The central dispute is whether using copyrighted content to train AI models counts as fair use, with publishers arguing the harm is concrete and ongoing, while AI companies push a broad fair-use defense.

Rights holders across media—music, publishing, and journalism—have filed numerous lawsuits against AI firms, arguing that training on their works without licenses is infringement.

A core question is whether training on copyrighted works is transformative use under fair use, and thus not infringing.

A Times spokesperson framed the DOJ stance as privileging trillion-dollar AI giants at the expense of American creators.

The case before a New York federal judge centers on these arguments and the discovery evidence gathered so far.

DOJ says its position could shape access to training data and licensing, affecting authors, publishers, and independent publishers beyond this case.

DOJ contends that training language models on vast copyrighted material is transformative and essential for innovation, growth, and national security; restricting it could hinder progress.

The DOJ argues that the NYT’s fair-use interpretation could chill AI development by constraining necessary training and misreading the fair-use doctrine.

Officials emphasize that large-scale training enables scientific and creative progress, presenting a public-benefit argument in favor of fair use.

The lawsuit began in 2023 with claims that journalists’ livelihoods are harmed and that AI summaries could substitute for original reporting, with supporters including several newspapers and investigative outlets.

A September 1 filing in Manhattan federal court marked the first government involvement in this wave of AI-copyright lawsuits, potentially aiding defendants facing similar claims.

The court filing underscores a broader regulatory moment in AI, copyright, and the economics of news publishing.