Analysis
The New York Times' copyright lawsuit against OpenAI and Microsoft moved into a critical new phase, with all three parties filing summary judgment arguments before a federal judge, Axios reported September 8. The Times argues OpenAI and Microsoft copied its journalism at scale to build commercial substitutes for its own product; OpenAI counters that scraping publicly available text to train large language models is protected under existing fair-use precedent.
OpenAI's filing leans on two 2025 California federal rulings -- Kadrey v. Meta and Bartz v. Anthropic -- both of which found that training AI models on copyrighted text was transformative fair use. The Times doesn't dispute those outcomes directly but argues they don't apply here because, unlike the plaintiffs in those cases, the Times can show ChatGPT outputs directly compete with and substitute for its own journalism -- a market-substitution argument neither Meta nor Anthropic's plaintiffs made as concretely. Pulse has tracked the case through an earlier dispute over hidden evidence and sanctions motions before it reached this summary-judgment phase.
The DOJ's unprecedented intervention
The most consequential new development is the Department of Justice's own brief, filed September 2, which sided with OpenAI's fair-use position -- the first time the federal government has taken a formal position in any AI copyright litigation, the Washington Post reported. That's a meaningful signal for every AI company facing similar suits, since a DOJ brief carries weight with judges beyond its narrow legal argument, and Anthropic, Meta, Google and Microsoft are all defendants in comparable pending cases that could benefit from the same precedent if the Times case sets one.
The case has been running since December 2023, making it one of the oldest and most closely watched of the current wave of AI copyright suits, and its outcome will likely shape licensing negotiations across the publishing industry regardless of which way it's decided -- a Times win would force AI labs toward the licensing-deal model OpenAI has already struck with several other publishers; an OpenAI win would remove much of the leverage publishers currently have in those negotiations.
The risk in reading too much into the DOJ brief is that it reflects the current administration's general posture toward AI industry regulation, not a binding legal conclusion -- judges are free to disregard government amicus positions, and Kadrey and Bartz are both still under appeal themselves, meaning the fair-use precedent OpenAI is leaning on isn't fully settled law yet either.
License versus litigate
The Times' litigation strategy stands apart from the roughly dozen major publishers that instead struck licensing deals with OpenAI rather than suing: News Corp agreed to a deal worth up to $250 million over five years in May 2024, covering the Wall Street Journal, the Times of London and the New York Post; Axel Springer, the Financial Times, the Associated Press, Vox Media, Conde Nast and Reddit have all signed comparable licensing arrangements. That split in the publishing industry -- license versus litigate -- means the Times case functions as a real-money test of which strategy publishers should have chosen: if the Times wins a large judgment or forces a much richer licensing structure through the courts, publishers that settled early for flat fees may look to have left money on the table; if OpenAI wins outright on fair use, the Times will have spent years and significant legal costs to arrive at the same market position competitors reached for free.
For AI labs, the case's outcome will likely set the template other publishers use in future negotiations regardless of settlement terms already signed, since a court ruling on the market-substitution question -- whether chatbot outputs compete with the underlying journalism -- would apply to every past and future licensing deal's leverage, not just this one lawsuit.
