and Copyright Law

In 2023, The New York Times sued OpenAI after the organization used Times articles to train large-language models (LLMs), on which its generative artificial intelligence (A.I.) services, including ChatGPT, rely. These LLMs incorporate information from datasets to build patterns and word recognition within a specific context. The complaint claims that OpenAI infringed on its copyright by utilizing Times material without permission to create these datasets, thus “build[ing] substitutive products.”

When users submit questions, LLMs produce results that are reproductions of finite datasets, some of which are substantially similar to the original material. Providing users with the means to generate such summaries undermines the Times’s journalistic efforts, precluding the need for users to pay the Times directly for access.

OpenAI argued publicly that its use of Times material to build datasets for LLMs is protected by the doctrine of fair use, primarily because its use was transformative. 

Under Section 107 of the Copyright Act, fair use is determined by balancing various factors, one of which is the purpose and character of the allegedly infringing use. This includes whether it is intended for commercial or transformative and educational purposes. Categorically, transformative works change an existing work to create something that has a new purpose, meaning, or creative expression. 

OpenAI’s use of Times material “to create products that substitute for The Times and steal audiences away from it” is not transformative, says the complaint. The lawsuit brings to question whether independent journalism is in direct competition with A.I. services, most others of which, such as co-defendant Microsoft, also build their datasets by scouring online sources.

In addition to the nature of fair use, under Section 107, courts must consider the potential effect of the allegedly infringing use on the market or value of the copyrighted work. Whether The Times has insofar provided evidence of actual harm to marketability is unclear, but it asserts in its complaint that its copyrighted material has been “substantially encoded” into LLMs, allowing users to bypass The Times’s paywall and diminishing market value of Times publications. “If individuals can access The Times’s highly valuable content… without having to navigate through The Times’s paywall, many will likely do so,” says the complaint. 

The precedent in a case like this would be in reference to Harper & Row v. Nation Enterprises (1985), this case helps us understand the framework for cases dealing with monetary loss and fair use. Harper & Row, the copyright owner of former President Gerald Ford’s memoir, had already negotiated a deal giving exclusive rights to Time Magazine so they could publish excerpts before the memoir's release. However, before the memoir’s publication, Nation Enterprises received the unpublished memoir from an unauthorized source and released an article with verbatim quotes from the copyrighted memoir; completely violating Harper & Row’s copyright and entirely eroding the value of Time’s deal with Harper & Row. Time Magazine had to cancel its article and broke off the deal with Harper & Row. The courts eventually decided that the Nation article was not protected under fair use because they used “the heart of the book” and hurt the market for the contracted first publication. 

As such, it would be in the favor of The New York Times to demonstrate that the content of its articles is substantively relayed via ChatGPT and OpenAI services in response to user queries in such a way that diminishes The Times’s exclusivity rights. The most recent news from the case is that on May 13, 2025 a preservation order now requires OpenAI to retain conversation logs, which The New York Times will search through in determining the scope and effect of the alleged infringement. 

The purpose of copyright is to protect authors and creators’ exclusive rights over their original, human-made work, and the implications of this case are not limited to this one individual matter. A ruling in favor of OpenAI would threaten the future of independent journalism, setting a precedent for the continued exploitation of news outlets and writers.