US Government Backs OpenAI's Fair Use Argument in New York Times Copyright Case

The U.S. government has filed a 20-page brief defending OpenAI's stance in its copyright fight with The New York Times, arguing that training AI models on copyrighted material can qualify as fair use. The Department of Justice stated that a broad ruling requiring AI companies to license copyrighted material for training could harm competition, innovation, and U.S. national security.

The brief, filed on September 1 in the U.S. District Court for the Southern District of New York, focuses on the use of copyrighted works for AI model training, not on potentially infringing outputs. The government contends that this training process transforms copyrighted text into material for learning language patterns, enabling models to make predictions without making the original works publicly accessible.

According to the government, the training is "transformative—spectacularly so," serving a different purpose than the original articles. It also warns that licensing requirements could favor large media companies and disadvantage smaller AI firms, potentially stifling competition. The filing asserts that the "training of AI models on copyrighted material, in and of itself, does not violate copyright laws."

Broader copyright liability for training is deemed "problematic—and legally incorrect" by the government, as it could generally make AI model training impermissible without licensing. This position aligns with broader U.S. policy objectives to maintain global leadership in artificial intelligence.

The New York Times sued OpenAI and Microsoft in December 2023, alleging unauthorized copying of its copyrighted journalism for the development of OpenAI's GPT models. The lawsuit claims the models can reproduce parts of Times articles, potentially bypassing paywalls.