A major U.S. newspaper's copyright case against OpenAI and Microsoft drew filings from outside groups on Friday, Oct. 30. It is the best known of 12 lawsuits against OpenAI combined in federal court in New York. It asks whether AI companies may copy news articles and other writing, without permission or payment, to build large language models. These are the kind of AI behind chatbots such as Claude and Gemini, trained on huge amounts of text. The newspaper says the companies used its journalism without permission to build products that compete with it. OpenAI and Microsoft say the training is fair use, the part of U.S. copyright law that allows some uses of copyrighted work without permission, such as for criticism, news reporting, teaching or research. Courts decide fair use case by case, weighing how a work was used, how much was copied and whether the copying hurts the market for the original.
On Sept. 1, the Justice Department filed in support of the AI companies, arguing, according to the lawmakers, that limiting the companies' use of copyrighted writing would threaten American AI development and national security. The government's filing, as excerpted by OpenAI, says limiting how these models are built "under a misunderstanding of fair use doctrine would thwart such creative and scientific progress while hindering American prosperity and economic mobility." On Oct. 30, Representatives Jamie Raskin, Jerrold Nadler and Hank Johnson, the top Democrats on the House Judiciary Committee and two of its subcommittees, asked the judge's permission to file a friend-of-the-court brief answering the Justice Department. "National security is not a fair use factor under the Copyright Act," the lawmakers wrote, saying the Justice Department was in effect asking for a national security exception that Congress never passed.
The Wikimedia Foundation, the nonprofit that runs Wikipedia, proposed a middle path. It argues that training a model should count as fair use when the result is something new and different from the original sources. But when a chatbot looks up news articles or other pages on the web while answering and passes that same content back to users, it says that should not be fair use, and companies should follow each publisher's license. For Wikipedia, that means giving credit and keeping adaptations under the same license. "Expanding access to knowledge through AI is a powerful goal, but it must respect the people, contributors, and communities that make reliable knowledge possible," said Bernadette Meehan, the foundation's chief executive.
Arguing against the AI companies, a group including Cate Blanchett's Human Consent Foundation, the talent agency CAA and groups for musicians and voice actors filed a brief saying the companies could license what they use, according to a news report. Their brief says the creative industries already clear rights for works with scattered owners "at scale and cost-effectively," according to the report. In an earlier filing, the actors' union SAG-AFTRA and the Writers Guild of America urged the court to "reject a categorical fair use exemption for AI training," a blanket rule that all AI training is fair use, according to the same report. OpenAI, for its part, published excerpts from filings backing it, from library, tech and authors' groups, a group for blind people, individual writers and scholars, and the government. Filings it quotes argue that training uses the writing for a different purpose and does not replace it. OpenAI said the newspaper's suit seeks "an undeserved payday."
These outside briefs do not decide anything. Judge Sidney Stein still has to rule on fair use, and the Wikimedia Foundation notes that only a few cases have taken up the question so far. The foundation expects his ruling to be highly influential as more courts face it.