Weighing in on a legal question with trillion-dollar stakes, the administration says AI firms don't need to pay for the millions of works they use without permission.
Trending on Billboard
The Trump administration is siding with AI companies in the nationwide legal battle over AI training, arguing in a first-ever court filing that labs like OpenAI are shielded from lawsuits when they ingest millions of copyrighted works to build their models.
In an advisory brief filed Tuesday (Sept. 1), the Department of Justice urges a federal judge to rule for OpenAI in a lawsuit filed by the New York Times — one of dozens of copyright cases filed by record labels, authors, movie studios and other creators against AI companies.
Weighing in for the first time on a legal question with trillion-dollar stakes, the feds say that AI companies are generally protected by copyright’s fair use defense when they train their machines on existing works, meaning they are not required to pay the original creators.
“Constraining [large language model] development under a misunderstanding of fair use doctrine would thwart… creative and scientific progress while hindering American prosperity and economic mobility,” the DOJ’s attorneys write in the filing.
The filing will be unwelcome news for major music companies, which have repeatedly sued Anthropic, the maker of Claude, and Suno, the leading music AI firm, for using millions of songs in training without permission. Like all such litigation, fair use is the make-or-break question in those cases.
In the new brief, DOJ argues that AI training is an “extraordinarily transformative” use of existing works that does not harm the market for the originals. Requiring payment for such training materials would “disproportionately benefit legacy media outlets” and merely provide them with “subsidies,” the feds say.
“It would be problematic — and legally incorrect — to impose broad copyright liability that would generally render training of AI models impermissible without licensing,” the DOJ’s attorneys write. “LLM training is consistent with that creative ‘progress’ that is the basic constitutional objective of copyright itself.”
The new brief takes a far more pro-AI stance than a report released last year by the federal Copyright Office, which said using “vast troves” of existing works to produce a flood of competing content “goes beyond established fair use boundaries.” Days after that report was released, both the Librarian of Congress and the Register of Copyrights, the top leaders at the Copyright Office, were fired by the Trump administration.