OpenAI rallies government and scholars to defend fair use

The legal showdown between OpenAI and The New York Times has often been framed as a billionaire tech giant trampling over journalism. But OpenAI just pulled back the curtain on its legal defense, showing it is far from standing alone.

OpenAI published excerpts from a wave of amicus briefs filed in federal court backing its stance that training AI on copyrighted works is a textbook case of fair use. The coalition backing the ChatGPT maker is remarkably diverse: the U.S. government, the American Library Association, the American Council of the Blind, the Electronic Frontier Foundation, Wired co-founder Kevin Kelly, and leading copyright scholars.

Learning patterns is not copying text

The common thread running through the filings is simple: teaching a model how words fit together is fundamentally different from distributing copies of an article.

The Computer and Communications Industry Association and law professor Matthew Sag argued that ingesting text to extract statistical patterns represents one of the most clearly transformative uses a court has ever evaluated. The goal is to build a general-purpose tool, not to resell someone else’s prose.

Legal scholars also took aim at The Times’ argument that AI models dilute their market. The Authors Alliance pointed out that copyright law protects original expression against unauthorized substitution, not against the arrival of new, competing works in the same genre. Professors Ed Lee and Rebecca Tushnet noted that facts and genres have never been copyrightable, and treating broad competitive pressure as legal harm would turn copyright doctrine on its head.

Beyond Silicon Valley's balance sheet

Perhaps the most compelling arguments came from groups outside the tech bubble, warning that shutting down fair use training would hurt real people.

A joint brief from the American Council of the Blind and research library groups warned that millions rely on downstream tools built on foundation models. These include visually impaired individuals using AI for greater daily independence, medical researchers analyzing massive datasets, and educators tailoring lessons to different learning needs.

Two independent creators, comedian Sarah Rose Siskind and musician Michael Wall, filed briefs detailing how AI keeps their small businesses afloat. Siskind noted she could barely run her company without AI helping level the playing field against mainstream corporations. Meanwhile, the Foundation for American Innovation urged the court to consider how gutting fair use would cripple American technological competitiveness and research speed.

Why it matters

We think this is a masterclass in coalition-building by OpenAI. By spotlighting librarians, blind advocates, and working artists, it effectively flips the script from a corporate land grab to a battle over public access and technological progress.

For developers and everyday users, the stakes could not be higher. If the court rules that reading publicly available text to train models requires bespoke licensing for every paragraph, open-source AI would largely become impossible to build legally, and subscription prices for consumer AI tools would skyrocket.

Still, rallies in court dockets do not guarantee a win. While the legal arguments for transformative use are rock-solid in theory, OpenAI still has to convince the judge that its models do not spit out memorized, verbatim text in practice. That line between learning and copying is where this entire war will be decided.