The Justice Department just told a federal court that making OpenAI pay to train on news articles could put the country at risk.
The DOJ filed a statement of interest in the consolidated publisher lawsuits against OpenAI, backing the company's fair use defense. The filing argues that licensing costs for training data would threaten national security, framing copyright liability as a competitiveness problem rather than a copyright one. The publishers bringing the case want compensation for the news content used to train OpenAI's models. The brief does not decide the case, but it signals where the administration stands on a fight that will shape how every AI company sources training data.
A statement of interest is not a ruling, but courts read them as a proxy for policy. If national security becomes an accepted argument for fair use, plaintiffs in any AI training-data case, not just publishers, face a harder path to compensation. The stakes are sharper because the EU runs the opposite system: its AI Act has no fair use doctrine and applies copyright and text-and-data-mining opt-out rules to any provider placing a model on the EU market, no matter where the training happened.
Same training run, different verdict depending on the border it crosses, and OpenAI's lawyers just got the DOJ's help arguing the American side.