On December 27, 2023, The New York Times (the “Times”) filed a complaint against OpenAI and Microsoft in the Southern District of New York based on their claimed use of the Times’s works in creating generative artificial intelligence (“AI”) products. (The New York Times Co. v. Microsoft Corp. et al., Case No. 1:23-cv-11195). The complaint alleges seven counts of intellectual property-related causes of action, including the purported copyright infringement of millions of the Times’s copyrighted news articles, investigations, opinion pieces, and other content. The products at the heart of the complaint are Microsoft’s Copilot (formerly known as Bing Chat) and OpenAI’s ChatGPT, two chatbot tools allegedly powered at least in part by the partnership between Microsoft and OpenAI in developing generative AI technology.
In its complaint, the Times argues that Microsoft and OpenAI engaged in “systematic and competitive infringement” by copying the Times’s content into their large-language models (“LLMs”) that fuel Copilot and ChatGPT. Put simply, LLMs are programs that require training on large datasets of text (often scraped from the Internet) to run. An LLM is fed text until it begins recognizing patterns and predicting appropriate word associations. This process culminates with the chatbot utilizing the LLM being able to provide an intelligible response to users’ prompts. The complaint alleges that the building, storage, and reproduction of text datasets containing the Times’s works in training LLMs infringes the newspaper’s exclusive rights in that material.
Further, the lawsuit contends that ChatGPT and similar generative tools will output portions of Times content; according to the lawsuit, the tools generate outputs that are substantially similar to the Times’s works (and in some instances, contain verbatim reproductions.) The Times alleges that these generative AI tools directly compete with it, insofar as they provide infringing substitute products for its content. According to the complaint, the Times contacted Microsoft and OpenAI in April 2023 to raise its concerns and begin licensing negotiations, but no agreement has been reached.
The filing anticipates fair use as a potential defense Microsoft and OpenAI may raise. Under copyright’s fair use doctrine, codified at 17 U.S.C. § 107, some unlicensed uses of copyrighted content will not be deemed infringing if used “for purposes such as criticism, comment, news reporting, teaching . . ., scholarship, or research.” (This list is illustrative rather than exhaustive.) Whether copyrighted material has been used in a manner consistent with fair use depends on the weighing of four statutory factors. Under the first statutory factor, which examines the purpose and character of the use, recent copyright case law has shifted emphasis onto whether the secondary use was transformative in nature. Many federal courts have previously deemed technological innovations utilizing copyrighted material as being sufficiently transformative to merit fair use’s protection.
The Times’s complaint states that Microsoft and OpenAI have publicly explained that their use of content to train generative AI models is transformative. The Times’s pleading rejects this proffered characterization, instead painting the conduct as theft of its audience by closely mimicking its expressive works.
In addition to the allegations of direct copyright infringement, the Times claims vicarious and contributory copyright infringement resulting from the collaboration between Microsoft and OpenAI in developing these products, along with a cause of action for the claimed removal of the Times’s copyright management information from the works. The Times is also alleging unfair competition and trademark dilution, arguing that the products have been known to “hallucinate,” i.e. to incorrectly quote works or misattribute authors and publications.
This lawsuit is the latest in a series of litigations that copyright owners have brought against OpenAI arising from its ChatGPT offering. The Times joins comedian Sarah Silverman, writer Paul Tremblay, writer Michael Chabon, writer Julian Sancton, and the Authors Guild as parties that brought copyright claims against the tech company in 2023. This wave of litigation against OpenAI comes as legal tensions rise between creatives and AI technology companies. Copyright practitioners will be watching these cases closely in the coming months to see how courts will weigh the plaintiffs’ claims against the fair use arguments that will undoubtedly form the core of OpenAI’s and other AI companies’ defenses.
