U.S. government backs OpenAI’s fair use defense in New York Times copyright lawsuit

The U.S. government told a federal court that training large language models on copyrighted works is ‘extraordinarily transformative’ and should qualify as fair use, siding with OpenAI in its legal dispute with the New York Times. In a court filing, the Department of Justice argued that a narrow interpretation of fair use would ‘thwart creative and scientific progress’ and undermine U.S. leadership in artificial intelligence. The letter states that the government has a ‘strong interest’ in how courts define fair use in the context of AI training, warning that a ruling against OpenAI could have ‘disastrous consequences’ for the industry.

The filing disputes the New York Times’ claim that OpenAI’s use of its articles was not fair use, asserting that the paper’s definition is ‘inconsistent’ with current copyright law. The government also argued that LLMs trained on copyrighted material do not directly compete with the original works, a factor courts consider in fair use analyses. It cautioned that equating AI training with output could imply that human writers who study others’ works—comparing the practice to a young Joan Didion typing out Ernest Hemingway’s stories—would similarly violate copyright.

The Department of Justice letter applies not only to the New York Times’ lawsuit against OpenAI and Microsoft but also to ‘all of the parties in this case and related cases,’ including other publishers and authors. Legal experts told WIRED that while the presiding judge is not obligated to follow the letter, such filings from the DOJ are typically given significant deference in copyright cases.

The filing arrives amid a wave of high-profile lawsuits over AI training data. Last year, Meta prevailed in a copyright case, though the judge noted insufficient evidence of harm, while Anthropic lost a separate case and was ordered to pay $1.5 billion in damages—though the judge ruled that AI training itself was fair use. Other recent lawsuits include Sony and Warner Music suing Anthropic over alleged unauthorized use of copyrighted music to train its models.