Author Kevin Kelly Files Brief Backing OpenAI's Fair Use Defense in Authors' Case

What happened? Kevin Kelly, the founding executive editor of Wired and a New York Times bestselling author, has filed a proposed friend-of-the-court brief backing OpenAI and Microsoft in the consolidated copyright cases in Manhattan federal court. His argument is that training AI models on copyrighted books is fair use.

The filing is dated Friday, October 16, 2026, and was entered on the docket as Exhibit A to ECF No. 2112 in In re OpenAI, Inc., Copyright Infringement Litigation (No. 25-md-3143, Southern District of New York), before Judge Sidney H. Stein and Magistrate Judge Ona T. Wang. It supports the defendants' motions for summary judgment and lists three member cases, including the authors' class case (No. 1:23-cv-08292).

What makes it stand out is who signed it. Most of the people suing OpenAI are writers. Kelly is a writer too, and he is asking the court to rule against them on the central question in the case.

Who Kelly is, and why he says he cares

The filing describes Kelly as "an author, like Plaintiffs, who creates original works of nonfiction and fiction." It cites two of his books, The Inevitable (2017) and The Silver Cord (2015), along with several 2026 essays on his own site about large language models.

He says copyright protection matters to him "based on his lifelong pursuits of writing." But he argues that a win for the plaintiffs on fair use would give them "far more than the limited monopoly to their original expression Congress intended to confer." Kelly is represented by Lehotsky Cohn LLP of Washington, D.C.

The document is labeled a proposed brief, the form a non-party uses when it asks the court's leave to weigh in. It runs 4,013 words by counsel's count.

The argument in three parts

Kelly's filing makes three main points.

Training works like human learning. He says AI "is not trained or designed to create copies of original works." Instead, he argues, models learn patterns across a huge body of text, much as writers absorb what they read over a lifetime. He calls the plaintiffs' books "an infinitesimal fraction" of what the models are trained on.

AI helps authors make new books. This is the part aimed most squarely at writers. The filing cites a survey reporting that 81% of authors use AI for research and 71% for editing. It then describes a self-published novelist, Aaron Black, who it says used AI to publish four novels in ten months "while writing all the prose himself." According to the filing, Black used AI to research historical facts, weather and the terrain of Connecticut towns, to track where characters and objects last appeared without rereading 90,000 words of manuscript, and to make visual references for scenes. "He retained all creative control over the novels and the prose," the filing says.

The public benefit outweighs the harm. Kelly points to uses outside publishing, such as drug discovery, math research and cold criminal cases, and argues that none of them hands readers copies of anyone's book. He also cites the 2025 ruling in Bartz v. Anthropic among the cases supporting his view.

A direct answer to the authors' class

Kelly's filing replies by name to one of the authors' main arguments. The authors' class, in a September filing PT reported on, warned that AI-generated books could displace "human-authored books." Kelly responds that this prediction "does not settle whether training is fair use," and that similar fears have followed earlier technologies that made new uses of copyrighted works.

He does not wave away every concern. "Amicus recognizes serious questions about AI's technical capabilities and its effects on society," the filing says. "Copyright law does not answer all of them."

His conclusion is one sentence long. The court, he says, "should hold that Defendants' use of copyrighted works to train AI models is a fair use."

Where the case stands

Summary judgment is the stage where a judge can decide a claim without a trial if the key facts aren't in dispute. Both sides have filed motions, and non-parties such as Kelly can ask to add their views. There is no ruling on fair use yet.

The case has been busy this fall. Earlier this month, Judge Stein struck an academic paper on AI and "dilution" from the record after OpenAI and Microsoft challenged how it came about.

Why it matters for authors

Writers are not all on one side. Most of the people suing OpenAI are authors. A filing from a well-known author arguing the other way gives the judge a writer's voice for fair use.

The "AI as assistant" pitch is now a legal argument. Research, continuity notes and reference images are tasks many indie authors already hand to AI tools. Kelly's filing uses exactly that kind of work to argue training should be allowed. How the court treats it could shape how those tools are built and sold.

Nothing changes yet. A proposed brief is one voice in a large record. The fair use question will be decided by Judge Stein, and either side could appeal. Authors in the class don't need to do anything because of this filing.

Sources

• U.S. District Court, S.D.N.Y.: Proposed Brief of Amicus Curiae Kevin Kelly in Support of Defendants' Motions for Summary Judgment, No. 25-md-3143, ECF No. 2112-1 (filed October 16, 2026)